Idiomatic Creativity - Perpustakaan STKIP Kusuma Negara

Transcription

<DOCINFO AUTHOR ""TITLE "Idiomatic Creativity: A cognitive-linguistic model of idiom-representation and idiom-variation in English"SUBJECT "HCP, Volume 17"KEYWORDS ""SIZE HEIGHT "240"WIDTH "160"VOFFSET "4">
Idiomatic Creativity
human cognitive processing is a forum for interdisciplinary research on the
nature and organization of the cognitive systems and processes involved in
speaking and understanding natural language (including sign language), and
their relationship to other domains of human cognition, including general
conceptual or knowledge systems and processes (the language and thought
issue), and other perceptual or behavioral systems such as vision and nonverbal behavior (e.g. gesture). ‘Cognition’ should be taken broadly, not only
including the domain of rationality, but also dimensions such as emotion and
the unconscious. The series is open to any type of approach to the above
questions (methodologically and theoretically) and to research from any
discipline, including (but not restricted to) different branches of psychology,
artificial intelligence and computer science, cognitive anthropology, linguistics,
philosophy and neuroscience. It takes a special interest in research crossing the
boundaries of these disciplines.
Editors
Marcelo Dascal, Tel Aviv University
Raymond W. Gibbs, University of California at Santa Cruz
Jan Nuyts, University of Antwerp
Editorial address
Jan Nuyts, University of Antwerp, Dept. of Linguistics (GER),
Universiteitsplein 1, B 2610 Wilrijk, Belgium.
E-mail: [email protected]
Editorial Advisory Board
Melissa Bowerman, Nijmegen; Wallace Chafe, Santa Barbara, CA;
Philip R. Cohen, Portland, OR; Antonio Damasio, Iowa City, IA;
Morton Ann Gernsbacher, Madison, WI; David McNeill, Chicago, IL;
Eric Pederson, Eugene, OR; François Recanati, Paris;
Sally Rice, Edmonton, Alberta; Benny Shanon, Jerusalem;
Lokendra Shastri, Berkeley, CA; Dan Slobin, Berkeley, CA;
Paul Thagard, Waterloo, Ontario
Volume 17
Idiomatic Creativity: A cognitive-linguistic model
of idiom-representation and idiom-variation in English
by Andreas Langlotz
A cognitive-linguistic model
of idiom-representation and idiom-variation
in English
Andreas Langlotz
University of Basel
John Benjamins Publishing Company
Amsterdam/Philadelphia
8
TM
The paper used in this publication meets the minimum requirements
of American National Standard for Information Sciences – Permanence
of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data
Andreas Langlotz
Idiomatic Creativity : A cognitive-linguistic model of idiom-representation
and idiom-variation in English / Andreas Langlotz.
p. cm. (Human Cognitive Processing, issn 1387–6724 ; v. 17)
Includes bibliographical references and indexes.
1. English language--Idioms. 2. English language--Variation. I. Title.
II. Series.
PE1460.L27 2006
427--dc22
isbn 90 272 2370 X (Hb; alk. paper)
2006040683
© 2006 – John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microfilm, or
any other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
JB[v.20020404] Prn:23/02/2006; 11:52
F: HCP17CO.tex / p.1 (48-111)
Table of contents
Tables and figures
Acknowledgments
chapter 1
Introduction
1.1
A preliminary definition of idiom 2
1.2 Introducing the problem: Idioms and creativity –
a contradiction in terms? 6
1.3 Towards a cognitive-linguistic approach to idiom
representation and variation 9
chapter 2
Idiom representation and variation – a hard nut to crack
2.1 Idioms as semantic units – the orthodox view 16
2.2 The compositional view 25
2.3 Proverbiality – the functional motivation of idioms 44
2.4 Outlook – desiderata for a cognitive-linguistic model of idiom
representation and variation 53
chapter 3
The cognitive architecture of meaning and language
3.1 Basic cognitive processes 57
3.2 The mental representation of knowledge and meaning 61
3.3 Complex patterns of semantic extension 66
3.4 Cognitive Grammar: The mental representation
of linguistic knowledge 75
3.5 Summarising overview 90
chapter 4
Idiom representation – a cognitive-linguistic model
4.1 Adapting idioms to the cognitive-linguistic framework 93
ix
xi
1
15
57
93
JB[v.20020404] Prn:23/02/2006; 11:52

F: HCP17CO.tex / p.2 (111-141)
4.2
4.3
4.4
4.5
4.6
4.7
A cognitive-linguistic account of compositeness 96
Institutionalisation and lexicalisation – the cognitive
entrenchment of an idiom 99
A cognitive-linguistic anatomy of the internal semantic
structure of idioms 106
Conceptual patterns shaping the internal semantic structure
of idioms 120
Typical patterns of figuration reflected by idioms 125
The cognitive functionality of idioms 135
chapter 5
The conceptual motivation of idioms denoting success, progress
and failure
5.1 What is success, progress and failure? –
A cognitive-model 144
5.2 The conceptual source domains for spf-idioms 145
5.3 Metonymic and metaphtonymyic motivation 166
5.4 Motivation by emblems 168
5.5 Opaque and constructionally-idiosyncratic spf-idioms 169
5.6 Implications for the psycholinguistic controversy about
metaphorical motivation 171
chapter 6
Idiom variation and variability – a cognitive-linguistic model
6.1 Frozenness vs. variability –
towards a cognitive-linguistic view 176
6.2 Idiomatic creativity 185
6.3 Subtypes of idiom variation 194
6.4 Principles of idiom variation 205
6.5 Cognitive constraints on idiom variation 215
chapter 7
The lexicogrammatical variation of idioms denoting success,
progress and failure
7.1
Database and analytical procedure 226
7.2 Articles and article-variation in spf-idioms 229
7.3 Number and number variation in idioms 245
7.4 Passivisation – idiom variation at the clause level 249
7.5 Adnominal modification 256
143
175
225
JB[v.20020404] Prn:23/02/2006; 11:52
F: HCP17CO.tex / p.3 (141-159)
Table of contents 
7.6
7.7
Lexical substitution 271
Controversial data 284
chapter 8
Conclusions and outlook
287
Notes
299
References
303
Appendix
www.idiomatic-creativity.ch
Author index
317
Subject index
319
JB[v.20020404] Prn:20/02/2006; 15:28
F: HCP17TF.tex / p.1 (46-108)
Tables and figures
Table 1.1
Table 4.1
Table 4.2
Table 6.1
Table 6.2
Table 7.1
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 3.7
Figure 4.1
Figure 4.2
Figure 4.3
Figure 5.1
Figure 5.2
Figure 6.1
Figure 6.2
Parameters for the definition of idioms
The -metaphor
Substructures of the idiomatic activation-set for grasp the nettle
Overview of idiom-variation classes
Overview of variation classes including variation principles
and variational constraints
Influencing factors for idiom passivisation
Deviant triangle
The     
  -metaphor
Constructional schemas
Categorisation of a usage event
Constructional schema for pencil sharpener
The creation of a composite construction
The multi-layered architecture of meaning and language
The composite structure of grasp the nettle
Geeraerts’s prismatic model of idiom semantics
Ontological correspondences underlying the isomorphism
of rock the boat
The conceptual ontology and structure of the -
Overview of source domains for -
Technical classification of idiom alterations
Idiomatic usage-events




















JB[v.20020404] Prn:23/01/2006; 11:00
F: HCP17AC.tex / p.1 (45-116)
Acknowledgments
This book grew out of my 2003 Ph.D thesis at the University of Basel, Switzerland. Some of the contents and ideas have also been addressed in pilot research
papers and conference proceedings (Langlotz 2001a, 2001b, 2004a, 2005; Langlotz forthcoming, under review).
My special thanks go to Jan Nuyts, Marcelo Dascal and Ray Gibbs for having accepted this book for publication in the Human Cognitive Processing series.
I am very grateful for their critical and stimulating comments, which helped
me turn my dissertation into a much better, more coherent and accessible text.
I would also like to thank the following people for having directly or indirectly supported me in writing the dissertation. I am very grateful to Willy
Elmer and Annelies Häcki Buhofer, who have motivated me to write the thesis and have always given me a great deal of intellectual and moral support. I
owe many thanks to David Allerton, Dmitrij Dobrovol’skij and Harald Burger
for the stimulating and inspiring discussions on phraseology and language and
their critical comments on many of my ideas. Very special thanks go to my
colleagues and friends Christoph Ebell, Daniele Ganser, Alexandra Guski, Regula Hohl, Danièle Klapproth, Nadja Nesselhauf, Andie Ochsner, Guy Schiltz,
Philipp Schweighauser, Paul Skandera and Judith Wieser, who have motivated
me – in all sorts of uncountable ways – to continue my work in good and not so
good times: without you, academia would be a much duller place. I feel as well
great gratitude to the many teachers and professors, especially Othmar Gisi,
Jürg Bauer and Hartwig Isernhagen, who have convinced me that English is
the most thrilling subject to study.
My deepest feelings go to all of my family, especially my parents and my
parents-in-law. Where would I be without your helping hands, great guidance
and emotional support?
There are no words to express what I feel for my wife Carla and my children
Fiona and Nicola – . . . . I thank you so much for being so patient, sympathetic,
strong and loving.
Basel, Autumn 2005
Andi Langlotz
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.1 (47-122)
chapter 
Introduction
‘All right,’ said the Cat; and this time it vanished quite slowly,
beginning with the end of the tail, and ending with the grin,
which remained some time after the rest of it had gone.
‘Well! I’ve often seen a cat without a grin,’ thought Alice; ‘but
a grin without a cat! It’s the most curious thing I ever saw in
all my life!’
(Lewis Carroll, Alice in Wonderland).
This book addresses a central problem in phraseological and linguistic analysis. The creative structure and the creative use of idioms. Let me therefore start
creatively, with a highly speculative metaphorical hypothesis: idioms are to linguists and language users what the Cheshire cat is to Alice. Idioms are peculiar
linguistic constructions that have raised many eyebrows in linguistics and often
confuse newcomers to a language. Indeed, the expression grin like a Cheshire
cat is an idiom. More precisely, it is an idiomatic comparison whose motivation has become opaque: as the Oxford English Dictionary (OED) indicates, the
phrase is of undetermined origin. So why should people who grin broadly grin
like a Cheshire cat?; after all, what, precisely, is a Chesire cat? These questions
point to the fact that according to the perception of most speakers of English
this expression behaves like its allegorical incarnation in Alice in Wonderland.
The idiomatic meaning of the expression, ‘grin broadly,’ leaves us with a grin
without a cat. This is what makes this idiom curious.
Many other idiomatic expressions behave in a similarly puzzling way.
When browsing through common idiom-dictionaries, one notices that the set
of linguistic expressions termed ‘idiomatic’ is very heterogeneous indeed: after
all, in any case, by and large, to put up with, grasp the nettle, to trip the light
fantastic, as dead as a doornail, birds of a feather flock together, good morning.
One can tell from a very superficial inspection that these units differ considerably with regard to their structural complexity, their lexical, morphosyntactic,
syntactic and semantic organisation as well as their discursive function. These
factors give them a linguistic nature that shares close correspondences with
the hardly ‘graspable’ character of the Cheshire cat. In short, idioms are very
complex linguistic configurations.
JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.2 (122-172)
The colourful linguistic spectrum of expressions called ‘idioms’ directly
reflects the considerable difficulties linguists face in finding an appropriate definition and classification of these linguistic phenomenona and to explain their
grammatical behaviour. The heterogeneity of idiomatic expressions stands in
a dialectical relation to the abundance of linguistic terminology developed to
capture and classify these constructions. Nunberg et al. (1994: 492) provide the
following diagnosis of the terminological state of the discipline:
In actual linguistic discourse and lexicographical practice, ‘idiom’ is applied
to a fuzzy category defined on the one hand by ostension of prototypical examples like English kick the bucket, take care of NP, or keep tabs on NP, and
on the other by implicit opposition to related categories like formulae, fixed
phrases, collocations, clichés, sayings, proverbs, and allusions – terms which,
like ‘idiom’ itself, inhabit the ungoverned country between lay metalanguage
and the theoretical terminology of linguistics.
Although various attempts have been made to structure this terminological
haze (see, e.g., Pilz 1978, 1981; Thun 1978; Burger et al. 1982; Gläser 1986,
1988, 1998; Barkema 1996; Mel’čuk 1995, 1998; Cowie 1998; Moon 1998),
phraseologists still cannot agree on a shared set of terms to describe the linguistic phenomena they discuss. Given this state of affairs, one has to develop
a suitable strategy to find one’s way through the terminological jungle to become capable of providing a systematic account of the Chesire-cat-like nature
of these curious linguistic phenomena.
. A preliminary definition of idiom
Many phraseologists have noticed that it is impossible to capture the linguistic
anatomy of idioms without relying on a set of different definitory dimensions
(see, e.g., Fernando & Flavell 1981; Burger et al. 1982; Barkema 1996; Fernando
1996; Burger 1998a; Moon 1998). Traditionally, idioms such as grasp the nettle, blow the gaff or trip the light fantastic have been described as conventional
multi-word units that are semantically opaque and structurally fixed. Thus the
internal organisation of idiomatic constructions can show more or less striking
a. semantic characteristics,
b. structural peculiarities and irregularities and
c. constraints or restrictions on their lexicogrammatical behaviour which
cannot be explained by the general grammatical rules of the given language. Nevertheless, idioms are
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.3 (172-253)
Chapter 1. Introduction
d. conventional expressions that belong to the grammar of a given language
and
e. fulfil specific discourse-communicative functions
In short, idiomatic constructions can be described as complex symbols with
specific formal, semantic, pragmatic and sociolinguistic characteristics. The
following table summarises these definitory features and patterns them along
the semiotic dimensions of form, meaning and grammatical status:
Table 1.1 Parameters for the definition of idioms
Semiotic dimension
Feature
Term
GRAMMATICAL
STATUS
FORM
Degree of conventionalisation or familiarity
institutionalisation
Formal complexity of construction: multiword unit
Lexicogrammatical behaviour: restricted
syntactic, morphosyntactic and lexical
variability
Meaning cannot be derived from constituent words but is extended/figurative.
compositeness
MEANING
frozenness
non-compositionality
Belonging to the grammatical system of a given speech community, idioms
are linguistic constructions that have gone through a sociolinguistic process of
conventionalisation. To capture an idiom’s degree of familiarity and conventionality within a given speech community, the term institutionalisation is used
(Fernando 1996: 3).
Compositeness refers to the fact that idioms are multi-word units that consist of two or more lexical constituents. For instance, grasp the nettle consists
of the constituents grasp and nettle, with nettle being conventionally used with
the definite article the. While idioms typically have the composite structure of
phrases or semi-clauses, idiomatic compounds (blackbird, chatterbox), phrasal
verbs (stand by, see through, come across) and proverbs (Birds of a feather flock
together) also belong to the group of composite idiomatic constructions. Some
linguists have included these constructions into their analyses of proper idioms
(see, e.g., Makkai 1972, also cf. Kuiper & Everaert 2004), whereas others exclude them (e.g., Rothkegel 1973; Moon 1998). For practical reasons the second
position is adopted in this study.
Fraser’s notion of frozenness is adopted as a generic term to capture lexicogrammatical restrictions (Fraser 1970). Frozenness can influence the variability of the lexical constituents and the grammatical behaviour. Therefore, I

JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.4 (253-299)
will further use the notion of fixedness to denote any syntactic and morphosyntactic restrictions and restricted collocability to capture paradigmatic constraints on the selection of lexical items (see also Barkema 1996: Section 3.4).
For example, trip the light fantastic is considerably frozen. It features fixedness
because it cannot be passivised for instance. Moreover, it is collocationally restricted because the constituents cannot be replaced by other lexical items: *trip
the heavy fantastic, *walk the light fantastic. Apart from being frozen this idiom
also features constructional idiosyncrasy. I will use this term to denote the presence of idiosyncratic (and irregular) lexical items and grammatical patterns.
Accordingly, blow the gaff is constructionally idiosyncratic too since it contains
the unique lexical constituent gaff.
Traditionally, semantic non-compositionality has been used as the primary
feature to define idioms. This notion points to the fact that the meaning of
these constructions is not the derivational sum of the meanings of their constituents:
The essential feature of an idiom is that its full meaning, and more generally
the meaning of any sentence containing an idiomatic stretch, is not the compositional function of the meaning of the idiom’s elementary parts.
(Katz & Postal 1963: 275)
In other words, the overall meaning of an idiomatic construction is a semantic
extension from the compositional result of the meanings of its lexical constituents. The relationship between an idiom’s overall meaning and the sum
of the meaning of the constituents reflects a pattern of figuration. Since idioms are institutionalised expressions, their extended meaning – and with it
the conveyed pattern of figuration – has become fixed in the lexicon of a given
speech community (cf. also Burger 1989). We must thus distinguish two levels
of meaning:
a. the literal meaning (the sum of the meaning of the constituents)
pattern of figuration
b. the idiomatic meaning (the lexicalised extended meaning of the construction)
The more discrepancy between the literal and the idiomatic meaning a construction features, the more opaque it is.
Finally, idioms can serve different communicative purposes involving different types of ideational, interpersonal and textual functions (see Fernando
1996: 1; Halliday 1978; also cf. Strässler 1982). Prototypical idioms primarily
serve an ideational function. For instance, grasp the nettle communicates an
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.5 (299-353)
experience or event (tackle a problem). In contrast, the address formula good
morning serves an interpersonal function, whereas the routine formulation in
a nutshell is mainly used to support textual structuring.
In general, these definitory parameters or typological axes can be applied
to any construction to anatomise and describe its idiomatic nature. Different
definitions and classifications of idioms can be distinguished with reference to
how many of these basic characteristics are selected or emphasised (for concise
overviews see Barkema 1996; Skandera 2004). The definitory and classificatory
problems in the study of idioms become more intelligible once one realises
that none of these discriminatory features is clear-cut. First, the feature of
compositeness creates some overlap with compounds (see Gläser 1986, 1998).
Moreover, institutionalisation, frozenness and non-compositionality are clines.
And while the discursive functions of different phrasemes are usually more distinct, they can nevertheless overlap. As a result, no definition of idiom can be
fully clear-cut. Rather, the definitory dimensions must be understood as continua that can be spotted on a given construction to find out about its degree
of idiomaticity, i.e. its degree of belonging to the class of idiomatic constructions. Only the sum of all dimensions – the full spectrum of all descriptive
spotlights – provides a holistic picture of the specific idiomatic nature of a construction. Here, I will depend on the following preliminary definition of idiom
or idiomatic construction:
An idiom is an institutionalised construction that is composed of two or more
lexical items and has the composite structure of a phrase or semi-clause, which
may feature constructional idiosyncrasy. An idiom primarily has an ideational
discourse-function and features figuration, i.e. its semantic structure is derivationally non-compositional. Moreover, it is considerably fixed and collocationally restricted.
This definition includes constructions such as red herring, buy the farm, take
the bull by the horns or fall flat.
In order to sketch the complex anatomy and behaviour of idioms further
and to approach a more fine-grained account of their Cheshire-cat-like nature,
it is fruitful to discuss idioms relative to the concept of creativity.

JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.6 (353-395)
. Introducing the problem: Idioms and creativity – a contradiction
in terms?
We commonly associate creativity with ingenious, artful or playful activities
that are unconventional in the very broad sense of not being subject to everyday routine. Thus, creativity is inseparably linked with intelligent human
behaviour. More specifically, it involves the mental ability to develop or invent
new and original ideas or products that have not been encountered before (for
a good overview of different approaches see Carter 2004: Chapters 1 and 2;
Salminen 1993).
Given the fact that most of the sentences and texts that we hear and read
every day have never been heard before, linguistic production can certainly
be regarded as a creative process. Most prominently, Chomsky (1965, 1971)
has placed the notion of linguistic creativity at the very centre of his approach
to language. By emphasising the rule-driven productivity of language, Chomsky opposed the behaviourist view, which describes linguistic productions as
the repetition of pre-heard linguistic stimuli. According to Chomsky, linguistic expressions are not the result of the reproduction of memorised linguistic
structures but are derived through the combination of memorised units – the
lexicon – by means of a formal combinatory apparatus – syntax. Given its focus on the combinatory formalisms underlying syntax, Chomsky’s generative
paradigm placed syntax in the foreground of linguistic analysis, while it pushed
the lexicon – regarded as the stock of pre-fabricated units – to the background.
In contrast, various linguists from different subfields have claimed that linguistic production cannot be fully explained on the basis of this syntax vs.
lexicon dichotomy (see, e.g., Halliday 1978; Coulmas 1979, 1981; Pawley &
Syder 1983; Tannen 1989). Following this line of reasoning, Sinclair introduced
the notion of idiom principle, which he contrasts with the open-choice principle
(Sinclair 1987: 319). According to him, the open-choice principle is inherent
in the segmental descriptions of grammatical structures and finds expression
in the “slot-and-filler” models provided by most grammatical approaches. The
open-choice principle cannot give a full account of linguistic production and
interpretation and must therefore be complemented by the idiom principle:
The principle of idiom is that a language user has available to him or her a
large number of semi-preconstructed phrases that constitute single choices,
even though they might appear to be analysable into segments. To some extent
this may reflect the recurrence of similar situations in human affairs; it may
illustrate a natural tendency to economy of effort; or it may be motivated in
part by the exigencies of real-time conversation. However, it arises, it has been
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.7 (395-461)
relegated to an inferior position in most current linguistics, because it does
not fit the open-choice model.
(Sinclair 1987: 320)
In other words, two principles seem to be at work in the production and interpretation of constructions: the creativity principle or open-choice principle
and the idiom principle. How, then, can these principles be related to idioms?
Phraseology is the linguistic discipline that analyses phraseological, i.e.
pre-established, constructions. So, obviously, phraseology focuses on those linguistic expressions which are closely connected to the idiom principle.1 Indeed,
traditional definitions of the notion of idiom have principally relied on demarcating these expressions from regularly-generated standard constructions
(see, e.g., Mel’čuk 1995, 1998). In this process of defining idioms in opposition to the norm of grammatical regularity, they have customarily been linked
to notions such as linguistic irregularity, arbitrariness or anomaly. For instance,
Weinreich (1969: 45) postulates that “[...] the semantic difference between idioms and their literal counterparts is, by definition, arbitrary in principle [...].”
With regard to the processing of idioms, this view further implies that idioms
are not freely formed grammatical constructions, i.e. they are not produced
through linguistic composition; rather, they are directly reproduced from the
mental lexicon. In this sense, idioms also seem to violate Chomsky’s notion of
linguistic creativity. Therefore they have typically been treated as lexical units
rather than regularly derived phrases in this framework.
The identification of idioms with non-creative, preconstructed and reproduced linguistic material is also implied in the following statement made by
George Orwell:
The invasion of one’s mind by ready-made phrases (lay the foundations, acquire a radical transformation) can only be prevented if one is constantly on
guard against them, and every such phrase anaesthetizes a portion of one’s
brain.
(George Orwell 1946; quoted in Fernando 1996: 26)
When making this claim, Orwell probably had a very literary or poetic conception of creativity in mind. And although this statement must not be regarded
as reflecting the state-of-the-art of linguistic research in idiom representation
and use, Orwell’s equation of “ready-made phrases” with non-creativity is also
conveyed by many traditional accounts of idiomatic language. Idiomatic constructions thus seem to epitomise non-creativity in linguistic processing. One
could even go a step further and claim that the use of idioms does not reflect originality, ingenuity or playfulness and should therefore not be taken as
a linguistic measure of human intelligence.

JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.8 (461-515)
Such a position obviously contrasts sharply with Lewis Carroll’s playful literary incarnation of the idiomatic Cheshire-cat. Clearly, Lewis Carroll’s comic
figure is a highly creative adaptation of the idiomatic construction. But the link
between the notions of idiom and creativity is far from being limited to such
artful outbursts of imagination. On the contrary, the intimate link between the
two phenomena can already be observed in seemingly trivial variation-data,
which challenges the unqualified identification of idiomaticity with routinised
linguistic activity. Consider, for instance, the following uses of the idiom grasp
the nettle as attested by the British National Corpus (BNC):2
(1) He does not appear, however, to have grasped the management nettle or to
have found a way of dealing with its stings. (K4T: 127).
(2) Conservation thinking has begun to grasp a few of the same nettles. (APN:
1380).
(3) Disappointing as the terms were from the peasant point of view ‘for the
Imperial government to grasp the nettle of Emancipation at all was a remarkable departure. (EA6: 823).
(4) It seemed to us that two nettles had to be grasped: the first was to decide
whether divorce law should attempt to remedy the ‘injustice’ meted out to
‘innocent’ spouses who are divorced against their will. (BNK: 1130).
(5) Moreover, deep-rooted public opposition to urban road construction
emerged in the 1970s, so that the only feasible way forward would have
been a policy of car restraint. But this nettle was never grasped; [...],
(C8F: 189).
(6) There is a nettle for a somebody – it might as well be the new Environment
Secretary – to grasp. (AAG: 73).
(7) But it was NME that embraced the intellectual nettle most firmly.
(CHA: 1670).
Although they consistently instantiate the relatively stable idiomatic construction grasp the nettle, these uses do not seem to reflect the activity of an uncreative, partially “anaesthetised” brain. Rather, they involve the systematic
application of regular grammatical processes including: adnominal modification, passivisation, lexical substitution, pluralisation and the flexible use of
determiners. In other words, lexicogrammatically, grasp the nettle is far more
variable than a simple process of lexical reproduction can explain.
The data reflect a specific form of creativity that I would like to describe
as idiomatic creativity. Idiomatic creativity reveals the Cheshire-cat-like nature
of idiomatic constructions: by analogy with the behaviour of this strange crea-
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.9 (515-562)
ture, idiomatic creativity captures the varied evocation of a relatively stable
idiomatic construction in a specific context of use. Thus, rather than just being
reproduced, idioms can be varied in discourse; in the same way as the Cheshire
cat, these constructions may change their conventional appearance: parts of
their conventional formal and semantic substructures can be highlighted with
others changing or fading away.
Idiom-variation data therefore question the strict dichotomy between the
creativity principle and the idiom principle. Rather than leading to a strict
opposition between regular standard constructions and irregular idiomatic
constructions, idiomatic creativity implies that idiom production and comprehension are subject to a dynamic tension between the two principles. This
further suggests that idioms cannot merely be described as lexical items; rather,
they seem to occupy a position between the lexicon and syntax, leading to a
fuzzy dividing line between the productive and reproductive aspects of linguistic competence.
. Towards a cognitive-linguistic approach to idiom representation
and variation
Chomsky (1975: 4–5) claimed: “By studying the properties of natural languages
[...] we may hope to gain some understanding of the specific characteristics of human intelligence.” With regard to the Chomskyan enterprise, Smith
(1999: 33–34) further specifies: “[...] to understand language it is sometimes
necessary to study extreme examples.” The relationship between idioms, idiom variation and the architecture of the human cognitive capacity is the
central theme pursued in this book. Although idioms have traditionally been
regarded as ‘extreme examples’ of linguistic structuring and processing, they
are universally-found ingredients of any natural language (Dobrovol’skij 1988).
Given this state of affairs, idioms are a particularly suitable phenomenon to
pursue Chomsky’s higher-order aim of linguistic analysis, although – and this
is the irony contained in my proposal – they have generally been regarded as
atypical linguistic structures in the Chomskyan paradigm (cf. also Baranov &
Dobrovol’skij 1991: 113). Going a step further, it can be claimed that the phenomenon of idiom variation and the question of its limitations is an even more
suitable mirror of intelligent human behaviour. Varied idioms represent ‘deviant’ instances of such ‘extreme examples’ and can therefore be polemically
described as reflecting the dynamics of the abnormal.

JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.10 (562-616)
The central aim of this book is to show that the cognitive dynamics reflected by idiomatic creativity is guided by very general cognitive processes
that underlie natural linguistic and non-linguistic processing. In this sense,
the phenomenon of idiomatic creativity is expected to reflect very general
cognitive abilities and principles that determine what we perceive as human
intelligence. In other words, idiom variation is claimed to reveal intelligent creative behaviour that exploits basic knowledge-resources and the information
processing capacities of the human mind.
In phraseology, idiom variation has long been recognised as an empirical
fact. Phraseological research has revealed that idioms show a greater degree of
formal and semantic flexibility than was traditionally conceded. Nevertheless,
the debate about the precise linguistic and mental quality of idioms is still undecided. This favours new approaches to the problem. In the last two decades,
linguistics has witnessed the increasing importance of cognitive linguistics –
the linguistic theory that is most closely associated with the names of Langacker, Lakoff and Fauconnier. Explaining the working principles of language
on the basis of general cognitive abilities such as categorisation or schematisation, cognitive linguists have developed a psychologically plausible, holistic
theory of linguistic representation and processing. This approach promises
new and deeper insights into the phenomenon of idiomatic creativity (see also
Baranov & Dobrovol’skji 1991: 112–113; Hessky 1995).
On the basis of a broader, cognitively motivated view of linguistic structure, the mental status of idioms as word-like lexical units can be questioned.
This opens new possibilities for the description of idiom representation and
the explanation of idiom variation. A cognitive-linguistic investigation into idiomatic creativity is thus forced to design an appropriate model of the mental
basis and correlates of this phenomenon. The model must accord with the
concepts, results and insights provided by phraseological analysis and, most
importantly, with empirical data reflecting the phenomenon. Accordingly, this
book addresses the following questions:
1. How are idioms represented and organised in a speaker’s cognitive grammar?
2. What cognitive structures and processes underlie this patterning?
3. Along which lines does idiom variation occur?
4. Can idiom variation be explained with reference to more general cognitive
processes that determine their mental representation and use?
To answer these questions, I will follow previous cognitive-linguistic accounts
of idiomaticity (Lakoff 1987: 446–453; Gibbs 1994; Kövecses & Szabó 1996).
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.11 (616-664)
The book develops a coherent cognitive-linguistic model of the cognitive status, the mental representation and the variation of idioms. This model is established on and tested against the basis of corpus-data that reflect actual idiomuse. Referring back to the introductory metaphorical analogy, the book is an attempt to anatomise the idiomatic Cheshire-cat from a cognitive-linguistic perspective to understand its intricate nature and to explain its puzzling behaviour.
More precisely, this book develops, models and tests the following argument:
Idiomatic creativity is based on a conceptualiser’s competence to construct,
structure, manipulate and construe conceptual patterns of figurativity. In idiom representation and idiom use, cognitive creativity therefore appears on
two levels; to trigger idiomatic creativity these two levels must interact:
a. Idioms were originally created as non-conventional metaphors or metonymies. Thus, their internal structure incorporates the systematic and creative extension of semantic structures. This, for instance, can be seen with
grasp the nettle. This idiom involves the creative extension of a literal meaning (grasp the nettle) to denote a more abstract process (tackle a difficult
problem). In this sense, idioms are structurally and semantically complex
linguistic constructions that are intrinsically creative.
b. The variation of an idiomatic construction in a specific usage-event involves variational creativity (as illustrated in (1)–(7)). The idiom as a
whole is manipulated as a mental configuration to adapt it (with its
substructures) to the communicative demands emerging in the specific
context of use.
c. Idiom variation is the result of the complex interaction of the two facets
of idiomatic creativity specified in (a) and (b). More specifically, the degree to which an idiom can be systematically and creatively manipulated
in discourse is dependent on the degree to which the idiom’s intrinsic creativity remains accessible to the language user or can be re-established by
him or her. Grasp the nettle can be systematically remotivated relative to
a set of conventional conceptual metaphors (see Sections 2.3.2., 3.3.1 and
Chapter 5). These metaphors render the idiom transparent and analysable.
This provides the basis for the syntactic and lexical manipulation of its
constituents. In other words, I claim that systematic idiom variation is the
effect of manipulating an idiomatic construction relative to the underlying conceptual correspondences that shape its creative, internal semantic
structure.
The cognitive-linguistic analysis is expected to shed more light on the linguistic
controversy centred around the representation and variational behaviour of

JB[v.20020404] Prn:20/02/2006; 14:08

F: HCP1701.tex / p.12 (664-722)
idioms and should provide some cognitive-linguistic suggestions to solve its
central problems.
The book is organised as follows. In Chapter 2, the linguistic and psycholinguistic controversy centred around the representation and processing
of idioms is reviewed. It will be shown that the linguistic and psycholinguistic approaches to idiom representation and variation have provided striking
insights into their heterogenous semantic nature and grammatical behaviour;
however, they have not yet produced comprehensive models to explain these
phenomena. The chapter therefore concludes with a proposal of how the bull
of idiom representation and variation can be taken by its horns with the help
of a cognitive-linguistic approach.
In Chapter 3, the cognitive-linguistic background, necessary to get the
cognitive-linguistic idiom-model off the ground, will be outlined. More specifically, this part mainly integrates concepts from Langacker’s Cognitive Grammar with Lakoff ’s Cognitive Semantics and his epistemological framework of
experientialist realism. In addition, some basic cognitive-pragmatic concepts
from Relevance Theory are introduced. The combination of these theoretical
frameworks results in a comprehensive cognitive architecture of knowledge,
meaning and grammatical structure, which also includes cognitive patterns of
figuration (metaphor, metonymy and blending). These complex patterns of semantic extension provide the basis on which the semantic structure of idioms
can be modelled.
The cognitive-linguistic re-interpretation of the grammatical status and
structure of idioms is performed in Chapter 4. Idioms will be modelled as
complex mental activation-sets with variable internal structures. The semantic
organisation of these activation-sets is explained with reference to the cognitive patterns of figuration. In Chapter 5, this model of idiom representation
and motivation will then be applied to idioms from the word-field of success,
progress and failure (spf) to reveal their systematic semantic organisation.
Alternative ways of how idiomatic configurations can be activated in discourse are discussed in Chapter 6. I will develop a cognitive-linguistic model
of idiom variation that includes alternative idiom-variation principles, i.e.
different strategies that underlie the variation of idiomatic constructions in
discourse. Relative to these strategies, a distinction between systematic idiom
variation and idiomatic wordplay will be established.
In Chapter 7, the explanatory potential of this cognitive-linguistic model
will be tested against empirical data. That is, the actual lexicogrammatical behaviour of English spf-idioms will be analysed on the basis of corpus-data
extracted from the BNC. The analytical part is to reveal if the model estab-
JB[v.20020404] Prn:20/02/2006; 14:08
F: HCP1701.tex / p.13 (722-732)
lished in the theoretical parts is powerful enough to account for the variants
extracted from the corpus. This cognitive-linguistic analysis is an attempt to
develop a idiom-variation grammar. It is intended to shed more light on the
controversy centred around the representation and variational behaviour of idioms and provides well-motivated cognitive-linguistic suggestions to solve its
central problems. Note that all the results from the analyses in Chapter 5 and
Chapter 7 are presented in detail in the electronic appendix (www.idiomaticcreativity.ch).
Finally, the conclusion summarises the major results and makes suggestions for further linguistic and psycholinguistic research.
With this plan in hand, we can now plunge into the idiomatic Wonderland.

JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.1 (47-131)
chapter 
Idiom representation and variation –
a hard nut to crack
An analysis of the lexical, morphological, and syntactic
flexibility of idioms would represent a very considerable
undertaking.
(Fellbaum 1993)
The central question of this study is whether idioms can be attributed a motivated internal semantic structure and if this structure influences their syntactic
and lexical flexibility. This question summarises an ongoing linguistic and psycholinguistic controversy concerned with the representation and status of idioms in grammar and the mental grammar of actual speakers. The controversy
centres around the two following, disparate views:
a. Idioms must be regarded as non-compositional, unanalysable, and unmotivated semantic units. Therefore, idioms have the status of lexical units
and they are processed non-compositionally by means of direct lexical retrieval. As a consequence, systematic idiom-variation cannot be explained
with reference to their internal semantic patterning, but must be explained
in terms of general and autonomous syntactic rules.
b. A great number of idioms can be attributed an internal semantic structure
which makes them semantically motivated and/or analysable. Such idioms
do not constitute semantic units and can therefore be processed compositionally. Idiom variability is a reflex of the internal semantic organisation
of these constructions.
While (a) can be called the traditional or orthodox view of idiom representation
and processing, (b) is generally termed the compositional view. To introduce
and review the two counterpositions, I will confront the orthodox view with
the compositional conception including both theoretical linguistic and psycholinguistic stances in the debate. While the linguistic approaches are basically
concerned with integrating idioms into a system of grammatical competence,
the psycholinguistic experiments primarily deal with idiom comprehension in
order to test hypotheses about idiom representation (cf. Häcki Buhofer 1999).
JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.2 (131-175)
. Idioms as semantic units – the orthodox view
The orthodox view of idioms as semantic units is best exemplified by early
transformational generative accounts of the linguistic status of idioms as well
as psycholinguistic correlates of these approaches. Transformational generative
frameworks, such as Weinreich’s (1969) idiom list model, are fundamentally
influenced by two central principles of this linguistic theory. First, the meaning of a grammatical construction is seen as determined by the principle of
compositionality. Second, syntax is regarded as the central component of linguistic structure. Given these principles, idioms, which are commonly seen as
semantically and syntactically idiosyncratic by definition, present a considerable stumbling block to the generative paradigm. The reconciliation of the
phenomenon of idiomaticity with the two fundamental theoretical principles
has therefore led to the characteristic description of idioms as complex phrases
with a unitary meaning.
.. Idioms as non-compositional phrases
The generative treatment of idioms is strongly determined by their characterisation as semantically non-compositional strings. The notion of noncompositionality derives from the fact that the meaning of these constructions
cannot be derived in terms of Frege’s principle of compositionality: “The
meaning of a sentence is determined by the meaning of its component parts
and the manner in which they are arranged in syntactic structure.” (O’Grady
et al. 1997: 260). To accord with the principle, idioms must be regarded as
linguistic exceptions that stand outside the grammatical norms of language:
Idioms are the ‘exceptions that prove the rule’: they do not get their meaning
from the meaning of their syntactic parts. If an idiom is treated as if it were
compositional, false predictions are made about its semantic properties and
relations.
(Katz 1973: 358)
Proceeding from the criterion of non-compositionality (cf. also Katz & Postal
1963: 275; Fraser 1970: 103), early generativist classifications of idioms proposed the distinction between “lexical idioms” and “phrase idioms” (Katz &
Postal 1963). These classes differ in terms of their internal syntactic complexity.
Lexical idioms are dominated by one of the lowest syntactic categories (N,
V, ADJ, etc.) and thus include exocentric compounds such as telephone, photograph or redneck. Since, from a syntactic point of view, lexical idioms thus
behave like ordinary lexical units, they do not present a serious challenge to
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.3 (175-224)
Chapter 2. Idiom representation and variation
generative theory: they can be inserted into the output of syntactic derivation
like ordinary lexical items. On the other hand, phrase idioms such as shoot
the breeze, take advantage or spill the beans have a complex internal syntactic
structure that is patterned on the phrase-level. Therefore, they cannot merely
be listed in the lexicon without avoiding unnecessary redundancy – given the
assumption that it would be uneconomic to list strings in the lexicon that
can be generated by the syntactic apparatus. Further, phrase idioms are less
exceptional from a syntactic perspective than they are semantically. A great
number of them follow regular syntactic patterns and all of them are subject to
inflectional agreement rules as reflected by (1)–(3).
(1) John and his friends are shooting the breeze.
(2) Advantage was never taken of the students.
(3) The beans were spilled, yesterday.
However, supported by the syntactic orientation of the generative paradigm,
it was soon recognised that, in most cases, the syntactic behaviour of idiomatic constituents is restricted. This is what Fraser (1970) calls frozenness.
Alternatively, the inability of phrase idioms to admit all grammatically possible transformations was described as transformational deficiency (Weinreich
1969: 47).
Given the preserved but restricted syntactic and morphological integrity
of phrase idioms, Katz and Postal (1963) suggested to divide the lexicon into
two parts: a lexical-item part and a phrase-idiom part. This allowed them to
assign unitary meanings both to higher level constituents, i.e. complex phrases
which maintain their syntactic character, and to terminal symbols, including lexical idioms. This idea was further developed by Weinreich (1969) in
his attempt to integrate phrasal idioms into the generative apparatus. Due to
their semantic non-compositionality, Weinreich assigns idioms to the lexical
component of the grammar. Their multi-word character combined with their
non-compositionality and transformational deficiency suggests two alternative
implementations in the dictionary (Weinreich 1969: 54):
a. “listed as a unit,” i.e. idioms are stored in the same way as all other lexical
items: the unit as a whole is given a separate phonological, syntactic and
semantic characterisation.
b. “listed in the dictionary constituent by constituent,” i.e. a given idiom is
divided into its lexical elements and a specific phonological, syntactic and
semantic description is attributed to each of them individually.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.4 (224-279)
Weinreich, however, rejects both of these treatments. Approach (a) must be
abandoned because it would force the grammarian to give up redundancy rules
in the phonological component. Second, the unitary treatment would lead to
ungrammatical tense marking or nominalisation, e.g. *shoot the breez-ed or
*shoot the breez-ing. In other words, it cannot account for the fact that the verb
shoot is inflected in the same way as in non-idiomatic expressions.
The constituent-by-constituent approach is able to compensate for the
phonological and grammatical disadvantages of the unitary treatment. The
individual components can be identified as verb, article and noun and their
phonological properties and inflectional specificities can be directly associated with the lexical constituents without creating any redundancy. However,
Weinreich (1969: 56) rejects this compositional solution on semantic grounds:
But this solution, too, has crass disadvantages, this time on the semantic level.
[...]. Notice that the segmentation of the paraphrase “chat/idly” is arbitrary
in relation to the idiom itself. Why not “chat idly/Ø”? The expression is idiomatic on all counts: the selection of unique senses of shoot and breeze is two
directional; it is determined by specific morphemes, as synonym tests prove
(thus, fire at the breeze or shoot the wind do not work); and the subsenses are
in a suppletive relation, since “chat idly” shares no semantic components of
any interest with “fire a projectile at” or “light wind.”
To work out a compromise that can accommodate the semantic unity of
idioms with the grammatical and phonological autonomy of the individual
constituents, Weinreich decides to adopt Katz and Postal’s (1963: 277) earlier
suggestion to slightly modify linguistic theory and to divide the lexicon into
a “lexical item part” and a “phrase-idiom part.” Weinreich (1969: 57) calls the
phrase-idiom part idiom list.
In the idiom list, each idiom is specified for its phrasal information and
its sense. That is, each entry contains specifications for obligatory operations
(e.g., shoot the breeze must always take an animate plural as its subject) and
inadmissible transformations (e.g., – passive, – of-nominalisation). The semantic characterisation is always unitary. In short, the idiom-list approach
maintains the grammatical status of the individual constituents, while characterising idioms as long words on the semantic level. Idiomatic frozenness, the
transformational deficiency of these formally-phrase/semantically-unit strings,
is specified as lexical information that has to be known. In other words, transformational deficiency is not characterised relative to the rule component of
the grammar, but it is treated as lexical knowledge.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.5 (279-323)
To integrate the list into the global generative apparatus, Weinreich
(1969: 58) devises a matching procedure – the idiom comparison rule. The idiom comparison rule is an instruction to compare literally processed material
with idioms that are stored in the list. Feeding on the idiom list, the idiom
comparison rule checks randomly generated terminal deep-structure strings
for their idiomaticity. If the rule finds a matching string in the idiom list –
this is the case when all specifications for the given entry are satisfied – it optionally replaces the semantic features of the original literal sentence by the
idiomatic meaning.
Apart from preserving the phrasal characteristics of idioms and from
avoiding redundancy, Weinreich (1969: 60) judges his approach to be of advantage with regard to idiom semantics:
It further wipes out any expectation of syntactic isomorphism between an
idiomatic expression and its paraphrase; the syntax of the sense description of an idiom list entry need not correspond to the formal syntax of the
expression itself.
In other words, Weinreich rejects the possibility of explaining the syntactic behaviour of idioms in terms of semantic regularities. In doing so, he underlines
the semantic irregularity of idioms and petrifies the orthodox view of idioms
as irregular semantic units in his system.
.. Direct look-up models – the psycholinguistic correlates of the
orthodox view
The orthodox view of idioms as semantically non-compositional, complex
phrases is implied in psycholinguistic models that treat idioms as word-like
lexical units to model idiom comprehension processes. Glucksberg (1993: 4)
therefore calls such models “direct look-up models.”
Weinreich’s transformational generativist model of idiom representation
provides the theoretical basis for Bobrow and Bell’s (1973) psycholinguistic
idiom-list hypothesis. The hypothesis postulates that a linguistic sequence containing a potential idiom is first interpreted literally.3 If the literal interpretation
of the string does not contradict the context, the comprehension process is
completed. But if the literal interpretation is contextually defective, the idiomatic meaning is activated by retrieving it from the mentally represented
idiom-list through direct look-up. Bobrow and Bell’s model is based on the
plausible assumption that it is impossible to identify a linguistic construction
as an idiom on the basis of its first constituent only. In other words, the model

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.6 (323-413)
assumes that a construction’s idiomaticity can only be realised if it cannot be
understood as a standard literal expression.4
Plausible as it may sound, the idiom-list hypothesis can be falsified on the
basis of psycholinguistic time-measurement experiments. Thus in the more
general context of figurative-language comprehension, Ortony et al. (1978)
show that potentially ambiguous idioms are generally processed faster when
used idiomatically, whereas literal uses slow down processing speed. This result is overwhelmingly confirmed by other psycholinguistic experiments (see
Swinney & Cutler 1979; Estill & Kemper 1982; Glass 1983; Gibbs & Gonzales
1985; Schweigert 1992; McGlone et al. 1994).
Psycholinguistic measurements of processing times thus suggest a figurativefirst conception of idiom comprehension. The data indicate that the idiomatic
meaning may be directly retrieved from memory before the literal processing of the idiomatic construction is completed. This view, for instance, finds
support in Gibbs’s direct-access hypothesis (Gibbs 1980, 1985, 1986).5 From a
linguistic point of view, this approach, at first sight, seems more appropriate
than the literal-first model because it takes into account that idioms are lexicalised constructions which should therefore allow for direct meaning retrieval
as is the case with other lexical units. In this sense, idioms can be regarded as
long words. Accordingly, literally used idioms are seen as special cases which
demand more complicated comprehension processes. This interpretation is
considered to explain the fact that understanding is faster when idioms are
used idiomatically. The direct-access view is also supported by Botelho da Silva
and Cutler’s (1993) results, which show that ill-formed idioms are processed
equally fast as well-formed idioms.
While these literal-first and the figurative-first models of idiom comprehension differ in their views of how the idiomatic meaning is activated, they
nevertheless share the same basic view of the mental status of idioms. Both
models imply that idioms are semantic units and must thus be attributed the
psycholinguistic status of lexical items. Bell and Bobrow’s model imports this
postulate by adopting Weinreich’s conception of an idiom-list in which idioms are stored as semantic units. The direct-access hypothesis contains this
claim more explicitly by equating idiom comprehension with word comprehension. The long-word view is most explicitly supported by Swinney and
Cutler’s (1979) lexical representation hypothesis. This model differs from the
direct-access model and the idiom-list hypothesis in assuming that the literal
and the idiomatic meaning are first processed in parallel with the literal analysis
being suppressed once the idiomatic meaning is stipulated.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.7 (413-429)
The configuration hypothesis provides a sophisticated alternative to these
direct look-up models without, however, qualifying the semantic description
of idioms (see Cacciari & Tabossi 1988; Tabossi & Zardon 1993, 1995). The
configuration hypothesis claims that idioms are not stored as simple formmeaning associations, but that they constitute complex arrangements of single
words. According to this model, idiom look-up only begins when the idiom is
itself recognised as a configuration, i.e. as a linguistic unit that is composed of
simpler lexical elements. A central claim of this model is that every idiom contains one or more lexical “keys” (Cacciari & Tabossi 1988: 678). A key works
as a kind of mental signal that makes the hearer evoke the idiomatic configuration as a whole, which leads to the activation of the idiomatic meaning.
The recognition of the key therefore marks the qualitative switching point between the idiomatic and the literal interpretation of an idiom. Before the key
is heard, the hearer tries to interpret the idiomatic string according to its literal meaning. But as soon as the key is recognised, the idiomatic meaning
can be activated. In other words, the key-constituents work as trigger-points
that must be recognised for the idiomatic configuration to be evoked. Consequently, the major factor causing the switch from the literal to the figurative
meaning is the position of the keys (Tabossi & Zardon 1993: 153–157). Very
often the key does not appear at the beginning of the idiomatic construction.
It can appear relatively early as in: when in Rome (do as the Romans do), intermediately, as with scream blue (murder), or late, as in be in seventh heaven.
Therefore, with a great number of idioms, the figurative meaning cannot be directly activated. However, in contrast to the literal-first hypothesis, the hearer
neither has to wait until the whole string is processed: as soon as the key is
heard, the figurative meaning can be switched on. A further important insight
of the psycholinguistic experiments related to the configuration hypothesis
is that the processing of an idiom’s literal meaning is not terminated after
the idiomatic meaning has been stipulated (see, e.g., Cacciari & Glucksberg
1991: 219–220). With the recognition of the key(s), the literal meaning is only
pushed to the background of semantic activation while the idiomatic meaning
becomes highly prominent. The impossibility to suppress the idiom’s literal
meaning is revealed by a great number of psycholinguistic studies (see Cacciari
& Glucksberg 1991; Blasko & Connine 1993; Colombo 1993; Flores d’Arcais
1993; Tabossi & Zardon 1993; McGlone et al. 1994; Cacciari 1993; Titone &
Connine 1999; Giora & Fein 1999). Thus, the configuration hypothesis can be
regarded as a more sophisticated compromise between the literal-first and the
figurative-first hypotheses.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.8 (429-467)
By emphasising the importance of the constituent meanings for the activation of the idiomatic meaning, the configuration hypothesis rejects the idiomsas-long-words view. This is in accordance with Weinreich’s account of idioms
as complex phrases. The configuration hypothesis provides compelling psycholinguistic evidence for the fact that although idioms are lexicalised phrases,
this does not prevent them from unfolding their structural complexity in actual
processing. By accounting for the fact that the idiom-constituents maintain
their lexical autonomy, it provides a basis for explaining their morphosyntactic flexibility. However, this strength is also a weakness. Since all idioms can
be attributed the status of phrasal word configurations, the hypothesis cannot
explain which idioms are systematically transformable and which ones are not.
This leads us to the problem of explaining idiomatic variability on the basis of
the orthodox view.
.. Consequences of the orthodox view for the explanation
of syntactic flexibility
In Weinreich’s syntax-based approach to idiom representation and derivation,
the stipulation of an idiom’s variation-potential remains purely descriptive and
without any explanatory power: an idiom’s transformational deficiency has to
be known as part of the given lexical knowledge.
Fraser (1970) tries to approach the transformational potential more systematically. Applying his concept of ‘frozenness,’ he develops a “frozenness
hierarchy” that is supposed to show that transformational restrictions apply
to idioms in terms of a seven-level scale (Fraser 1970: 39). Although Fraser’s
hierarchy can be used to describe the transformational deficiency of idioms
more systematically, the model has two major shortcomings. First, it is neither based on nor tested against empirical data. Second, and most importantly,
Fraser does not provide any explanation of why certain idioms are transformationally more deficient than others (this point of criticism is also raised
by Burger 1973: 70 and Schenk 1995: 255). Fraser’s non-explanatory treatment
idiom variability is a direct consequence of his syntax-based approach to the
phenomenon.
Following Burger (1973: 77), one can claim that any syntax-based approach
to idiom transformation is doomed to failure (see also Dobrovol’skij 1999b):
Ein syntaktisch fundiertes (Standard-)Modell Chomskyscher Art wäre auch –
wie oben angedeutet – höchstens geeignet, eine Deskription solcher Defekte
zu liefern; keineswegs könnte in diesem Rahmen eine Begründung für die jew-
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.9 (467-540)
eilige Beschränkung gegeben werden. Alle, die in den transformationellen Defekten eine charakteristische Eigenschaft der Idiome sahen, mussten zugeben,
dass zumindest in vielen Fällen der syntakische Defekt eine Konsequenz der
semantischen Beschaffenheit des Idioms ist (auch Fraser kommt um diese
Einsicht nicht herum).
Arguing in the same vein, Chafe (1968) regards the very existence of the phenomenon of ‘idiomaticity’ as a proof of the inadequacy of the transformational
paradigm. In his view, the transformational generative framework is incapable
of explaining the very existence of idiomaticity, which must therefore be considered an “anomaly in the Chomskyan paradigm” (Chafe 1968: 110–111). In
contrast, Chafe’s alternative model places semantics at the heart of the generative process. Accordingly, Chafe approaches the syntactic behaviour of idioms
in semantic rather than purely syntactic terms: transformational deficiency
does not have to be characterised superficially as being controlled by lexical
knowledge, but can be explained (at least to some extent) as a reflex of an
idiom’s semantic structure. For instance, the idiom kick the bucket cannot be
passivised because its meaning corresponds to the meaning of the intransitive
verb die. Thus, since intransitive verbs cannot adopt the passive voice, the idiomatic meaning of kick the bucket blocks this syntactic transformation on the
semantic level. Moreover, modifications of the NP the bucket are impossible
because this phrase does not have a correspondent on the level of the idiomatic
meaning (see Chafe 1968: 122).
Very similar approaches to explaining transformational deficiency are also
proposed by Newmeyer (1974) and Burger (1973: Chapter 5). In their views,
a syntactic transformation such as passivisation can only be applied to an idiomatic construction if both its idiomatic meaning and its literal meaning are
open to passivisation. Thus, the string pull someone’s leg can be passivised because both the literal verb pull and the idiomatic meaning ‘to tease’ allow this
voice. In contrast, blow one’s top cannot be passivised because the idiomatic
meaning ‘be very angry’ describes an intransitive process.
On balance, Chafe’s, Newmeyer’s and Burger’s semantic approaches to the
representation and syntactic behaviour of idioms offer an explanatory basis to
account for transformational deficiency. In contrast, syntax-based frameworks
as the ones proposed by Weinreich and Fraser remain purely descriptive and
therefore work as classificatory tools rather than adequate theoretical models
to capture the systematic variational behaviour of idioms.
But Weinreich’s approach is characteristic for the generativist treatment of
idioms because it is primarily based on the notion of ‘non-compositionality’.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.10 (540-567)
The non-compositional view, which characterises idioms as “anomalies of
interpretation” (Nunberg et al. 1994: 505), is predominant and common in
most generative approaches to the linguistic status of idioms (see also Chomsky 1980; Radford 1997). As a consequence, the semantic structure of idioms
(their potential metaphoricity or different types of semantic specialisation) is
generally left unanalysed in these frameworks. Given the long-lasting predominance of the Chomskyan paradigm in linguistic theory, the rejection of the
existence of any internal semantic organisation of idiomatic strings has exerted a strong influence on their perception as anomalous, exceptional and
unsystematic linguistic phenomena (cf. Keil 1997: 52). The orthodox idiomsas-semantic-units view must be seen as a logical consequence of the syntaxcentristric and bottom-up conception of linguistic composition advocated in
transformational generative grammars. The atomistic and unidirectional view
of linguistic composition unavoidably leads to the restrictive equation of noncompositionality with semantic unanalysability, i.e. semantic unity. Indeed,
after a short discussion of the concept, Weinreich (1969: 76) explicitly rejects
the notion of motivation: “In short, I feel that the relation between idiomatic
and literal meanings is so unsystematic as to deserve no place in the theory.”
Weinreich (1969: 76–77) further justifies this rejection by arguing that there is
no “grammar, [...] of functional extension;” in addition, he claims that “semantic rules to connect basic with derived meaning, [...] are chimerical.” However,
as we have seen, Weinreich’s loyalty to the basic principles of the transformational generative paradigm leads to a dead end with regard to the explanation
of idiom variability.
Even Chafe, who regards the syntax-based transformational generative
paradigm as falsified by the very existence of idiomaticity, does not question
the restrictive identification of idiom meaning with word meaning:
[...] the meaning of an idiom, [...] is not some amalgamation of the meanings
of the parts of the structure. Rather, the meaning of an idiom is comparable
to the meaning of a single lexical item.
(Chafe 1968: 111)
Thus, although Chafe escapes the trap of syntax-centrism and is capable of
revealing a number of regularities in idiom transformation, his insistence on
the definition of idioms as semantic units leaves many idiomatic stones unturned.6 More specifically, the orthodox view cannot explain actual variational
data such as the following:
(4) All this has upset the applecart of the relation of fertility to prosperity.
(EDK: 1433).
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.11 (567-620)
(5) Then, in the 1960s, the bull of controversy was grasped firmly by the horns
as the term ‘Political Education’ came into fashion. (APE: 456).
(6) ‘The Government has shown in the past that it is willing to grasp nettles
that others have shied away from, and I have no doubt we shall grasp this
one if we have to.’ (A2E: 12).
(7) Ever since Mr Gorbachev attended our G7 meeting last year and Mr Lamont went to inspect the Russian chaos, Britain has effectively adopted the
Soviets and we will be pressing hardest to help – even George Bush is now
jumping on this bandwagon. (AL2: 215).
(8) You know I mean can you think of anything more ridiculous with there
you go they probably haven’t got any other leg to stan I mean that’s no leg to
stand on but they’ve got nothing else to say really you know. (HUX: 114).
(9) This way of thinking has to be one of the blinder alleys that we have been
led up by psychoanalysis. (A6D: 158).
How can idiomatic constituents be postmodified by an of-complement (4),
postmodified by an of-complement within a passive construction (5), pluralised and postmodified by a restrictive relative clause (6), used with a demonstrative determiner (7), qualified (8) or used in the comparative form (9),
if these constituents do not possess some autonomous semantic value? In
terms of a conception of idiom semantics which claims universal semantic
unity for all idiomatic strings, these data are inexplicable. Consequently, the
semantic-unity view has negative consequences for the explanation of the syntactic behaviour of idioms. Since the idiomatic meaning cannot be related to
the constituent parts, the syntactic behaviour of idioms cannot be explained in
semantic terms. This, exactly, is the weak spot against which the compositional
view is directed.
. The compositional view
Reviving ideas developed in earlier accounts (Nunberg 1978; Wasow et al.
1983; Gazdar et al. 1985), Nunberg et al. (1994) strongly attack the generative conception of idioms as non-compositional semantic units and propose
an alternative conception which refines the semantic characterisation of idiomaticity.
Nunberg et al.’s criticism is explicitly directed against a number of widely
accepted syntactic arguments in the generative paradigm that strongly feed on

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.12 (620-679)
the semantic-unity view. One of these arguments, for instance, claims that
syntactically-transformed idioms prove the existence of autonomous syntactic transformations. Nunberg et al. (1994: 510, numbers adapted by author)
summarise this argument on the basis of the following examples:
(10) a. Pat spilled the beans.
b. The beans were spilled by Pat.
(11) a. The cat is out of the bag.
b. The cat seems to be out of the bag.
If surface structures are generated directly, so the argument goes, then the a
and b sentences must be distinct idioms. By contrast, a transformational analysis of the b sentences allows each idiom to be listed only once. The idioms
can be inserted into D-structures in their contiguous (a) forms and transformationally broken up, yielding their b forms. This simplifies the grammar and
captures a generalization.
Nunberg et al. correctly observe that this argument can only be maintained as
long as idioms are regarded as arbitrary form-meaning associations. Thus, if it
can be proved that the strict non-compositionality conception is wrong, then
idioms provide no evidence in favour of syntactic transformations.
.. A broader view of idiom semantics
Nunberg et al.’s alternative characterisation of idiom transformations is based
on a more precise description of idiom semantics. According to the authors,
the traditional description of idioms as non-compositional semantic units is
too simplistic. Nunberg et al. (1994: 495) rightly interpret the traditional notion of idiomatic compositionality in terms of “whether a native speaker would
recover the sense of the idiom on hearing it in an ‘uninformative context’.” Although the traditional conception of idiomaticity seems to be straightforward
at first sight, Nunberg et al. (1994: 495) reject it for its unnaturalness:
[...] this definition is actually quite unnatural and false to the way most idioms
are presumably learned: it would require the listener to ask explicitly what each
idiom meant, which is relatively uncommon.
They exemplify their claim with reference to the Spanish idiom tener una
lengua de trapo which is non-predictable in isolation, but becomes fairly transparent once it appears with appropriate contextual information: no vamos
a pasar un dia tranquilo, Juan tiene una lengua de trapo, i.e. Juan is very
talkative. In other words, while appearing outside a usage-context a great
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.13 (679-726)
number of idioms are clearly non-predictable if their meaning is not known.
However, when appearing in a usage-context, they are often fully comprehensible and motivated.7 In other words, generative conceptions of idiom
semantics blindly equate the semantic structure of idioms with bottom-up
non-compositionality. Nunberg et al. challenge this orthodox view by considering the additional possibility of top-down analysability. While the traditional conception of non-compositionality does not allot the existence of
post-compositional re-analysis, the authors argue that this process cannot be
simply ignored:
To justify this claim [that idioms have a simplex semantic representation] for
any particular case, however, it has to be shown not just that the meaning of
the idiom could not be predicted on the basis of a knowledge of the meanings of its parts, but that once the meaning of the idiom is known (say by
hearing it used in a sufficiently informative context), it cannot be devolved on
the constituents of the expression. And this is not entailed by simple nonpredictability.
(Nunberg et al. 1994: 496)
By postulating the possibility of the top-down semantic re-analysis of a derivationally non-compositional construction, they argue for the existence of semantically compositional idioms.
Nunberg et al. (1994: 496) use the term idiomatically combining expression to refer to semantically compositional idioms and define this idiom type
as follows:
To say that an idiom is an idiomatically combining expression is to say that
the conventional mapping from literal to idiomatic interpretation is homomorphic with respect to certain properties of the interpretation of the idiom’s
components.
This is illustrated on the basis of the idiom pull strings. This expression can
be regarded as semantically compositional because it is possible to devolve
the idiomatic meaning homomorphically on the individual constituents. In
the context of the phrase, strings denotes something like ‘personal connections’ and pull means ‘exploit (the connections).’ Thus both linguistic constituents can be attributed a separate semantic interpretation relative to the
idiomatic meaning on the basis of a top-down analytical process. These meaningful correspondences can be established even though the idiomatic meaning
is not predictable through bottom-up composition. The expression combines
idiomatically, since the two constituents can only be attributed individual figurative meanings when they are co-selected. In other words, pull and strings do
not possess autonomous figurative senses, but their separate, derived mean-

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.14 (726-788)
ings are activated when they appear within the conventionalised collocational
pattern evoked by the expression pull strings. I will call this semantic characteristic analysability. In short, analysability captures the fact that with some
idioms elements of the idiomatic meaning can be distributed over the literal
constituents.
Nunberg et al. emphasise that not all idioms can be described as idiomatically combining expressions. Idioms such as kick the bucket and shoot the
breeze or saw logs are not semantically decomposable, even though some of
them (e.g., saw logs) are motivated. Nunberg et al. (1994: 497) call the class of
non-compositional idioms idiomatic phrases.8
.. The syntactic behaviour of idiomatically combining expressions
With the distinction between idiomatically combining expressions and idiomatic phrases the class of idiomatic constructions that have the status of
semantic units is reduced to a much smaller set. This is directly relevant for
the explanation of the syntactic behaviour of idioms:
[...] there are compelling reasons to believe that the majority of phrasal idioms
are in fact semantically compositional, and that the very phenomenon of idiomaticity is fundamentally semantic in nature. Much of the literature on the
syntax of idioms is thus based on the misconception that no such semantic
compositionality exists. [...] the various constraints on idioms that have been
observed are better explained in consequence of either semantic properties or
else broader tendencies in the figurative use of language.
With reference to the class of idiomatically combining expressions, Nunberg
et al. are able to explain a great number of syntactic modifications without
having to rely on the postulation of autonomous syntactic transformations or
to disclaim such modifications as wordplay as it is often done in the literature
(cf. Schenk 1995; Mel’čuk 1995). For instance, adjectival premodification as
in leave no legal stone unturned can be explained on the basis of the phrasebound figurative meanings of the constituents. Since stone can be attributed the
metaphorical meaning ‘measure,’ the modification of this constituent makes
immediate sense. If this and other transformations are applied to idiomatic
phrases, however, only the literal reading is possible:
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.15 (788-824)
a. Quantification:
b. Topicalisation:
c. VP-ellipsis:
d. Anaphora:
e. Passivisation:
touch a couple of nerves
*he kicked a couple of buckets
Those strings, he wouldn’t pull for you
*Those logs, he didn’t saw
My goose is cooked, but yours isn’t.
*They boys only wanted to shoot the breeze, and it was.
Kim’s family pulled some strings on her behalf, but
they weren’t enough to get her the job.
*John seemed to kick the bucket, but it wasn’t there.
The beans were spilled by John.
*The dust was bitten by John.
(all examples taken from Nunberg et al. 1994:
500–503).
As this comparison suggests, it becomes possible to explain the syntactic flexibility of idioms in semantic terms:
We are thus in effect proposing to explain a variety of ‘transformational deficiencies’ of idioms by positing a bifurcation between idiomatic phrasal constructions and idiomatically combining expressions, [...]. [...] this approach
predicts a strong correlation between semantic analyzability and ‘transformational productivity’.
The absence of the idiomatic interpretations for syntactically altered idiomatic
phrases supports the predicted correlation between semantic analysability and
syntactic flexibility. This insight justifies the authors’ claim that the behaviour
of idioms cannot be taken as a measure to defend the autonomous existence of
syntactic transformations.
In sum, Nunberg et al.’s theoretical account of the linguistic status of
idioms is based on a rejection of the semantic unity view. Introducing
the distinction between semantically compositional idiomatically-combiningexpressions and semantically non-compositional idiomatic-phrases, they establish a refined conception of idiom semantics on the basis of which the
syntactic variability of these constructions can be predicted. However, as will
be seen in the next sections, the idioms-as-compositional-strings conception
has not remained undisputed.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.16 (824-885)
.. Criticism against the compositional view
A number of more recent linguistic studies have tried to put forward evidence
against the compositional view. Most of these studies stem from grammatical
models connected to the generative paradigm.
Schenk (1995), who argues from the context of the Rosetta grammar, defends the idioms-as-semantic-units view by claiming that many aspects of the
syntactic versatility of idioms can be explained on the basis of the distinction
between syntactic operations that are only applicable to meaningful expressions and others which are applicable to both meaningful and semantically
empty expressions. Thus, Schenk attempts to show that idioms only allow operations of the second type, which he takes as an indirect proof of their semantic non-compositionality (Schenk 1995: 253). Schenk discusses three syntactic
operations: topicalisation, raising and control. While topicalisation and control
can only be applied to meaningful constituents, Schenk argues that raising is
also possible with empty constituents such as existential there and prop it (the
following examples are reproduced from Schenk 1995: 259–264, the numbers
have been adapted by the author):
a. Topicalisation:
(12) a. He believes there to be unicorns.
b. *There he believes to be unicorns.
(13) a. John spilled the beans.
b. *The beans John spilled.
b. Raising:
(14) a. John believes there to be ghosts.
b. There tends to be a lot of rain in California.
(15) a. John believes the beans to be spilled.
b. The beans tend to be spilled.
c. Control:
(16) a. John instructs the dog to fetch the newspaper.
b. The dog tries to eat.
(17) a. John instructs it to fetch the newspaper.
b. It tries to eat.
(18) a. *John instructs there to be people.
b. *There try to be people.
(19) a. *John instructs the beans to be spilled.
b. *The beans try to be spilled.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.17 (885-942)
At first sight, these examples, which Schenk proposes on the basis of introspection, seem to support his argument. However, his analysis is problematic for a
number of reasons.
First, Schenk claims that raising is possible for meaningless elements. However, in my view this operation is not applicable to fully opaque expressions
such as kick the bucket or shoot the breeze:
(20) a. *Mary believes the bucket to be kicked by John.
b. *The bucket tends to be kicked by John.
(21) a. *John believes the breeze to be shot
b. *The breeze tends to be shot.
The impossibility of (20) and (21) runs counter to Schenk’s claim that all idioms have the same semantic status, i.e. the status of non-compositional units.
In other words, his basic introspective argument is severely challenged by such
controversial data. In a similar same vein, Schenk is also criticised by Abeillé
(1995: 18), who recognises that more syntactic variations are possible than
predicted in his framework:
Few of these predictions are truly borne out upon careful examination, and, as
shown by Newmeyer (1974) or Wasow et al. (1983), it is hard to find syntactic
rules that would not apply to any idiom.
For instance, the control sentence Some strings are harder to pull than others
seems to be semantically well-formed.
Further, in my opinion, (19a–b) are flawed independent of the syntactic
operation discussed because they are semantically deviant on the literal and
the idiomatic level. Both beans and secrets are inanimates. Therefore, they can
neither be instructed nor can they try to do something. It is due to these semantic constraints that the control operation cannot be applied to spill the beans,
rather than the semantic non-compositionality of the expression.
Finally, one must question the general validity of the syntactic analysis proposed by Schenk. With reference to their discursive functions idioms are highly
unlikely to appear in the syntactic frames discussed by Schenk. In other words,
the awkwardness of some operations, even those that Schenk regards as possible for all idioms, is also caused by discursive and stylistic factors rather than
the general applicability of syntactic operations.
Nicolas (1995) analyses prenominal modification of idiomatic NPs. He
rejects the view that such idiom-internal NPs have identifiable meanings as
postulated by the compositional view of idiom semantics. In contrast, his central claim is that internal modification must be systematically interpreted as

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.18 (942-998)
external, adverbial modification of the whole idiom. Consequently, he argues
that internal modification does not verify the compositional view of idioms,
but that such syntactic variation can be integrated into a non-compositional
view of idiom semantics. Nicolas bases his argument on earlier ideas developed
by Verhagen (1990). Verhagen (1990: 7, quoted according to Nicolas 1995: 236)
proposed that:
[A]n adverb which is morphologically related to an adjective Adj and which
modifies an SVC phrase can play the same role as the adjective Adj which
modifies the ... [predicative noun] in that SVC phrase.
1. The lady quickly carried out an attack [...] on the tramp.
2. The lady carried out a quick attack [...] on the tramp.
In other words, NP-modifying adjectives in V-NP structures can perform the
same semantic function as an adverb that modifies the whole phrase. Suggesting the same idea in the context of idiom-internal modification, Arnold and
Sadler (1989) introduce a distinction between mere syntactic modification and
normal semantic modification. Thus, while (22a–b) belong to the first type
and can therefore be regarded as synonymous, (23a–b) reflect true semantic
modification, i.e. the two sentences have different interpretations:
(22) a. He came a nasty cropper yesterday.
b. He failed nastily yesterday.
(23) a. They bought a wonderful dress yesterday.
b. *They bought a dress yesterday, wonderfully.
Following these ideas, Nicolas (1995: 237) labels the former type of modification “syntactic modification” while the normal type is termed “semantic
modification.” This distinction is identical to what Ernst (1981: 52) calls “semantically external” modification as opposed to “semantically internal modification” of an idiom. Since Nicolas advocates a non-compositional view of
idiom semantics, he claims that well-formed idiom modification can only be
syntactic. Semantic modification is only possible as idiomatic wordplay.
Nicolas scrutinises his claim by checking variations for 75 V-NP idioms on
the basis of data extracted from a 50-million-word newspaper corpus in combination with an introspective method. To analyse the data, Nicolas develops
a systematic typology of syntactic modification which is based on an adverbial
classification system (see Nicolas 1995: 238–239).
For instance, a viewpoint modification is present in the varied expression
call the political tune, which can be rendered as ‘dominate politically speaking.’
Or manner modification occurs with make rapid headway ‘progress rapidly’
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.19 (998-1049)
(Nicolas 1995: 239). Applying the classification-system to the variation data,
Nicolas finds out that at least 90% of his idiom sample show tokens with internal NP-modification; viewpoint modification being the most frequent type
covering about 85% of all variation tokens. Nicolas (1995: 249) therefore concludes: “Idiom-internal modification of the type examined semantically modifies the idiom as a whole; no cases of modification that forced a genuinely
internal (NP-modifying) interpretation were found.”
Nicolas seems to provide compelling evidence in favour of the noncompositional idiom conception. However, two points of criticism can be
raised against his analysis. First, Nicolas’s strategy to re-interpret internal adjectival modification in terms of external adverbial modification raises the
question of why the adjectival type is so predominant in the first place. His
basic argument, which is mainly designed to find a way to defend the noncompositional idiom-view, is not concerned with such an explanation. On the
other hand, Nicolas’s approach to internal modification does not preclude a
compositional analysis of this phenomenon. Thus, for instance, call the political tune can be equally interpreted as an instance of semantic modification in
the sense of ‘determine the political activities; take the political decisions.’
Second, the focus on internal NP-modification excludes other types of adnominal modification like of -complements or determiner modification. Nicolas seems to be aware of this problem and points to controversial data in
his corpus:
(24) There was only one man on the stage last night who can fill that bill (taken
from Nicolas 1995: 243).
(25) The deputy leader, Roy Hattersley, raised Labour eyebrows (taken from
Nicolas 1995: 243).
Nicolas (1995: 243) tries to interpret these modifications according to his
framework: “fill the bill from that viewpoint” for (24) and “surprised Labour,
i.e., the Labour party” for (25). In my view, these interpretations are not
fully satisfactory. In (24) the demonstrative pronoun clearly points to a specific referent that has already been mentioned before. Therefore, it makes far
more sense to interpret the variation compositionally as ‘qualify for that issue/situation.’ The nominal premodifier Labour in (25) seems to be metonymically related to the action of raising eyebrows. This idiom is transparent in the
sense that the bodily reaction described in the idiom stands for the emotional
cause of surprise. Therefore, the modifier Labour seems to be related to the
members of the Labour party as such, rather than the Labour party in general.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.20 (1049-1104)
In sum, Nicolas’s investigation seems to pose a serious challenge to the
compositional view of idiom semantics. Nevertheless, Nicolas’s argumentation and his empirical results cannot falsify this conception. Rather, a more
comprehensive empirical analysis including alternative types of adnominal
modification is needed.
Abeillé (1995) proposes a formalisation of the lexical status of idioms
within the framework of Lexicalised Tree Adjoining Grammars (TAGs).
Abeillé’s model is based on an extensive analysis of 2000 French V-NP idioms.
On the basis of this corpus, she analyses different syntactic operations: raising,
passive, cleft sentences and adnominal modification. Discussing the results of
her analysis Abeillé (1995: 24) concludes that
semantic factors obviously play an important role in explaining the syntactic
flexibility of idiomatic expressions. However, Wasow et al.’s proposal oversimplified the problem, and a systematic decomposition of the idiomatic meaning
in not necessarily helpful.
More specifically she observes that
1. “there is no clear correlation between passive and semantic analyzability”
(Abeillé 1995: 24). To substantiate this point she claims that the following
expressions are opaque, but can nevertheless take the passive: prendre le
taureau par les cornes (engl. take the bull by the horns, mettre des bémols (to
attenuate), or avaler des couleuvres (to accept an affront).
2. “there is no clear correlation between NP extraction [topicalisation, clefting] and semantic decomposability” (Abeillé 1995: 24). This said, Abeillé
emphasises that clefted idiomatic constituents must be accompanied by an
adnominal modifier as in:
(26) C’est le taureau des privatisations qu’il a pris par les cornes de
l’actionariat populaire (taken from Abeillé 1995: 20).
According to the author, the inserted element des privatisations can be
interpreted as an argumental adjective.
3. “the insertion of quantifiers and relative clauses is not a clear sign of compositionality” (Abeillé 1995: 24). Rather she sees the relativisation of an
idiom NP as determined by the variability of the determiner of the NP.
Abeillé (1995: 24) therefore draws a careful conclusion:
I do not mean that no regularities can be found in the syntax of idioms, [...].
Neither do I mean that all idioms should be considered noncompositional.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.21 (1104-1157)
But I did not observe a clear correlation between such syntactic regularities
and compositionality.
As a consequence, she proposes her TAG analysis of idiom flexibility, which
primarily relies on a noncompositional view of idiomatic strings.
As a minor point of criticism one can follow Schenk (1995: 255) and point
out that Abeillé’s TAG approach to the syntactic flexibility of idioms remains
purely descriptive. One can oppose this criticism by emphasising that Abeillé’s
discussion should be appreciated for its unbiased approach with regard to the
compositionality vs. non-compositionality controversy. Nevertheless, I think
that her observations listed above can be readily accommodated to the compositionality view. A number of idioms that she regards as opaque and open
to passivisation are clearly analysable and transparent in my opinion. For instance, prendre le taureau par les cornes can be rendered as ‘tackle a problem
directly’ with taureau denoting the problem and prendre meaning ‘tackle.’ Consequently, this idiom can be passivised because taureau has its independent
metaphorical meaning in the context of the phrase. The same is true for the
cleft extraction and adnominal modification in (26). The constituent taureau
can be postmodified by the complement des privatisation due to its metaphorical sense. Thus, the problem referred to in the statement is specified as a specific
privatisation problem. It is only on the basis of this specification that it makes
sense to extract the NP taureau des privatisation and focus on it contrastively
by means of a cleft construction, as is correctly observed by Abeillé.
The same explanation holds for the correlation of NP relativisation with
the variability of the determiner. As Fellbaum (1993) shows, determiner variability in idioms is dependent on the compositional semantic structure of
these constructions. In such idioms, the determiner works as an indicator of
the noun’s referential status (Fellbaum 1993: 272). This is not true for noncompositional, unanalysable idioms (trip the light fantastic, buy the farm),
whose NPs are “nondenoting” (Fellbaum 1993: 279). Consequently the determiner cannot be altered in these expressions. Thus, given the dependence of
determiner variability on semantic compositionality, the correlation between
the relativisation of an idiom’s NP and the variability of the determiner observed by Abeillé must itself be interpreted as an effect of the analysability of
these constructions.
In sum, on the basis of empirical and introspective data two major points
of criticism against the compositional view have been raised. First, it is argued
that there are no clear correlations between syntactic flexibility in general and
idiom analysability. Second, some syntactic alterations like internal modifi-

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.22 (1157-1201)
cation and raising seem to be applicable to semantically empty constituents.
Consequently, some variation processes are assumed to be explicable without
reference to compositionality. In what follows, the compositional view will be
discussed in the light of its psycholinguistic correlates.
.. The psycholinguistic correlates of the compositional view
The decomposition hypothesis introduced by Gibbs and Nayak (1989) and Gibbs
et al. (1989) is based on the linguistic classification of idioms proposed by
Nunberg (1978). Since this study constitutes the historical source of Nunberg
et al.’s compositional view of idioms, the decomposition hypothesis can be
regarded as the psycholinguistic correlate of this linguistic model of idiom representation. Following the compositional view, the decomposition hypothesis
argues that a great number of idiomatic constructions are semantically decomposable or analysable with the specific meanings of their parts contributing
independently to their overall figurative meanings. Indeed, the results of relevant experiments show that the degree of semantic decomposition correlates
with comprehension speed:
These data suggest that people attempt to do some compositional analysis
when understanding idiomatic phrases. When an idiom is decomposable, subjects can assign independent meanings to its individual parts and will quickly
recognize how these meaningful parts combine to form the overall figurative
interpretation of the phrase. [...]. [Further,] the ability to assign the parts of an
idiom with independent meanings motivates why many readers view normally
decomposable idioms as being syntactically productive and lexically flexible
[...].
(Gibbs et al. 1989: 587)
Thus, the longer processing times for non-decomposable idiomatic phrases
can be regarded as a consequence of the impossibility to process these phrases
compositionally. The reaction times suggest that people usually try to subject
these constructions to a compositional analysis as part of the comprehension
process. For decomposable idioms, this compositional analysis supports the
process, whereas for non-decomposable ones it interferes with direct comprehension. Thereby, it is important to emphasise that the compositional analysis
does not apply to the literal meanings of the constituents, but that it refers
directly to the specialised idiom-bound figurative meanings of these words
(see Gibbs et al. 1989: 588–589). These specialised figurative senses of the constituents are claimed to be stored in the mental lexicon and triggered in the
idiomatic context. Decomposable idioms can therefore be regarded as lin-
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.23 (1201-1259)
guistic units that can be produced and comprehended in terms of semantic
(de)composition.
Like the configuration hypothesis, the decomposition hypothesis thus undermines the view of idioms as long words. However, it is important to emphasise that the two models differ with regard to how they attack the long-word
conception. While the configuration hypothesis highlights the importance of
the constituent meanings for idiom comprehension processes, the decomposition hypothesis focuses on the possibility of a structured relationship between
the overall idiomatic meaning and these literal meanings. In other words,
the configuration hypothesis underlines the actual composition of the literal
meaning in idiom comprehension, whereas the decomposition hypothesis regards idiom compositionality as a top-down process. The two models thus
highlight two alternative types of compositionality: literal compositionality
(idioms as word configurations) and figurative-literal decomposition (idioms
as analysable linguistic strings). Note that both of these compositionality types
do not directly correspond to the orthodox view of compositionality, which
implies the bottom-up derivation of the idiomatic meaning from the senses of
the lexical constituents.
With regard to syntactic variability, the decomposition hypothesis predicts
that decomposable idioms are more flexible than non-decomposable ones.
This view is supported by the results of Gibbs and Nayak’s (1989) experimental study, which tested native speaker intuitions about the analysability
and syntactic frozenness of idioms by means of a judgement task. Their results show that decomposable idioms allow more syntactic modifications than
non-decomposable ones.
Along similar lines, Glucksberg (1993) attempts to answer the question:
How can people determine the meaning of idiom variants? Glucksberg’s psycholinguistic model is based on the following argumentation: an occasional
idiom variant such as got up on the right side of the bed (derived from: got up on
the wrong side of the bed) is not stored in the mental lexicon but it is produced
through cognitive processing. Consequently, the meanings of the constituents
must be evoked to determine the variant’s meaning. According to Glucksberg
(1993: 10), there are two alternatives to explain how people arrive at this meaning: the “multistep model” and the “phrase-induced polysemy model (PIP).”
The sequential multistep model is in accordance with the view of idioms as linguistic units whose meanings are retrieved as semantic units through direct
look-up. Following this model, the meaning of the variant would be constructed by first recognising the novel idiom as a variant of a conventional
idiom, looking up the idiomatic meaning of the original idiom, then activat-

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.24 (1259-1298)
ing the constituent meanings of both idioms, identifying the correspondences
of those meanings to the original’s overall idiomatic meaning and inferencing
the overall meaning of the variant on the basis of these semantic representations. Glucksberg’s PIP model offers a much simpler alternative. Namely, it
starts from the assumption that the constituents of familiar and decomposable
idioms become polysemous, i.e. they acquire idiom-specific figurative senses
through their frequent evocation in the idiomatic context. For instance, the
verb spill in spill the beans has the additional word meaning reveal in the idiom context. This phrase-induced polysemy of the verb makes it possible to
comprehend the variant he didn’t spill a single bean by means of ordinary linguistic processing. Thus, the PIP model can be regarded as equivalent to the
compositional view.
The PIP model can account for experimental insights ascertained by a
study conducted by McGlone et al. (1994). The study revealed that the comprehension times for original, unmodified idioms (used according to their
base-form and prototypical syntactic frames) are processed faster (direct lookup) than idiom variants (linguistic processing). The same result is attested by
Botelho da Silva and Cutler (1993) and Van de Voort and Vonk (1995). But,
more interestingly, the study also showed that an idiom variant such as he didn’t
spill a single bean is processed as fast as the literal paraphrase he didn’t say a single word (McGlone et al. 1994: 171). Neither the direct-access nor the lexical
representation hypothesis are equipped to explain comprehension when idioms display their phrasal, compositional qualities as reflected in such instances
of semantic productivity. Additional assumptions are needed to account for
the result that idiom variants can be understood as rapidly as their literal paraphrases. The PIP-model offers an adequate framework to account for the latter
insight. Given the dual meaning of spill in the idiom context, the string can be
comprehended in terms of linguistic processing rather than inferencing. For
modified decomposable idioms, comprehension can thus be said to work in
the same way as ordinary sentence comprehension (Glucksberg 1993: 13).
In a nutshell, psycholinguistic experiments demonstrate that decomposable, i.e. compositional/analysable, idioms are more flexible (both lexically
and syntactically) than non-decomposable ones. These results thus support
the compositional view of idiom semantics, i.e. the existence of top-down
analysability.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.25 (1298-1356)
.. Towards a hybrid model of idiom representation and processing
Proceeding from his ideas on phrase-induced polysemy, Glucksberg highlights
the empirical fact that even nondecomposable and opaque idioms can be variable as illustrated by the following modifications (Glucksberg 1993: 13; see also
McGlone et al. 1994: 184; Glucksberg 2001: 72–73):
Speaker A:
Speaker B:
Speaker A:
Speaker B:
By and large, people are much better off now than they
were five years ago.
By and not-so-large. Have you seen the latest figures on
poverty and unemployment in urban areas?
Did the old man kick the bucket last night?
Nah, he barely nudged it.
These productive variations would be impossible, if idioms were processed
like long words. Rather, the variations further support the well-attested fact
that, one the one hand, the literal meanings of idiom constituents are activated when an idiom is processed and that, on the other, the literal meaning is
subject to a compositional analysis if it is well-formed (for the present examples this is only true for kick the bucket). Highlighting the accessibility of literal
compositionality,9 Glucksberg (1993: 20) goes as far as to claim that all idioms
maintain a minimal degree of compositionality: “Indeed, purely noncompositional idioms may not exist at all.”10 Thus, taking the accessibility of the literal
meaning level in idiom comprehension and production into account, the constituents’ word meanings can be said to exert a considerable influence on the
full idiomatic configuration. More precisely, the idiomatic meaning and the
semantics of the idiom constituents seem to constrain idiom use and comprehension jointly (see also Glucksberg 1993: 20). How, then, can the above
variations be explained if not by semantic decomposition?
Following Cacciari and Glucksberg (1991), Glucksberg (1993: 14) explains
these variations in terms of their communicative functionality: “When an
idiom’s constituents bear functional relations to the idiom’s meaning, then
operations such as quantification, antonymy, and negation will be productive provided that a plausible communicative intent can be inferred.” For the
above examples, a plausible functional relation between the variants and the
idiomatic meaning can be easily established. In the first example, the negation
of large (not so large) can be directly projected onto the semantically related
idiomatic meaning ‘to a great extent,’ while in the second one, the fine-grained
conceptual relations between kick and barely nudge are exploited and mapped
back onto the idiomatic meaning. Thus the potential for the semantic and dis-

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.26 (1356-1403)
cursive productivity of idioms and their syntactic and lexical flexibility can only
be explained if idiom constituents remain psychologically autonomous semantic units. Following Glucksberg, most idioms can be claimed to have at least two
active levels of semantic representation: their stipulated idiomatic meanings
and the literal meanings of their constituents. Further, some compositional
idioms can have phrase-induced meanings that make it possible to devolve
the idiomatic meaning on the constituents (cf. also Glucksberg 1993: 16–19).
When constructing the actual meaning of a sentence containing an idiomatic
expression, listeners must integrate these levels of meaning with the discourse
context. At this point, however, it remains unclear how this works.
Thus, psycholinguistic discussions of idiom variation and variability reinforce the results of the configuration and the decomposition hypothesis. Some
patterns of idiom variation cannot be explained if one does not assume some
degree of compositionality, i.e. semantic analysabilty, whereas others strongly
depend on the availability of the literal meaning, conceptual knowledge associated with this meaning and their evocation together with the idiomatic
meaning. Thus, psycholinguistic insights into idiom variation clearly favour
a comprehensive view of the semantic structure of idioms. In general, the
findings from psycholinguistic studies point to evidence that speak for both
non-compositional (meaning stipulation) and compositional (word configuration and semantic decomposition) pathways in idiom processing (cf. also
Burger et al. 1982: 187). What one misses at this point is an integrated model
that is able to provide a synthesis between these alternative processing routes.
In other words, more hybrid models of idiom processing are needed that are
able to integrate the well-attested results on idiom processing.
With regard to idiom comprehension, a hybrid view of idiom processing is suggested by advocates of simultaneous processing models (Gibbs 1990;
Schweigert 1992).11 According to the simultaneous processing view, both the
literal and the figurative interpretation of an idiom are processed in parallel. In
the receptive process all elements of a given expression are related to semantic
representations. These elements can consist of single words, idiom constituents
and keys, and, once it is recognised, the entire idiomatic meaning. Thereby,
the lexical status of each element (word meanings, composite literal meaning,
phrase-bound figurative meanings, the meaning of the entire idiomatic configuration) is part of the idiom’s semantic representation. If there are different
possible semantic representations for a constituent or constituent string, all of
them are potentially activated. Through the interaction of these semantic representations with contextual information, the suitable interpretation (literal or
idiomatic) is selected.
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.27 (1403-1451)
Since they take the full semantic potential of an idiomatic configuration
into account, simultaneous processing hypotheses can account for the contradictory results of time measurement experiments. If the two semantic levels are
processed in parallel, they can be regarded as presenting two possible interpretative alternatives. According to Gibbs (1990: 428) the simultaneous activation
of the literal and the figurative meaning therefore results in a ‘horse race’
between the two semantic levels:
However, people may process the literal meanings of idioms simultaneously
to determining its figurative meanings [...], resulting in a horse-race that facilitated participants’ responses to idioms over the time required to process
literal phrases where only one meaning is processed. This horse-race model
can be extended to account for the reading time advantages of decomposable over nondecomposable idioms by suggesting that the literal and figurative
meanings of decomposable idioms work together to facilitate people’s speeded
responses to these phrases. On the other hand, processing the literal meanings
of semantically nondecomposable idioms as [sic!] and interferes with people’s
comprehension of these phrases.
As Dobrovol’skij (1997: 17) points out, the outcome of the ‘race’ is determined
by various factors such as the semantic structure of the idiom as a whole
(improbable literal meanings, unique constituents, decomposable structure,
etc.), the participants’ knowledge of the idiom (its familiarity), the position
of the key elements and the specific speech context. Since the simultaneous
processing model offers a more holistic model of idiom comprehension, it
seems to be better equipped to account for different processing alternatives.
The simultaneous processing view therefore seems to offer a more adequate
framework from a cognitive-linguistic perspective than literal-first or figurative
first models.
A sophisticated view of simultaneous processing is also proposed by Titone
and Connine (1999). Their model is an attempt to integrate the decomposition
hypothesis and the configuration hypothesis. They argue that both the compositional and the non-compositional views of idiom processing are flawed
if they are considered in isolation. Therefore, they propose a hybrid view of
idiom representation and processing: idioms are stored both as unitary wordcombinations and compositional word sequences.
On the one hand, idioms may be compositional as described by Nunberg et al. On the other hand, they are word-like due to their conventionality:
“the association between the particular configuration of words and a specific
idiomatic meaning is highly overlearned” (Titone & Connine 1999: 1667). It
is this cognitive facet of conventionality which plays an essential role for the

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.28 (1451-1504)
processing of idioms: conventionalisation provides the basis for the direct stipulation of the idiomatic meaning. Since many idioms can be regarded as highly
conventionalised units, their meaning is very likely to be activated by direct
stipulation regardless of whether the string is decomposable, transparent or
non-decomposable (see also Giora & Fein 1999).
The probabilistic correlation of direct meaning activation with conventionalization is in accordance with Tabossi and Zardon’s (1993: 156) interpretation of the notion of ‘idiom key’:
Perhaps the best way of specifying an idiom key is in probabilistic terms, as
the point in the string after which the probability of the fragment to continue
idiomatically is very high, even though a different, literal completion still may
be possible.
In other words, the activation of the idiomatic key and the stipulation of the
idiomatic meaning is not necessarily a semantic process (Titone & Connine
1999: 1667), but works as a probabilistic effect of the typical co-occurrence of
words that are embedded in idiomatic word configurations. This form-based
meaning activation is the decisive processing characteristic to make idioms
appear like words.
The direct stipulation model based on the probabilistic co-occurrence of
the idiomatic key words has another advantage. Namely, it can explain the processing disadvantages for idiom variants in comparison to the use of idioms
according to their base-forms. If an idiomatic base-form is stored in the mental lexicon as a probabilistic co-occurrence pattern, then idiom modification
can be said to disturb the prototypical connections in the configuration. As a
consequence, the configuration cannot be recognised immediately, which slows
down processing speed.
In Titone and Connine’s hybrid model, both compositional and noncompositional processing are combined in the discursive modulation of an
idiom’s meaning. This includes the following steps:
1. In the discourse representation the idiomatic word configuration is processed literally.
2. With the activation of the idiomatic key elements the idiomatic meaning is
directly retrieved.
3. Literal processing continues but now interacts with the idiomatic meaning.
4. a. For decomposable idioms the constituental meanings support the integration of the configuration into the sentence meaning, because they
can be directly mapped onto the idiomatic meaning (phrase-induced
polysemy).
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.29 (1504-1560)
b. For non-decomposable idioms the literal meaning slows down the integration, because the configuration remains contextually-ambiguous for a
longer time.
To determine the degree to which idiomatic and literal meanings are computed
during idiom processing and to find out in how far these meanings are integrated into the overall sentence meaning, Titone and Connine conducted an
eye tracking experiment. The results of this experiment support their model.
Concluding their investigation, Titone and Connine (1999: 1672) therefore
point to the extended range of applicability of their hybrid model:
[...] a model that incorporates the word-like aspects as well as the metaphorical and compositionally derived aspects of idioms is likely to be more generalizable across many classes of idiomatic and other nonliteral sequences. The
advantage of such a model is that it accounts for the similarities and differences in representation and processing of many classes of idiomatic sequences
as well as other nonliteral sequences.
By integrating the configuration with the decomposition hypothesis, Titone
and Connine’s hybrid model of idiom representation and processing achieves
to reconcile these two alternative views on idiom processing. By characterising idioms in probabilistic terms as conventionalised patterns of word cooccurrence whose direct meaning retrieval is dependent on the key-based
recognition of the mentally entrenched standard configuration, the model offers an appropriate psycholinguistic framework to capture the complex cognitive status of idioms. With the increasing conventionalisation of a pattern of
semantic extension, the activation weights of its semantic substructures shift
from the literal level to the level of idiomatic meaning. However, as most
psycholinguistic studies show, these transfers in semantic salience, are not accompanied with a complete suppression of the literal meaning (see also Gernsbacher & Robertson 1999). Rather, the literal meaning is processed to work as
a semantic background against which the idiomatic meaning is established.
By integrating the semantic dimension of compositionality (including
both literal-meaning composition and figurative-to-literal decomposition),
Titone and Connine further take into account that idioms are not reducible
to entrenched form-meaning associations, but that many of these configurations must be attributed a complex internal semantic structure. Still, their
psycholinguistic account does not answer the question of how decomposition
comes into existence. The notion of decomposition has the status of a heuristics
rather than a well-motivated theoretical concept (for corresponding criticism
see Dobrovol’skij 1997: 90).

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.30 (1560-1615)
Thus although psycholinguistic studies on decomposition have revealed
important regularities in the comprehension of different idiom-types, the
psycholinguistic results have not been related to a comprehensive and wellbalanced linguistic model of idiomatic constructions. Therefore, most of the
insights have remained in an empirical vacuum so far.
. Proverbiality – the functional motivation of idioms
.. Proverbiality
Nunberg et al. do not merely propose their refined semantic characterisation
of idioms as a mere heuristic device, but try to explain differences in the semantic structure of idioms in terms of a more general view of what idioms
are. One major point of criticism they direct against the generative treatment
of idioms is that the generative aspiration to find an optimal, context-free
grammar-system has lead to an over-grammatisation of idiomaticity. In contrast, Nunberg et al. do not leave the phenomenon of idiomaticity with its
semantic peculiarities in descriptive isolation but try to explain it in terms
of the figurative processes that underlie idiomatic constructions and the discursive functions that they serve. In their view, the discursive functionality of
idioms resides in what they call proverbiality (Nunberg et al. 1994: 493):
Typically, a proverb or proverbial expression invokes a concrete situation
(pulling strings, showing a flag, breaking ice) as the metaphorical model for a
recurrent culturally significant situation involving abstract relations or entities
(e.g. exerting influence, making one’s opinions known, easing the formality of
a social encounter).
Nunberg et al. base this insight on the significant statistical tendency of idioms to map concrete elements in the literal meaning onto abstract referents
in the idiomatic meaning. In other words, by pointing to the proverbiality
of idiomatic constructions, they recognise a more general discursive and cognitive function underlying the phenomenon of idiomaticity: a great number
of idioms work as metaphorical models that make it possible to characterise
abstract conceptual configurations on the basis of concrete, i.e. directly comprehensible scenarios. Unfortunately, the authors are unable to propose a theoretical model that allows them to explain the internal semantic organisation
of idiomatic expressions on the basis of distinct patterns of figurative extension
(the same point of criticism is also raised by Schenk 1995: 255). Nunberg et al.
(1994: 509) show that they are aware of this problem:
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.31 (1615-1663)
The distinction between idiomatic phrases and idiomatic combinations in and
of itself can provide only a partial account of the puzzling variable distribution of idiomatic interpretation. Interacting factors, mostly having to do with
the nature of the discourse function of particular constructions and the particular figures underlying various idiomatic combinations, have a critical role
to play as well. Providing such an account would require explicit formulations of the semantics and pragmatics of numerous constructions, together
with detailed analyses of a representative sample of idioms, showing how their
idiomatic and literal interpretations are related.[...] Such an undertaking is beyond the scope of this paper, but the approach outlined here would seem to be
an essential first step toward solving these long-standing mysteries.
Thus, the authors can take the credit for pointing to a functionally motivated
account of idiomaticity, but the precise correspondences between the proverbiality of idioms and their internal semantic organisation remains open in their
framework. Precisely this, however, is the focus of the present book.
.. Idiom motivation and conceptual metaphors
In order to unfold the functional proverbiality of idiomatic expressions, speakers must be able to motivate them. The idea that the users of a given language
can (re)motivate idioms semantically to work as metaphors has a long tradition in phraseological theory.12 Motivation refers to a speaker’s ability to make
sense of an idiomatic expression by reactivating or remotivating their figurativity, i.e. to understand why the idiom has the idiomatic meaning it has with
a view to its literal meaning. For instance, the idiomatic meaning of tip of the
iceberg (part of a serious problem or situation) can be motivated on the basis
of our knowledge about icebergs and the problems they cause for navigation.
In contrast, the idiom red herring (piece of information used for the deliberate
distraction of an opponent) is unmotivated for most speakers of English; they
have lost the metaphorical ground to make sense of this idiom.13
A number of psycholinguistic studies indicate that idiom motivation is not
only a classificatory construct, but that it plays an important role in the mental
representation of these constructions. Unfortunately, these studies generally do
not consider the extensive linguistic research on idiom motivation developed
in the field of phraseology (cf. also Burger 1998b: 30).
Discussing the “the place of idioms in a literal and metaphorical world,”
Cacciari (1993), for instance, tested the interpretation strategies that adults
and children use to assign a motivated meaning to idiomatic expressions. She
concludes that “the ability to make sense of idioms is not restricted to the

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.32 (1663-1714)
simple mechanism of retrieving an idiom’s meaning from lexical memory”
(Cacciari 1993: 41). Unfortunately, Cacciari does not interpret these strategies
any further but simply regards them as being caused by “general mechanisms
for meaning extractions” (Cacciari 1993: 43). Nevertheless, the results provide
strong support that a great number idioms are not stored as lexical units, but
that their motivated meaning is activated “together with information much
more complex than a simple paraphrase” (Cacciari 1993: 42).
A cognitive-linguistic model of idiom motivation is offered by Lakoff
(1987). Using the example of spill the beans, Lakoff (1987: 446–453) discusses
the motivation of idioms on the basis of conceptual metaphors (for a definition
see Section 3.3.1). Moreover, he tries to explain this process in relation to the
conventional images and the associated knowledge evoked by this expression:
The container for the beans is almost always the size of the human head; it
is not barrel size, or silo size, or size, say of a small mustard jar. The beans
were supposed to be kept in that container. The spilling is, or appears to be,
accidental. The beans are never spilled into a neat pile; instead they go all over
the place. They are never easy to retrieve. The spill is always messy.
(Lakoff 1987: 449)
The actual linguistic expression spill the beans does not describe or specify any
of these details, because the idiom’s formal structure does not explicitly refer to
a specific container or the situation after the spilling. Consequently, these details are conveyed through the image and the conceptual knowledge associated
with the literal-scene and not through the meanings of the individual words:
the conceptual knowledge transcends the amount of information yielded by
the individual meanings of the constituents.
Following Lakoff, people can activate rich conceptual knowledge by means
of which the literal meaning of the idiom can be mapped onto its idiomatic
meaning. This mapping process is supported by a set of conceptual metaphors.
In this case, the relevant metaphor is the conduit meaphor [...]. According to the conduit metaphor, the mind is a container, ideas are entities,
and communication involves taking ideas out of the mind, putting them into
words, and sending them to other people.
(Lakoff 1987: 450)
The meaning of the idiom is motivated because the conceptual knowledge
associated with the spill-the-beans scene and the metaphorical knowledge provide a link that makes it possible to relate the literal meaning to the idiomatic
meaning: “The beans correspond to information. The container corresponds
to the head” (Lakoff 1987: 450). The conduit-metaphor maps the conceptual
scenario of spilling beans onto the target-domain of mental activity and com-
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.33 (1714-1766)
munication (cf. Reddy 1993). As a result, the actual meaning ‘to reveal a secret’
can be motivated and the expression becomes analysable (spill = reveal, beans
= secrets).
In a previous study, (Langlotz unpublished 1998), I have employed this
heuristics in a comparative analysis of the semantic structure of English and
German idioms denoting mental fitness and weakness. The study revealed that
these idiomatic expressions are coherently structured by a well-defined set of
conceptual metaphors although they reflect a considerable degree of variation
on the level of specific lexical implementation. For instance, the following four
idioms to have a tile/slate loose vs. einen Dachschaden haben and to have a chink
in the brain vs. einen Sprung in der Schüssel haben are structured by a shared set
of conceptual metaphors: the mind as a container, ideas as objects, the
mind as (the upper part of) a building (for the first pair), the mind as a
bowl (for the second pair), insanity is a physical defect (resulting from
a physical force that destroys the container). In other words, the interlinguistic lexical variation between these English and German idioms is guided
by abstract cognitive principles and conceptual patterns that work underneath
them. For the comparative study, the cognitive-semantic notion of conceptual
metaphor therefore proved to be a good tool to discover underlying conceptual
regularities in the semantic structure of the respective idioms.
Lakoff ’s model of idiom motivation on the basis of conceptual metaphors
was also substantiated by various psycholinguistic experiments conducted by
Gibbs and his followers. The importance of conceptual metaphors for the comprehension of idioms is supported by the experiments performed by Nayak
and Gibbs (1990) and Gibbs (1990, 1992, 1993, 1994, 1995). Nayak and Gibbs,
for example, show that the potential motivation of idioms by conceptual
metaphors influences judgements about their appropriate use and interpretation in particular discourse situations and that people are sensitive to the supposed coherence of the conceptual metaphors underlying particular idioms.
Gibbs therefore concludes:
The traditional dead metaphor view provides no explanation for the contextsensitivity of idioms other than to suggest that the use of idioms is an arbitrary
matter of convention. The conceptual view of idiomaticity suggested here offers a motivated reason for why idioms mean what they do and are used in
specific kinds of discourse situations.
(Gibbs 1993: 71)
The vividness of metaphorical structures in idiomatic strings and their patterning by conceptual metaphors is further supported by Gibbs and O’Brien’s
(1990) study on the mental imagery evoked for particular idioms. It is shown

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.34 (1766-1817)
that informants can produce consistent images for those idioms that can be
structured by conceptual metaphors (e.g., spill the beans).
.. Criticism against the conceptual metaphor view of idiom motivation
The Lakoffian model of idiom motivation was criticised by a number of proponents from the Gluckbergian school of metaphor and idiom comprehension
(Glucksberg et al. 1993; Cacciari & Glucksberg 1995; Glucksberg & Keysar
1993; Glucksberg & McGlone 1999; Keysar & Bly 1999).14 Keysar and Bly
(1999), for instance, try to show that idioms can be remotivated but that
conceptual metaphors do not play a role in this process; rather,
intuitions about idiom transparency reflect the mind’s interpretative strategies. [...] when people learn an idiom’s meaning, they attempt to map elements
of that meaning onto the linguistic constituents of the idiom. This mapping
makes the stipulated meaning seem transparent, while it obscures the possibility that the idiom conceivably could mean something else.
(Keysar & Bly 1999: 1559)
To test their claim, Keysar and Bly presented two groups of informants with a
set of obsolete, and therefore unfamiliar, idioms such as the goose hangs high.
The researchers made each group learn a different postulated meaning for this
idiom. While the first group was biased towards a positive interpretation of the
expression (things are looking good), the other group was led to internalise the
opposite reading (things are not looking good). In the second test-phase, the
informants had to predict how an uninformed person would understand the
expression. These predictions were understood to reflect the internal motivations that the subjects mapped onto the idiom’s literal meaning. The results
of Keysar and Bly’s experiment showed that the informants’ predictions were
consistent with what they believed to be the ‘true’ meaning of the expression.
This outcome indicates that the actual meaning of an idiom constrains the way
the expression is motivated. Therefore, the authors postulate that conceptual
metaphors cannot be claimed to underlie the interpretation of idioms, but that
motivation is the effect of general interpretation strategies.
However, Keysar and Bly’s argument contains a number of basic problems.
First, their argument would only be justified if conceptual metaphors did only
apply to single idiomatic expressions. However, as all Lakoffian studies have
shown, and as will be shown in this study (Chapter 5), a great number of idioms
belonging to the same word field can be systematically motivated on the basis of
a restricted and coherent set of conceptual metaphors. The very systematicity of
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.35 (1817-1871)
metaphorical models and their relation to idiomatic expressions indicates that
the models underlie the motivation of these strings and effect their cognitive
functionality, rather than being merely interpreted into an idiom’s semantic
structure.
Further, Keysar and Bly’s experiment seems to be based on the implicit assumption that conceptual metaphors make idiomatic meanings predictable to
some extent. Thus, the researchers expected that the goose is high should have
been motivated according to the model good is up (Keysar & Bly 1999: 1572).
Counter to this implicit assumption, one must emphasise that Lakoff ’s description of idiom motivation does not assume predictability at all (Lakoff
1987: 450). Rather, the metaphorical models offer a conceptual basis on which
a given expression can be seen to ‘make sense.’ In other words, Lakoff ’s view
of idiom motivation is based on a top-down reinterpretative process which
takes the idiomatic meaning for granted. Thus if different speakers learn opposite meanings for an unfamiliar idiom, it cannot be expected that they activate
the same conceptual metaphors to render the expression transparent. In the
specific case of the opposite meanings for the goose hangs high, alternative
metaphor models exist to make sense of both interpretations. The positive
interpretation ‘things are looking good’ is rendered transparent on the basis of the conceptualisation good is up. The negative reading ‘things are not
looking good’ can also be motivated by another coherent, but fundamentally
different, system of conceptual metaphors: absence is distance, a valuable
resource is good food, security is being on the ground, uncertainty
is being high in the air (off the ground) or unknown is up. Therefore,
the informants’ ability to render both meanings for the idiomatic expression
transparent reflects their freedom to choose between different conceptualisations to support their interpretation. Moreover, working as guidelines in this
process, the conceptual models prevent nonsensical sense attributions.
Third, and most importantly, Keysar and Bly do not offer a plausible alternative to explain the process of rendering the two opposite readings transparent. What they very generally describe as “the mind’s interpretative strategies”
is not explained in terms of a coherent theory of figurative competence. Why
are both meanings for the goose hangs high possible? What is the implicit knowledge to render both meanings plausible? In my view, the facility with which
both meanings are adopted by the informants can only be explained in terms of
a cognitive theory of metaphor. More precisely, to be able to interpret a spatial
scene (the goose hangs high) as an evaluative statement, the interpreter must resort to his or her encyclopedic knowledge. Referring to the memory of specific
positive and negative experiences related to high-hanging objects, it becomes

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.36 (1871-1940)
possible to project the evaluative dimension onto the scene. The assumption
of conceptual metaphors offers a theoretical basis to explain why idiomatic expressions are used in the first place. Working as a pre-formulated conceptual
concretisation – the goose hangs high conveys a concrete scenario with reference
to which a given target-activity can be evaluated – idioms function as micromodels that provide easier cognitive access to complex and abstract conceptual
relationships as, in the present case, an evaluation.
In another study, Cacciari and Glucksberg (1995) review and criticise
Gibbs’s experiments on mental imagery and its structuring through conceptual metaphors. However, their criticism is based on a somewhat tendentious
interpretation of the former study. More precisely, they interpret the influence
of conceptual metaphor on mental images as making these images figurativemeaning-based rather than literal-meaning-based:
Such automatic access to the metaphorical source of an idiom’s meaning
would enable people to bypass the concrete literal meaning of idiom strings
when generating images for those strings. But can people bypass literal meanings?
(Cacciari & Glucksberg 1995: 46)
It is important to emphasise that Gibbs and O’Brien do not argue for a suppression of the literal meaning. Rather, they claim that the informants’ literal
images are patterned by the metaphorical models in terms of abstract conceptual relationships like causality, intentionality, reversibility or spatial relations
(e.g., the size of the bean-container). Thus, the literal images are influenced
by the abstract relations contained in the figurative meaning, but they are not
deactivated by them. Cacciari and Glucksberg’s results show that the images
evoked by informants are predominantly literal for all types of idioms. These
results, which correspond to the insights of an earlier study by Häcki Buhofer
(1989), lead them to conclude that Gibbs and O’Brien’s assumption cannot be
maintained:
The results of this study thus call into question Gibbs and O’Brien’s assumption that the mental images that people generate from idiom strings directly
reflect idiomatic meanings, or the conceptual metaphors that presumably underlie those meanings. Instead, the images tend to reflect the concrete actions
and events denoted by an idiom’s constituent words and phrases, [...]. As is
strongly suggested by the literature on imagery, concrete concepts are far easier to image than are abstract concepts, and so one would expect, on a priori
grounds, that the concrete, literal meaning of an idiom would be far easier to
generate than an abstract, figurative meaning.
(Cacciari & Glucksberg 1995: 53)
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.37 (1940-1962)
In my view, Cacciari and Glucksberg’s cognitive interpretation of their results is
fully right. The evocation of imagery is strongly dependent on the concreteness
of a concept. One can therefore conclude that an idiom’s potential to evoke imagery correlates with the concreteness of its constituents. Cacciari and Glucksberg’s results further underline the practical impossibility to suppress the literal
meanings of the idiom constituents in processing. However, their simultaneous
rejection of the influence of conceptual metaphor on the motivation of an idiom cannot be justified because Lakoff does not claim that conventional images
explain or stand for the figurative meaning of an idiom. Rather, the images
are seen to support the activation of conceptual source-domain knowledge,
which is then mapped metaphorically onto the target-domain of the idiomatic
meaning. To fulfil its most important cognitive function: the structuring of an
abstract domain in terms of a concrete one, conceptual metaphor, by its very
nature, depends on the conceptual richness of the source domain. If those idioms that can be interpreted as instantiations of conceptual metaphors did not
provide access to the concrete source-domains, the metaphors would not be
able to fulfil their concretising function. In other words, the very functionality, i.e. the proverbiality, of these idioms is dependent on the evocation of
source-domain-related cognitive representations, including imagery. The results reported by Gibbs and O’Brien do not deny the accessibility of such
representations. Rather, they point to the influence of the more abstract structuring function of the conceptual metaphor on the informants’ interpretations
of these images. In my view, the two studies therefore cannot generally be
regarded as contradictory (also cf. Dobrovol’skij 1997: 45).
Accordingly, I am of the opinion that it makes most sense to understand
the process of idiom motivation according to this compromise. People are able
to ascribe transparency to idioms by understanding the idiomatic meaning relative to the literal meaning in terms of a re-interpretative process. This process
is guided by underlying conceptual metaphors that are implicitly presupposed
in the interpretation. Often, the conceptual metaphors alone are not sufficient to determine the full motivation of a transparent idiomatic construction
(cf., e.g., spill the beans). Full conceptual motivation depends on the concrete
knowledge associated with the literal scene. Since conceptual metaphors are
relatively abstract figurative patterns, they do not fully predetermine what conceptual substructures from the literal source-domain are mapped onto the
idiomatic-meaning domain. These fine-grained mappings can only be established on the basis of the specific encyclopedic knowledge associated with the
literal meaning. Thus, while the motivation of spill the beans is guided by the
conceptual conduit-metaphor, it is not only this superordinate mapping that

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.38 (1962-2016)
motivates the actual idiomatic meaning. Rather, the specific motivation is enriched by one’s specific knowledge about the entailments of the situation of
beans being spilled.
From the perspective of the European phraseological tradition, Burger
(1998a: Chapter 4) and (1998b) also criticises the treatment of idioms by the
Lakoffian school of conceptual metaphor. Although Burger concedes that this
cognitive-linguistic approach provides a valuable perspective for the semantic description of idioms, he criticises that in many cases the notion of idiom
is harnessed for the purposes of pro-conceptual-metaphor arguments in an
overly generalising way. More specifically, Burger raises the following points of
criticism against the Lakoffian approach:
1. Not all idioms are metaphorical (see Burger 1998b: 30).
2. Not all idioms that are metaphorical are structured by general conceptual
metaphors. A great number of idioms cannot be described according to
systems of conceptual metaphor (see Burger 1998a: 91).
3. The actual occurrence of idioms in texts does not suggest that the underlying metaphorical models are used in a consistent way (Burger 1998b: 36).
4. The definition of metaphorical models on the basis of idioms often violates
the linguistic data. The metaphorical patterns in idioms are often more
specific than the generic models suggest (Burger 1998a: 91).
5. The supposed universality of conceptual metaphors neglects the historical and cultural specificity of the phraseological system of a given language (see Burger 1998a: 91; see also Dobrovol’skij & Piirainen 1997: 115,
Footnote 9).
As concerns (1) and (2), the notions of idiom and conceptual metaphor are
obviously not congruent. As will be shown, conceptual metaphors are conceptual patterns of semantic extension and therefore exist independently of
their symbolic realisation in idioms. Moreover, many idioms (e.g., spick and
span, red herring, etc.) cannot be related to conceptual metaphors. (To be fair,
one has to emphasise that this point is also raised by Lakoff (1987: 451)). Although a considerable number of idiomatic constructions can be motivated on
the basis of conceptual metaphors, the two notions point to two qualitatively
different phenomena: conceptual structures on the one hand and linguistic
constructions on the other. There seems to be systematic overlap between the
two phenomena, but they are not congruent. The degree to which conceptual metaphor plays a role for the motivation of idiomatic constructions has
to be scrutinised. To evaluate items (3)–(5), further empirical investigation in
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.39 (2016-2070)
the respective areas (discursive behaviour of idioms, cross-cultural contrastive
phraseology, metaphor theory) is therefore needed.
The precise relationship between proposed conceptual structures and idiomatic data constitutes the central focus of the present cognitive-linguistic
analysis. This leads me to my hypothesis and major points and contentions.
. Outlook – desiderata for a cognitive-linguistic model of idiom
representation and variation
Linguistic and psycholinguistic evidence speaks for the representation of idioms as complex, composite word-configurations rather than lexical units.
Mentally, these word-configurations nevertheless have unit status because their
idiomatic meaning can be unfolded through direct stipulation once the most
salient constituent – the key – is heard. Psycholinguistic evidence further shows
that many of these sequences can be decomposed, which influences their processing and syntactic flexibility. The semantic complexity of idioms is further
enhanced by the fact that the composition of the literal meaning cannot be
suppressed in actual processing. Finally, considerable evidence speaks for the
motivated semantic structure of a number of idioms. While the influence
of conceptual metaphor on idiom representation has remained controversial,
it cannot be disclaimed as an important basis for the semantic structuring
of idioms.
In short, all this evidence supports a hybrid view of idioms as complex constructions. Such a hybrid view can best account for the fact that the form as well
as the internal semantic structure of idioms can be very diverse. At one pole,
one can find semantically unanalysable units such as spick and span whereas
at the other pole there are highly transparent expressions such as the proverb
people who live in glass houses should not throw stones. For metaphorical idioms
or proverbs of the latter type we have a clear idea of the constituents’ contribution to the overall meaning. On the semantic continuum between word-like
and metaphor-like idioms there are a number of motivated and/or analysable
strings such as spill the beans which tend to be processed like compositional
constructions. Given this diversity of semantic motivation and analysability,
one must also assume corresponding diversity in idiom comprehension and
production. This diversity must be accounted for by a model of idiom representation, which does not reduce these linguistic units to long words, but regards
them as complex mental representations that have the potential of unfolding
various levels of structural and semantic representation.

JB[v.20020404] Prn:20/02/2006; 14:10

F: HCP1702.tex / p.40 (2070-2120)
The previous review and discussion thus shows that in order to become
able to explain, rather than describe, the grammatical status of idioms and their
variability, we need a linguistic theory that can adequately handle the structural
and semantic heterogeneity of idiomatic constructions and that is powerful
enough to find reasons for their variation-potential and variation-constraints,
respectively. Moreover, it must find explanations for the underlying analysability, motivation or opacity of idiomatic constructions rather than only describe
or postulate them.
So far, the notion of semantic compositionality, as proposed by Nunberg
et al., has only served us as a heuristic concept because the phenomenon is
not explained on the basis of a more general lexical-semantic theory. (This
point of criticism is also highlighted by Schenk 1995: 255 and Abeillé 1995: 17).
As we have seen, Nunberg et al. are aware of this problem, but cannot solve
it within the context of their theoretical framework. A powerful explanatory
model of idiom representation and variation must be therefore be able to
explain and model
1. different types of semantic specialisation and structural idiosyncrasy in
idioms on the basis of a well-motivated theory of semantic extension,
2. the ways in which these specific semantic and structural characteristics are
instantiated in actual use,
3. the systematic lexicogrammatical variability of idioms as an effect of their
semantic and structural characteristics,
4. the potential existence of other factors, rather than semantic ones, that
underlie the actual use of idioms,
5. the criteria underlying the distinction between systematic variability and
wordplay
As we have seen, Lakoff ’s model is a first programmatic attempt to explain
the semantic motivation of idiomatic constructions in a cognitive-linguistic
framework. The model implies that the semantic structure of many idioms can
be coherently shaped by conceptual metaphors. Idioms therefore inherently
transport this principle of cognitive creativity. Is it then also possible to explain
the grammatical variability of idioms in actual use on the basis of the same
framework? The expression grasp the nettle does indeed seems to be influenced
by conceptual structuring in terms of the following metaphors: problems are
dangerous natural forces (→ nettle = problem) and dealing with a
problem is manipulating/tackling a physical object (→ grasp = deal
with problem). These conceptual correspondences motivate the idiom’s semantic structure and render it analysable. This influences the grammatical
JB[v.20020404] Prn:20/02/2006; 14:10
F: HCP1702.tex / p.41 (2120-2140)
malleability of the construction. For instance, example (6) on page 8 can be
paraphrased as:
(27) There is a nettle (a problem) for a somebody – it might as well be the new
Environment Secretary – to grasp (to deal with).
This analysis suggests that an application of this cognitive-linguistic heuristics to model the phenomenon of idiom representation and variation is wellmotivated.
The present study is an attempt to solve the central controversy about
idiom representation and variation by the help of a cognitive-linguistic approach. Investigating deeper into the structural characteristics of idioms and
their mental correlates, my cognitive-linguistic approach to idiom representation provides a differentiated and cognitively-motivated view of what an
idiom is. This, in turn, should provide a well-motivated basis on which idiomvariation can be explained. In what follows, the cognitive-linguistic heuristics,
i.e. the tools for my model and analyses, are established.

JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.1 (46-112)
chapter 
The cognitive architecture
of meaning and language
We humans are all model builders, building models of the
world and of ourselves within that world.
(Lamb 2000)
In this chapter the theoretical foundations will be established on the basis of
which idioms can be modelled in cognitive-linguistic terms. First, a set of basic mental processes that underlie the cognitive organisation of knowledge are
characterised. Showing how these processes shape the cognitive construction
of mental representations, the cognitive architecture of knowledge is sketched
in a second step. This architecture will then be extended by different processes of semantic extension, most prominently metaphor and metonymy. As
we have seen, these processes are assumed to be of particular importance for
the mental representation of idioms. Finally, the place of linguistic representation and processing in this framework will be outlined in line with Langacker’s
Cognitive Grammar.
. Basic cognitive processes
Human cognition depends on a number of fundamental cognitive processes
that make it possible to acquire, structure, store and process information and
knowledge. Rather than working as a static box that is filled with thoughts
and concepts, the mind functions as a dynamical network which is constantly
re-shaped by mental processing: “Mind is the same as mental processing” (Langacker 1987: 100). This dynamic conception of cognition is motivated by recent
interactionist, connectionist and emergentist models that describe mental activity and growth in terms of the brain’s neurological organisation: information
is processed and stored by the spreading activation or inhibition of neuronal
groups, which are organised in terms of neuronal networks (see, e.g., Bates
et al. 1998; Bechtel & Abrahamsen 1991; Elman 1998: 501–502). Such net-
JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.2 (112-163)
works contain a huge number of nodes that are interlinked via connections
with different connection weights. Meaningful representations emerge through
strong associations between these nodes. In other words, knowledge resides in
the specific connection patterns in the neuronal network. Cognitive-linguists
have adopted this network view as the most important heuristics to explain
cognitive representations and processes, including language.
On the basis of his dynamic conception of the mind, Langacker characterises mental experience as a very complex sequence of a myriad of both
transitory and recurrent cognitive events. An event describes “a cognitive occurrence of any degree of complexity, be it the firing of a single neuron or
a massive happening of intricate structure and large-scale architecture” (Langacker 1987: 100). To structure mental experience, these storms of cognitive
events must be ordered. A primary factor for this is the strengthening of connections in neuronal networks. By the frequent and strong activation of a given
network, the connections between the nodes are enforced to become discrete
activation-sets (Bates et al. 1998: 591). Langacker (2000a: 3) calls this process
entrenchment. Entrenchment underlies the cognitive process of routinisation.
Cognitive routines are events that have reached a degree of entrenchment and
automatisation that gives them the status of units. Mental units thus emerge as
“pre-packaged assemblies” or “habits” that do not demand the “constructive
effort required for the creation of novel structures” but can be directly activated
due to automatisation (Langacker 1987: 57). Units must be distinguished from
transitory events, i.e. occasional cognitive events that are not entrenched. Since
automatisation is dependent on frequency and stimulus strength, entrenchment must be seen as a gradual and dynamic process. As a result, the unit-status
of a given cognitive event is subject to constant reorganisation: units can be reinforced and become more deeply entrenched, whereas scarcity of activation
leads to the deautomatisation of a routine (Langacker 2000a: 10).
For the crystallisation of structured mental experience, cognitive events do
not only have to become entrenched but must also be coordinated. Coordination between events occurs when a number of nodes or clusters of nodes are
connected to create a complex event (Langacker 1987: 177). Complex events
thus result in complex activation-sets that are established between coordinated
nodes. By definition, then, coordinated subevents are immanent in the complex
event: relative to the overall structure of the complex event immanent events
are included as substructures.
The process of comparison allows human beings to weigh up one complex
event against another event and notice similarities and discrepancies between
their immanent substructures (Langacker 1987: 101). The recognition of dis-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.3 (163-223)
Chapter 3. The cognitive architecture of meaning and language
crepancy between cognitive events is very central for the imposition of discrete
boundaries onto mental experience. I will call this process segmentation (Lamb
2000: 177). The grouping of cognitive events into segmented, discrete entities
and the recognition of their interconnectedness or autonomy is the result of
highly complex comparison procedures that scan the mental representations
that are triggered by perceived stimuli.
While the coordinated construction, entrenchment and comparison of
complex mental events underlies the human ability to create elaborate and
distinct mental representations, our mental capacity for abstraction makes it
possible for us to conceive such representations at different levels of detail:
(1) a. This black silk Armani shirt costs 2000$. →
b. This shirt is expensive. →
c. The thing is expensive.
When reading (1a–c), one notices that the same fact can be formulated with
different degrees of precision (Langacker 1998: 5). This example reflects our
cognitive capacity to derive schemas from concrete cognitive events to conceive a given object or situation at different levels of schematicity or specificity
(Langacker 1987: 132). A schema has the quality of an abstract, higher-order
mental network that is derived from more concrete and dense information. Information that can be subsumed under the schema is called an elaboration or
instantiation of the schema (Langacker 1987: 68). For example, in (1), sentence
(1c) can be seen as a schema for (1a) and (1b). These latter sentences thus work
as elaborations or instantiations of this information. In the case of elaboration,
the schema is thus immanent to its instantiations (Langacker 1987: 438).
Schemas are of primary importance for the organisation and interpretation
of experience. They can be activated and projected onto lower-level mental
structures to integrate and structure them:
Figure 3.1 Deviant triangle
Objectively, the three sets of lines do no cohere. Since, however, our cognitive
apparatus is conditioned to organise perceptual stimuli into meaningful struc-

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.4 (223-279)
tures, we connect these lines unconsciously to construct a meaningful entity
– a triangle. How can this be explained? With reference to our set of mental
processes, this cognitive construction-process amounts to the projection of the
entrenched triangle-schema onto perceptual stimuli for the sake of grouping
them. More explicitly, the bottom-up perception of the visual stimuli results in
segmentation, i.e. the recognition of the three discrete groups of lines or angles,
respectively. To organise this flow of perception-events further, an entrenched
mental unit can be activated to work as a standard with which this novel experience is compared. A standard provides a cognitive routine relative to which
a target – the novel experience – can be assessed (Langacker 1987: 101). The
triangle-schema constitutes such a standard. Since the overall configuration of
the perceived lines is immanent in the triangle-schema, this schema can be activated. By projecting the triangle schema top-down, the lines can be further
coordinated. This makes us perceive a triangle.
On the basis of this description, the processes of projection and grouping
can be defined: projection describes the top-down process of mapping an entrenched mental standard, S, onto a cognitive target, T, to group it. Note that
the standard can only be projected onto the target if it is activated on the basis
of some substructures that are immanent to both S and T. Accordingly, grouping involves the process of organising the substructures of T relative to S (also
cf. Indurkhya 1992: 96).
The projection of schemas to group cognitive targets underlies the fundamental process of categorisation. Through categorisation information is organised into groups of similar entities by abstracting away from their individual
differences (see Schwarz 1996: Section 3.3.2.1; Taylor 1995; Ungerer & Schmid
1996: Chapter 1). A category thus amounts to a schema that is designed to
group lower-level cognitive entities.
Langacker’s account of categorisation provides a synthesis between the disparate views proposed by prototype-theory and classical taxonomic models
(for an extensive discussion see Taylor 1995). Langacker’s model is based on
two distinct but related categorisation relationships:
a. categorisation by schema (categorisation by elaboration of a schema)
b. categorisation by prototype (categorisation by extension from the prototype)
While the notion of prototype implies that some members are more representative for a category than others (see Rosch 1978), taxonomic models proceed
from the idea that the members of a category share a distinct set of necessary and sufficient features that are incorporated in the overarching category-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.5 (279-316)
schema. To exemplify the correlation of the two modes, one can categorise the
concepts hammer and computer relative to the hyperonym tool. The Collins
Cobuild English Language Dictionary defines this concept as follows: “A tool
is any instrument or piece of equipment that you hold in your hand in order
to help you to do a particular kind of work.” Let us assume that this definition represents the tool-schema. Our concept of a hammer fits this definition
prototypically and thus elaborates the schema fully. In contrast, a computer is
not usually held in one’s hand. Relative to the category-prototype, hammer,
computer thus stands in a relationship of extension. The tool-schema is only
partially immanent in this categorisational target: although the tool-schema
overlaps with the properties of computers, the computer-concept does not
elaborate it fully. Thus, a computer can be categorised by the tool-schema,
but hammer is a better, prototypical exemplar of this category.
With this short overview of basic cognitive processes in hand, we are now
in a position to model the mental organisation of knowledge.
. The mental representation of knowledge and meaning
.. Conceptualisation
From a cognitive perspective, knowledge consists in meaningful conceptual
structures that are stored in the mind and can be projected onto experienced
phenomena to understand them (for an overview see Schwarz 1992). For the
mental system to work, it is indispensable that conceptual categories channel
the enormous amount of external information that stimulates the human perceptual apparatus. To group the incoming stimuli, human beings must establish meaningful categories to help them reduce this information in a functional
way. In short, the human mind is forced to simplify.
Concepts fulfil the central functions of representing, storing, structuring
and organising information about the world economically and thus allow us to
retrieve, process and communicate knowledge in an efficient way. We can, for
instance, imagine a cat in our mind’s eye without actually seeing, hearing or
touching a genuine cat. This indicates that we possess a cat-concept that integrates visual, auditory and tactile information about these animals (cf. Lamb
2000: 183). Working as a piece of meaningful knowledge, we can project this
conceptual cat-schema as a categorisation-standard to make sense of any concretely perceived target-creature that features a certain degree of ‘catness.’ Thus,
meaningful experience is not only related to our immediate perception of the

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.6 (316-369)
physical world, but draws upon conceptual routines. Concepts allow us to make
sense of our environments by converting the sum of diffuse external information into discrete mental entities. In other words, conceptual units works as
entrenched high-level categorisation-standards that can be mapped onto the
flow of experience to group and organise it.
The ability to evoke a concept, e.g. a mental cat-image, speaks for the
relative autonomy of conceptual representation and processing. Langacker
(1987: 112) therefore distinguishes between autonomous and peripherallyconnected cognitive events. While peripherally-connected events emerge through
the direct sensorimotor interaction with the physical world, autonomous
events can be processed without reference to such direct experience. Concepts
are cognitively constructed through concept formation. This process transfers
concrete peripherally-connected stimuli to the more autonomous mode of
conceptual representation (cf. Schwarz 1996: 94). For example, the cat-concept
must abstract from the concrete and rich information that we perceive when
we encounter different real cats. Concept formation thus amounts to a process
of deriving an autonomous conceptual schema that reduces the informational
content of peripherally-connected perceptual representations.
In short, the creation of knowledge thus proceeds through the following
stages. Physical stimuli from the external world serve as input to the cognitive
agent’s perceptual apparatus who transforms these stimuli into a sensorimotor data set, i.e. a “reality that is made available for conceptualization through
the sensorimotor apparatus of the cognitive agent” (Indurkhya 1992: 132). The
sensorimotor data set is the pre-selected and pre-grouped cognitive representation of what the cognitive agent experiences as his environment. To understand this environment, the sensorimotor data set must be further grouped
and organised by deriving conceptual structures that dissect it into functionally appropriate meaningful units. Therefore, ‘reality’ as we perceive is it the
constructive result of projecting conceptual schemas onto immediate sensorimotor experience. Concepts provide us with relatively autonomous models of
the world as we understand it. A cognitive analysis of knowledge and meaning
thus amounts to an analysis of the structure and growth of these “projected
worlds” (Lamb 2000: 17; cf. also Johnson-Laird 1983).
.. Sensorimotor experience and preconceptual structures
To explain the acquisition of conceptual structures, Lakoff and Johnson (1980)
and Lakoff (1987) propose a psychological epistemology termed experientialist
realism or experientialism (see also Lakoff & Johnson 1999). According to this
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.7 (369-411)
view, meaning is only constructed when cognitive agents interact with their
physical and social world of experience (cf. also Hays 2000: 162). Meaning is
thus conceived as embodied, i.e. it is determined by the concrete human interaction with their experiential realities. Embodied meaning is determined by
the specific human biology, the physical world, the culture(s) we live in and
our social organisation. Experience in this broad sense determines the shape of
our meaningful conceptual structures; these cognitive constructs are not only
individual but constitute collective and shared cultural systems of knowledge.
According to Lakoff (1987: 267), embodied meaning is based on a “directly
emergent” preconceptual form of bodily experience that exists previous to conceptual structures and provides the experiential ground for rich meaning to
grow. Basic forms of preconceptual structuring make it possible for us to perceive objects and their boundaries, to realise alternative relationships between
objects and to distinguish objects from processes (see also Lamb 2000: 175).
With reference to Langacker, we can employ the notions of thing, process and
interconnection as technical terms to get a handle on these dimensions (for a
comprehensive definition see Langacker 1987: Chapters 5, 6 and 7). For example, the conceptualisation of a triangle depends on its preconceptual grouping
into specific entities (sides, angles) that constitute its structure. To perceive a
triangle as a thing, these entities must be grouped as being interconnected,
i.e. as defining a salient region in visual space. Going further, Lakoff distinguishes two more specific types of preconceptual structuring that organise the
sensorimotor data set: “image-schematic structure” and “basic-level structure”
(Lakoff 1987: 267).
Image-schemas such as container, path, balance, up-down, frontback, part-whole are very simple force-dynamic, orientational and relational
structures that emerge from recurrent patterns, shapes and regularities in our
daily activities (Lakoff 1987: 267; see also Johnson 1987). For instance, babies can move from one point to another without having a concept of way
or distance. The structure experienced when they make this movement consists of a starting point, an endpoint and a link between the two points. This
path-schema is implied in a number of phenomena: walking from one point
to another, throwing a ball to somebody, hitting somebody, etc. (Johnson
1987: 28). Thus, image-schemas are abstract kinaesthetic patterns that transcend particular modes of perception. They underlie the cognitive structuring
of interconnected sensorimotor experiences and also pattern the perception of
force-dynamic relations between objects (Talmy 1988, 2000). In short, imageschemas structure the whole of our most fundamental direct experiences of the

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.8 (411-488)
world and therefore work as conceptual archetypes (see also Langacker 1993: 3;
Gibbs & Colston 1995).
On the basis of archetypal image-schematic structures, the evolution of
first concepts can be explained. Basic concepts like way, path or distance are
derived from the path-schema presented above. In this sense, orientational
concepts emerge as “directly comprehensible concepts” (Lakoff & Johnson
1980: 58). The relation of these concepts to sensorimotor bodily experience is
preserved in a fully developed conceptual system (Lakoff 1987: 267–268).
With image-schemas, basic-level concepts constitute the second basis on
which complex conceptual structures are established. The basic level works as
an effective entrance point into a network of conceptual categories (see Rosch
et al. 1976; also cf. Berlin et al. 1974):
superordinate level:
basic level:
subordinate level:
animal
dog
dachshund
plant
tree
oak
Within such categorisation hierarchies the basic level is psychologically more
prominent than the other levels. At the basic level the most relevant information of a conceptual category is stored at an intermediate level of granularity. It
thus works as the most economic level of abstraction in comprehensive mental
representations. For instance, when seeing a dachshund, it is more probable for people to refer to this creature by using the term dog than by calling
it dachshund. Only if it were necessary to make the communicative distinction between a dachshund and another dog, say a poodle, the term dachshund
would be used. This preference for basic-level categorisation is determined by
cognitive economy and functionality: the basic level is related to the perception
of gestalts, i.e. holistic entities, while the subordinate and superordinate levels
differentiate and schematise these gestalts, respectively: “At that level, things are
perceived holistically, as a single gestalt, while for the identification at a lower
level, specific details (called distinctive features) have to be picked out [...]”
(Lakoff 1987: 33).
The shift from preconceptual structures of bodily experience to conceptual
structures leads to the emergence of basic-level concepts. The basic structure
of bodily experience that is related to the concept sit (down/up), for instance,
consists in the acts of standing, sitting down, sitting and standing up plus the
bodily interaction with the various kinds of seat. Working as a gestalt, the
basic-level concept sit groups these basic-level experiences and summarises
the processes to constitute a unified whole.
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.9 (488-538)
.. Cognitive models
On the basis of image-schemas and basic-level concepts more complex conceptual representations can be constructed. Lakoff (1987: 68) calls such complex
mental networks of world knowledge idealised cognitive models or ICMs. (Alternatively, such arrangements of conceptual knowledge have been termed frames
(Fillmore 1975, 1985; Fillmore & Atkins 1992) or scripts (Schank & Abelson
1977)). For example, the ICM of week structures our experience of time in a
systematic way. This cognitive model is idealised because in nature weeks do
not really exist. Indeed, other cultures use other ICMs to organise the flow of
time (Lakoff 1987: 69). Consequently, each ICM expresses a particular worldview, i.e. a projected world which is superimposed on our experience in order
to comprehend it in a coherent way. Constituting autonomous cognitive constructs, ICMs can be evoked to group highly complex experiential phenomena
(e.g., day-night cycles) and thus create what we experience as reality: “Since we
act in accord with our conceptual systems and since our actions are real, our
conceptual systems have a major role in creating reality” (Lakoff 1987: 296).
To fulfil their cognitive function, cognitive models must be have a clearcut internal organisation: “Each ICM has an ontology and a structure (Lakoff
1987: 285).” This internal set-up of cognitive models is usually determined by
image-schemas and basic-level concepts. For example, the journey-ICM is
shaped by the following conceptual ontology: a) a point of departure, b) a
route, c) a destination, d) a traveller moving from a starting point to an end
point along a path by means of e) a vehicle. The ontology of a cognitive model
characterises the discrete things contained in it. Any element of the ontology constitutes an intrinsic conceptual category. The structure of the cognitive
model captures the coordinated interconnections between these conceptual elements. In the present example, the ontological concepts are arranged by the
path-image-schema as well as the dynamic movement of the traveller along the
path. Thus, cognitive models, as the journey-model, become directly meaningful because they can be directly related to our bodily and social experience
via basic-level concepts and image-schematic structures. Of course, more complex cognitive models must be created to understand intricate phenomena.
This works by further abstracting or differentiating the basic-level “by the projection from basic level categories to superordinate and subordinate categories”
(Lakoff 1987: 268). In this way more abstract or more detailed ICMs can arise
from image-schematic structures and basic-level concepts.
The structured nature of ICMs implies that individual concepts are not
stored in isolation but densely networked. Thus, concepts are embedded in

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.10 (538-575)
what Langacker (1987: Chapter 4) calls domains. The notion of domain points
to the fact that ICMs function as mentally represented experiential contexts
or backgrounds relative to which individual conceptual units are understood.
For example, the concept finger can only become meaningful relative to
the concept hand in which it is embedded. hand itself constitutes a subdomain of the ICMs arm and body. This illustrates that conceptual categories
are not only organised at various taxonomic levels of schematicity and specificity, but that they are further patterned in layers of conceptual embedding.15
Moreover, most concepts demand more than one experiential domain for their
full description. Langacker uses the notion of domain matrix to describe this
domain-multiplicity (Langacker 1987: 147, 1991b: 4). For instance, the concept
cat must be defined relative to a matrix of different domains. First, cats can be
related to their function as pets and are thus embedded in the domain household. Further, they figure among ferocious carnivores which demands their
conceptualisation in the domain of wildlife. Finally, when being black, they
belong to the complex experiential context of superstition.
In a nutshell, knowledge is organised in mentally constructed networks of
conceptual categories and cognitive models that represent world knowledge in
an autonomous cognitive format. Concepts and ICMs function as cognitivelyrepresented schemas relative to which new experience is understood.
. Complex patterns of semantic extension
Following Lakoff and Johnson’s (1980) pathbreaking insights, many cognitivelinguists have shown that the cognitive architecture of knowledge is not restricted to directly comprehensible ICMs but that conceptual structures are
figurative to a great extent. These figurative conceptual structures are based
on the semantic extension of concrete conceptual categories and conceptual
domains by means of the fundamental cognitive mechanisms of metaphor,
metonymy and blending (see Lakoff 1987; Johnson 1987; Lakoff & Johnson
1999; Gibbs 1994; Fauconnier & Turner 1998; Turner & Fauconnier 2000).
.. Conceptual metaphors
Conceptual metaphors are metaphorical concepts, i.e. complex ICMs that are
cognitively constructed by mapping a concrete source-domain onto a more
abstract target-domain (Lakoff & Johnson 1980: 4). For instance, expressions
such as our relationship has hit the buffers; We’re at a crossroads; look how far
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.11 (575-639)
journey:
source domain
development of love:
target domain
Ontology (conceptual nodes):
Ontology:
travellers
starting point
route
destination
vehicle
lovers
beginning of relationship
‘direction’ of development
purpose/goal (happiness)
relationship
------------→
------------→
------------→
------------→
------------→
Structure (interconnections):
Structure:
path-image-schema
temporal development
----------→
Figure 3.2 The development of a love relationship is a journey
we’ve come; It has been a long and winding road; We may have to go our separate ways are patterned by the conceptual metaphor the development of
a love relationship is a journey.16 This general and systematic conceptual
metaphor maps the journey-ICM (with its immanent ontology and structure)
onto the abstract target-domain development of love relationship. The
mapping establishes multiple, clearly structured correspondences between the
ontologies and structures of the two domains (see Figure 3.2, based on Lakoff
1993a: 206–208)
The systematic correspondences between the two domains make it possible to manipulate the source-domain in order to draw metaphorical inferences
about the target-domain. For example, we can describe a critical phase in our
partnership as: Darling, I think our relationship is spinning its wheels. We know
that a car that is spinning its wheels has got stuck and cannot move forward.
This situation can only be mended by some significant effort: either the travellers push the car out of the mud or it is taken in tow by some recovery vehicle.
Due to the metaphorical correspondences, these entailments can be transferred
to the target to establish analogical inferences. Thinking about car-journeys
thus allows the cognitive agent to draw conclusions about love relationships:
the love relationship can only be recovered by some significant effort or by
external help (Lakoff 1993a: 207). In other words, the conceptual metaphor
fulfils the function of providing easier cognitive access to the more abstract
target-domain. Instead of manipulating the empirically-abstract domain love
relationship, the cognitive agent can draw inferences about it by referring
to the more clearly structured and more directly comprehensible domain car
journey. By projecting the source-domain onto the target-domain, the ontol-

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.12 (639-690)
ogy and structure of the target can be grouped according to the experiential
entailments immanent in the source-domain network.
This example illuminates the importance of metaphor for cognitive functioning. Conceptual metaphors fulfil the function of structuring and ontologising many empirically-abstract domains (emotions, careers, love relationships, time, life, cognition, etc.). Since these domains cannot be tested
through direct sensorimotor manipulation, they are not open to the direct formation of fine-grained cognitive models, but can only be accessed indirectly.
To understand these domains, i.e. to make them accessible to cognitive manipulation, they must be given a concrete structure and ontology. This can
be achieved through the metaphorical projection of directly comprehensible
cognitive models. In this sense, conceptual metaphors reflect cognitive creativity. The comprehension of a conceptual target in terms of a well-entrenched
source-domain demands the conceptualiser’s cognitive ability to actualise connections between mental representations that have not been exploited previously. In other words, the process of establishing coordinated links between
the source and the target networks leads to a creative elaboration of conceptual network relationships. Metaphor thus works as a fundamental, creative
cognitive-process for enlarging and structuring our conceptual representations
(cf. also Indurkhya 1992).
Following Langacker one can pare metaphorical mapping down to an
elaborate process of categorisation by extension (see Langacker 1987: 143; Langacker 2000a: 17–18). Metaphor emerges as an unconventional categorisation
event in which a complex conceptual standard, e.g. the journey-ICM, is
evoked as a source or vehicle to refer to an actual topic or target (love relationship) which lies outside its conventional categorical value.17 As a consequence,
the novel target is categorised by extension. This extension of the scope of the
source-concept creates a semantic tension between its conventional meaning
and its actual target-meaning (cf. Langlotz 2004b). As a result, two levels of
interpretation are evoked: a literal level (the level of the conventional meaning
journey) and a metaphorical level (the level of the actual meaning love relationship). To motivate this metaphorical categorisation event, i.e. to establish
the link between the actual (metaphorical) and the conventional (literal) conceptualisation, the conceptualiser must recognise or establish some link or
common ground between the two conceptual representations, i.e. he or she
must activate or create relevant similarities shared by the two concepts.18 This
recognition or creation of similarity thus amounts to the extraction of a superordinate schema which is instantiated by both the standard and the target.
In the development of a love relationship is a journey-metaphor the
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.13 (690-746)
similarity between the two domain is created by means of the path-imageschema: the schema structures the temporal development of the relationship
as a spatial process.
The cognitive functionality of conceptual metaphor can be readily embedded in the multi-layered architecture of cognitive structures and explained by
means of this metaphor-as-categorisation-by-extension account. By projecting
the journey-model, the abstract concept of love relationship is grouped
and structured in terms of our concrete understanding of the source-domain.
In other words, the source functions as a projected world; it works as conceptual background of comparison that helps us understand the abstract targetexperience. In this sense, conceptual metaphors fulfil the same basic function as
non-metaphorical cognitive models (cf. Indurkhya 1992). By conceptualising
and concretising empirically-abstract experiences, they shape our world-views
and actions and therefore constitute what Lakoff and Johnson (1980: 55) call
“metaphors we live by.”
Being systematic and coherent concepts, conceptual metaphors are shaped
by the same cognitive processes as all other cognitive structures. Thus, the basic
cognitive processes of schematisation and elaboration also affect these complex patterns of semantic extension. For instance, the development of a love
relationship is a journey-metaphor is an instance of the more schematic
event structure-metaphor, which itself consists of the more basic metaphors
states are locations (I am in/out of love, in a deep depression), changes
are movements (they fell into love, our friendship turned into love), causes
are forces (She drives me crazy) (see Lakoff 1993a: 220–225). The development of a love relationship is a journey-metaphor constitutes one
instantiation of this metaphorical event-structure configuration. Thus, conceptual metaphors do not occur in isolation from one another but they are
organised in complex networks of taxonomic inheritance hierarchies (Lakoff
1993a: 222).
Moreover conceptual metaphors can be networked through coordination
to create highly complex composite metaphors (Lakoff & Turner 1989: 70).
Metaphorical composition is involved in the career is a journey-metaphor
for instance. Since, metaphorically speaking, careers are success-oriented journeys, career goals are not only conceived as target-destinations but are also
conceptualised relative to the abstract concept of success. Being abstract, this
concept is itself patterned by the conceptual-metaphor success is up (as reflected in be on the up, be in the upswing, up and coming). To conceptualise
the success-oriented nature of career-goals, this metaphor is coordinated with
the destination-slot in the journey-domain: career goals are high des-

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.14 (746-810)
tinations. The two metaphors career is a journey and success is up
are thus coordinated via the target-concept career-success. This composite
metaphorical relationship is implied in the idioms: come up/go up/move up in
the world, a mountain to climb, the dizzy heights, reach the top, keep one’s feet
on the ground, the summit of one’s career, etc. In short, individual conceptual
metaphors are networked with other metaphors to form complex and interlocking metaphor systems (see Lakoff & Johnson 1980: Ch. 16; Lakoff & Turner
1989: 86–87; Lakoff 1993a: 219–229).
.. Conceptual integration through blending
So far, metaphor has been described as the process of mapping the informationstructure of a concrete conceptual source-domain onto a more abstract targetdomain. Turner and Fauconnier have argued that for some metaphorical transfers this view is too restrictive. In their opinion, metaphorical mapping cannot
always be described as a unidirectional projection of informational structure
from the source to the target, but very often the projection must be regarded as
interactive.19 It is therefore necessary to complement the previous account of
metaphorical extension by Fauconnier and Turner’s description of conceptual
blending alternatively called conceptual integration (Fauconnier 1997; Fauconnier & Turner 1998; Turner & Fauconnier 2000).
The process of conceptual blending was originally described in the context
of Fauconnier’s mental space theory (Fauconnier 1994, 1997). Mental spaces are
abstract conceptual structures that are constructed in the minds of cognitive
agents when they process meaning in discourse. Conceptual blending is one
central cognitive process that depends crucially on mappings between mental
spaces. Thus, Turner and Fauconnier argue that when a metaphor is processed,
a blend is constructed in which important cognitive work is accomplished: in
the blend conceptual elements of the source and target domains are related and
thus interact (see Fauconnier 1997: Section 6.4). In other words, the blend consists of an emergent conceptual structure which feeds on the cognitive content
of the source and target domains and is created by the partial mapping and integration of conceptual elements from both of these domains. The source and
target domains thus work as input spaces for the hybrid conceptualisation in
the blend.
Blending increases the structural complexity of the conceptual network
relationships that are established when a metaphor is processed (for a good
illustration see Fauconnier 1997: 150). To exemplify this, Turner and Fauconnier (2000: 133) analyse the expression: “if Clinton were the Titanic, the ice-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.15 (810-846)
berg would sink.” This blended metaphor cropped up in 1998 when the film
‘Titanic’ was very popular and President Clinton was about to survive the
Lewinsky sex-scandal. The metaphor has a source (the titanic-scenario) and
a target domain (the sex-scandal-scenario), which function as input spaces
to the blended space. Following conventional metaphorical extensions conceptual correspondences are established by projecting informational structure
of the source onto the target: the titanic and iceberg slots in the source
are mapped onto the corresponding bill clinton and sex scandal slots in
the target. Consequently, the following metaphorical analogy is established:
titanic : iceberg = clinton : sex scandal. However, the metaphorical
counterfactual cannot be understood on the basis of these correspondences
only. To understand the actual meaning of the statement, a blended space
has to be constructed. The blend involves the hybrid conceptualisation of the
clinton-titanic hitting the sex-scandal-iceberg. The blend imports most
of its emergent domain structure form the titanic-scenario. Thus, it includes
a ship that is travelling to a destination but runs into an iceberg. However,
the causal structure and the blend’s overall event-shape are imported from
the sex-scandal-scenario. In this real-life scenario Clinton is not removed
from office but survives the Republican accusations. The blend thus creates
the emergent domain structure in which the clinton-titanic runs into the
sex-scandal-iceberg with the consequence of the iceberg sinking.
In short, blended metaphors cannot be understood on the basis of a unidirectional mapping process but depend on the interactional mapping of conceptual structure from the source to the target, and from these domains into
the blend.20 This results in the dynamic construction of an emergent hybrid
domain (clinton-titanic runs into sex-scandal-iceberg) that is empirically unmotivated, i.e. it cannot be related to any concrete actual experience.
Conceptual blending thus nicely illustrates the creative power of conceptual
projection and is a strong indicator of the autonomous cognitive modelling of
imaginary projected worlds.
.. Metonymy
In the same way as metaphor, metonymy plays a fundamental cognitive role
for the construction of elaborate conceptual structures. Metonymy involves the
process of “establishing connections between entities which co-occur within
a given conceptual structure” (Taylor 1995: 123). For instance, when saying
Yesterday, I drank seven bottles the conceptual connection container for
content is established; or in Hollywood has been producing a lot of nonsense

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.16 (846-888)
recently the mapping place for institution is evoked. Cognitive linguistic analyses describe metonymy as a mapping procedure that establishes a
coordinated link between contiguous conceptual substructures (for a good
overview see Barcelona 2000a, 2000b). For instance, the container-concept
bottle is a salient substructure in the cognitive-model drink. The contiguous
conceptual association between bottle and liquid in the bottle underlies
the metonymic expression I drank seven bottles. The metonymic standard or
metonymic source bottle thus works as a conceptual “reference point” or peg
to provide easier mental access to the associated, but less salient conceptual
substructure content – the metonymic target (Langacker 1993: 5).
Metonymy thus differs from metaphor because it does not establish a
mapping relation between two distinct conceptual domains; rather, it involves semantic extension within the same experiential domain (cf. Barcelona
2000a: 4).21 Moreover, metonymy is not based on similarity but on conceptual contiguity (see Lakoff & Johnson 1980; Croft 1993; Taylor 1995; Barcelona
2000a).
In a nutshell, with metonymic extension the most salient concept within
a given conceptualisation is evoked as a standard to access a conceptually
contiguous target.
.. Emblems
In their contrastive study Symbole in Sprache und Kultur, Dobrovol’skij and
Piirainen (1997) analyse the phenomenon of cultural and linguistic symbolism
in idiomatic expressions. This conceptual phenomenon is generally ignored in
the cognitive-linguistic literature. I will include it in my framework because
it plays an important role for the semantic structuring of idioms. However,
since the term symbol is heavily loaded in linguistics, I prefer to use the term
emblem as a substitute for Dobrovol’skij and Piirainen’s corresponding notion
of Symbol.
An emblem is a stereotypical conceptual prototype that works as the material representation of a very abstract quality or attribute. Consider expressions
like man is a wolf, he is such a pig, she is wise as an owl, he is a lion. Rather
than reflecting the general conceptual metaphor people are animals, these
expressions derive their meaning from the conventionalised emblematic values attributed to the animal concepts. In many cultures, wolf, pig, owl and
lion are seen as incarnations of the very abstract attributes evil, dirtiness,
wisdom and courage. In other words, the status of these animal concepts as
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.17 (888-943)
emblems consists in their function to work as figurative vehicles to represent
the stereotypical attributes ascribed to them within a given cultural system.
Developing a typology of emblems, Dobrovol’skij and Piirainen (1997:
Chapter 2) show that emblems can have different degrees of strength. A strong
emblem is commonly known in a linguistic or cultural community. For instance, in Christian cultures the cross-emblem has a well-entrenched and
highly salient cognitive status. In contrast, the fish-emblem, which was used
with approximately the same function by early Christians, can no longer be
regarded as generally known. Since the knowledge of its emblematic value is
restricted to a number of experts or specific subcultures, it must therefore be
classified as a weak emblem.
.. The cognitive interaction of metaphor, metonymy,
blending and emblems
Discussing the process of metaphor composition, we have observed that conceptual metaphors may involve the coordinated activation of different sourcedomains to construct a complex metaphorical conceptualisation of the targetdomain (e.g., career as an upward journey to a higher destination).
We have further seen that source and target domains may interact in terms
of conceptual blending to create a hybrid, blended conceptualisation for a
given constellation (e.g., the clinton-titanic ramming the sex-scandal
iceberg). These complex conceptual interactions between different cognitive
models thus reflect the creative but coherent cognitive construction of novel
cognitive models. In the same way as conventional cognitive models, these
models function as projected worlds: they are mentally constructed to group a
given experience for the sake of easier mental access or better comprehension.
The complex interaction between alternative cognitive-models is not only
restricted to metaphorical mappings. In cognitive-linguistics it has become
a truism that metaphor and metonymy (and with them blending and emblems) do not appear in isolation, but may interact in intricate ways (see Lakoff
& Turner 1989: 104–106; Goossens 1990; Gibbs 1994: 449–451; Barcelona
2000a: 10). Thus, the coordinated activation of different conceptual networks
may involve the cognitive interaction between all processes of semantic extension outlined above. This reveals the great creative potential underlying the
human construction of cognitive models of the world.
The conceptual interaction of metaphor and metonymy results in complex
patterns of metaphtonymy (Goossens 1990: 323). For instance, the idiom grit
one’s teeth has a metonymic basis. When trying hard to suppress their pain,

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.18 (943-992)
people literally grit their teeth. The action thus constitutes a metonymically associated physical symptom for the underlying process of repressing the feeling.
As soon as grit one’s teeth is used to refer to the suppression of emotions and
thoughts without the affected person actually gritting the teeth, the expression
becomes metaphorical: the minister seems to have gritted his teeth at yesterday’s
meeting. The idiom can therefore be regarded as a metaphor from metonymy
or a metonymically motivated metaphor.
A more intricate interaction of metonymy and metaphor underlies the expression shoot one’s mouth off (talk foolishly about things one does not know
or should not talk about). The idiom metaphorically associates the uncontrolled use of a gun with a more general ill-considered activity (see Goossens
1990: 334). However in the source-domain of the metaphor the concept gun
is replaced by mouth (shoot the mouth off ). This substitution is motivated by
metonymic mapping. mouth metonymically stands for the domain speech
faculty. By integrating this metonymy with the metaphor silly activity is
shooting off a gun, a hybrid cognitive model is created. In other words, this
idiom is shaped by a process of metaphtonymic blending: ill-considered speech
can thus be construed metaphtonymically as shooting off the mouth-gun (also
cf. Turner & Fauconnier 2000).
Complex conceptual interaction with other metaphors and metonymies
also affects emblems (cf. Dobrovol’skij & Piirainen 1997: Section 3.3.4). For
instance, the idiom be in seventh heaven combines the number emblem seven
with the conceptual metaphor better is up. A metonymy-emblem combination can be found in a golden handshake (a large payment given to a person
leaving a company or organisation): the handshake metonymically stands for
leaving and golden works as an emblem for wealth.
Summing up, metaphor, metonymy, blending and emblems do not have to
occur in isolation, but are frequently combined to form elaborate conceptual
constructs. These constructs have the status of hybrid conceptualisations that
are established to model, group and organise a given target-conceptualisation
by exploiting directly accessible, empirically concrete cognitive models. Taking
the intricate complexity of patterns of semantic extension and their potential interaction into account, the cognitive space of conceptual structures must
be regarded as an extremely complex system of “hyperdimensional meanings”
(Bates et al. 1998: 590); the architecture of conceptual representations thus constitutes a multi-dimensional web of meaningful configurations that may be
potentially projected onto a given experience to group and comprehend it.
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.19 (992-1053)
. Cognitive Grammar: The mental representation
of linguistic knowledge
The central stance of cognitive linguistics is that grammar is organised by the
same basic cognitive processes as all other mental representations (cf. Langacker 1987: 1). For the cognitive basis of grammar this entails the following
correspondences:
a. By analogy with concepts, i.e. knowledge units which work as meaningful
schemas, symbolic units work as basic linguistic schemas.
b. Functional embodiment corresponds to the usage-based nature of Cognitive Grammar: interaction with the world of linguistic experience is the
source of grammar formation.
c. The creative construction of complex conceptual models is reflected in the
ability to create complex and novel grammatical constructions.
d. The relationship between concepts and the experiences that they categorise
and group corresponds to the linguistic coding problem, i.e. the relationship between symbolic units and the richer conceptualisations that they
represent.
In the following sections these correspondences are outlined.
.. Symbolic units and linguistic semantics
Following the Saussurian characterisation of linguistic symbols as formmeaning associations, linguistic units are defined as bipolar symbolic units
that consist of a semantic and a phonological pole. Form and meaning are
linked through symbolisation, i.e. the entrenched association between the semantic and phonological structures (Langacker 1987: 77). Since phonological
structuring is not of direct relevance for the analysis of idioms, the following
presentation only focuses on the semantic pole.
To denote the semantic pole of a linguistic unit (either morpheme or construction), Langacker uses the notion of predication (Langacker 1987: 97). The
cognitive-linguistic approach to meaning is conceptual in nature; conceptualisations provide the actual content of predications (cf. Langacker 1997: 235–
236; Croft 2001: 19). Moreover, cognitive-linguistic views of semantics reject
the “building-block view of linguistic meaning;” according to this view, words
are containers for semantic features which can be added compositionally to derive the meaning of a linguistic expression (Lakoff & Johnson 1980: 202–203;
cf. also Kemmer & Barlow (2000: xxi). Cognitive linguists claim that linguis-

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.20 (1053-1114)
tic items do not compose meanings, rather, they construe meanings. Using the
notion of designation, Langacker shows that different linguistic constructions
attribute different degrees of prominence to particular substructures in a given
conceptualisation (Langacker 1987: 116). Depending on the designation, the
same conceptual content can be construed in different ways: the book on the table (designating the book), the table with the book on it (designating the table),
the book is on the table (designating the relationship between book and table).
Alternative cognitive construals of the same conceptual content thus lead to
different semantic representations. To delimit its precise semantic value, every
predication therefore includes: i) the conceptual content evoked by the symbolic unit, and ii) its conventionalised construal, i.e. the designation, of this
content (Langacker 1991b: 3–4, 1998: 4).
The construal of conceptual content depends on the cognitive ability to
make a variety of focal adjustments that mould a given conceptualisation (Langacker 1987: 117). To delimit the scope of predication of a linguistic unit, the
cogniser must focus on a specific area within an extensive network of conceptual relations (Langacker 1991b: 7). This process of selection makes it possible
for him or her to determine the specific aspects of a conceptualisation that are
in focus (Langacker 1987: 493). Within this selected scope of predication every
predication further highlights a particular substructure. This substructure is
called the profile; the background against which the profile is set off is its base
(Langacker 1987: 183). For instance, the verb buy selects the complex commercial event-ICM as its scope of predication. Relative to this conceptual base
(consisting of the elements buyer, goods, seller, money), buy profiles the interrelation of the buyer and good(s) concepts. A linguistic predication (e.g.,
the meaning of buy) thus involves both a conceptual base (a conceptual representation with a delimited scope, e.g. the commercial event-script) and a
superimposed profile (the highlighted buyer and good(s) interrelation).
Since predications are construed conceptualisations, this symbol-based
profiling of mental experience involves a further cognitive grouping-process.
The specific construal that a symbolic-unit imposes on a conceptual base constitutes the language-specific aspect of meaning: linguistic meaning is “conceptualisation tailored to the specifications of linguistic convention” (Langacker
1987: 99); “language acts to filter or emphasise what is foregrounded” (Hays
2000: 167). Linguistic meaning thus adds a further dimension to the structuring of knowledge. Working as a conventionalised and socially-shared “preconfigured conceptual system” (Hays 2000: 161), language provides us with “a
system of shared thought – a system of ‘collective representations”’ (Kroenenfeld 2000: 198).
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.21 (1114-1179)
.. Grammar as a structured inventory of symbolic units
Cognitive Grammar is characterised as a complex network that consists of a
myriad of symbolic-units. Langacker’s model of grammar is an emancipatory
counterproposal to the modular and algorithmic models of language offered by
the Chomskyan paradigm. A very central point of criticism against Chomskyan
models is their “building-block” conception of language, which is implied in
the modular view of linguistic knowledge and involves the description of sentence formation as the combination of lexical items according to the combinatory principles provided by a formal syntax (Langacker 1987: Section 12.1.2.;
Lamb 1999, 2000). Instead of establishing a dividing line between syntax and
the lexicon, Langacker (1987: 57) understands a cognitive grammar as a “structured inventory of conventional linguistic units,” which function as entrenched
and autonomous communication routines. Traditionally, the idea of an inventory of linguistic units fits only part of the grammar, namely, the lexicon.
Following his content requirement, Langacker, however, extends the notion of
symbolic unit unorthodoxically to include grammatical morphemes, syntactic
categories and any other type of grammatical construction:
[...] the only units permitted in the grammar of a language are (i) semantic,
phonological, and symbolic structures that occur overtly in linguistic expressions; (ii) structures that are schematic for those in (i); and (iii) categorizing
relationships involving the structures (i) and (ii). I call this the “content requirement,” and consider it to be intrinsically more restrictive [...] than the
constraints generally imposed on algorithmic models. (Langacker 1991b: 18)
In accordance with this requirement, syntax is not seen as a relatively isolated formal system. In contrast, it is interpreted in semantic terms. In the
same way as lexical items, syntax is a means to construe and structure conceptual content. Along with this semantisation of syntax, the distinction between
discrete grammatical modules (lexicon vs. syntax) is abolished and replaced
by a “continuum of symbolic structures” (Langacker 1987: 54). On this continuum traditional syntactic, morphological or lexical structures constitute
form-meaning pairings with different degrees of complexity and schematicity (Langacker 1991b: 16). The zero point of complexity is expressed by the
traditional notion of morpheme – the smallest meaningful grammatical unit.
Starting from this minimum degree of complexity, symbolic units can virtually achieve any degree of structural complexity. Thus, even a proverb like Too
many cooks spoil the broth constitutes a symbolic unit because it is subject to
cognitive routinisation (for a good overview see Croft 2001: 17, Table 1.3).

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.22 (1179-1181)
Following the rejection of an autonomous syntax and lexicon, Langacker
further replaces the idea of autonomous syntactic rules by the notion of
constructional schema (Langacker 1991b: 17). Constructional schemas are internalised formulation-templates that have the form of complex and highly
schematic symbolic units. In accordance with the content requirement, constructional schemas can only arise via abstraction from the complex forms and
meanings that occur in actual use. Constructional schemas comprise the commonalties observed in sets of expressions that are symbolically complex. By nature, then, these templates are exactly parallel to the complex expressions that
they categorise except for their lesser degree of specificity. For instance, the sentence John sliced the tomatoes can be related to various constructional schemas
that define its structural description (Langacker 2000a: 17), (see Figure 3.3).
While the abstract schema [sbj process obj] structures the sentence as a transitive clause, the templates [[process] [past]] and [[thing] [plural]] specify
the morphological structure of the verb sliced and the noun tomatoes, respectively. Note, however, that the schematic values [sbj], [process], [obj], [past],
[thing] and [plural] are seen as abstract semantic structures rather than autonomous syntactic labels. Thus, in accordance with the content requirement,
the structural organisation of constructional schemas is motivated and patterned by (abstract) semantic relationships. The morphosyntactic nodes of the
construction constitute very abstract, image-schematic profiles of conceptual
[TRANSITIVE CLAUSE]
[THING]
[[THING] [PLURAL]]
[SBJ PROCESS OBJ]
[tomato]
[PROCESS] [[PROCESS] [PAST]]
[SBJ slice OBJ]
(tomatoes)
[slice]
(SBJ slice the tomatoes)
(sliced)
(SBJ sliced the tomatoes)
(John sliced the tomatoes)
Figure 3.3 Constructional schemas (cf. Croft 2001: 26, Figure 1.11)
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.23 (1181-1229)
substructures that are characterised relative to schematic conceptual domains
(see next section).
The example further illustrates how symbolic units are organised in the
form of a structured inventory rather than unstructured lists. This inventory
has the shape of a complex network. In this network, symbolic units of various types constitute the nodes which are linked taxonomically. To fully represent a speaker’s knowledge of the conventions of his or her language, any
symbolic unit with unique or idiosyncratic phonological, morphological, syntactic, lexical, semantic, pragmatic or discourse-functional properties must be
represented as a specific node. In neuro-cognitive terms symbolic units and
constructions thus have the status of complex mental activation patterns or
activation-sets:
[...] linguistic units are seen as cognitive routines. Such units are nothing
more than recurrent patterns of mental (ultimately neural) activation; as such
they are not stored in any particular neural location, [...]. During linguistic processing, linguistic units are part and parcel of the system’s processing
activity: they exist as activation patterns. When no processing is occurring,
the information represented by such units simply resides in patterns of connectivity (including differential connection strengths) resulting from previous
activations.
(Kemmer & Barlow 2000: xii)
Thus in accordance with the general description of knowledge as a system of
mental schemas that are derived from peripherally-connected sensorimotor
experience for the sake of organising it, the nature of these linguistic activationsets amounts to a schematic representation of concrete linguistic performance.
.. The usage-based nature of language and its relevant coding
Feeding back into novel utterances, the schematic grammatical distillate of
networked symbolic units and constructions is constantly re-instantiated in
the production of novel utterances. Via comprehension processes, the resulting performance again works as input to the linguistic knowledge systems of
other speakers and so forth. In this way, the cognitively represented language
systems of interacting speakers and hearers are accommodated and result in
the construction of a mutually shared grammar (cf. Croft 2000: Chapter 4;
Kroenenfeld 2000: 203). In other words, Cognitive Grammar is usage-based
(Langacker 1987: 46, see also Langacker 1991b: Chapter 10), i.e. the source
of linguistic knowledge resides in the concrete experience of linguistic performance in actual language use. Grammar constitutes “a competence to perform”

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.24 (1229-1296)
that is derived from and specifically designed for the necessities of actual communication (Lamb 2000: 95).
In concrete events of linguistic processing, the mental activation-sets that
define a speaker’s linguistic competence must be activated to coordinate the
production and comprehension of actual utterances. This is what Langacker
(1987: 65–66) calls the problem of coding:
The task of finding an appropriate linguistic expression for a conceptualization can be referred to as the problem of coding; its solution is a target
structure [...]. The target is therefore a usage event, i.e. a symbolic expression
assembled by a speaker in a particular set of circumstances for a particular purpose: this symbolic relationship holds between a detailed, context-dependent
conceptualization and some type of phonological structure (in the case of
speech, it is the actual vocalization).
According to Langacker (2000a: 10) the coding problem reduces to “a matter of categorisation.” Linguistic units can be employed by a speaker/hearer
to categorise and activate rich conceptualisations and vocalisations in order to exchange these information structures with communicative partners
(Figure 3.4).
Linguistic production and comprehension thus coalesce to become two
facets of linguistic categorisation events: in production, linguistic routines (activation patterns) are retrieved to code a rich target-conceptualisation in a
symbolic format. In relevance theoretical terms, such a coding target is embedded in the mentally constructed cognitive environment that constitutes the cogniser’s manifest representation of context (Sperber & Wilson 1986: Chapter 8).
To share this conceptualisation as a mutually-manifest cognitive environment
with the hearer, the speaker must code his intended communicative-target
in an optimally relevant way. This works by grouping the conceptualisation
in terms of matching linguistic units that seem to guarantee the principle of
relevance (cf. also Sperber & Wilson 1995: Chapter 2). In comprehension, a
rich vocalisation must be compared to the more schematic phonological standards provided by linguistic knowledge to evoke a corresponding semantic
representation, which can be mapped to construct a context-specific targetconceptualisation, i.e. to derive contextual implicatures.
Obviously, the process of coding is not an individualistic activity that takes
place in a cognitive vacuum. Rather, coding is always embedded in a pragmatic
scenario of communication between a producer (speaker/writer) and one or
more (implied) receivers (hearers/readers). To share the target conceptualisation with the receiver, the producer must perform a complex communicative
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.25 (1296-1319)
USAGE CONTEXT
GRAMMAR
symbolic unit
semantic unit
usage event
coding
phonological
unit
conceptualisation
vocalisation
coding
STANDARD
TARGET
CATEGORISATION
Figure 3.4 Categorisation of a usage event; adapted from Langacker (1987: 67, 77)
tightrope walk. The major communicative problem is that the meaning of a
linguistic construction almost always underspecifies the rich contextual conceptualisations evoked with a given thought, idea or feeling. That is, linguistic
constructions cannot be fully explicit about the information that the producer
attempts to encode. Therefore, the communicator must attempt to trigger additional contextual effects in the hearer to coordinate and accommodate their
cognitive environments (Sperber & Wilson 1986: Chapter 7). Thus, in (1), B
communicates information that does not explicity answer A’s question.
(1) Speaker A: Want some tea?
Speaker B: I just had coffee, thanks.
But rather than being an irrelevant answer, B’s contribution points to additional, implied information that A must reconstruct cognitively. Thus, to
attune his cognitive environment to the supposed intention of B, A must derive
additional contextual effects that transcend a mere negation:
a.
b.
c.
d.
A learns that B has just had coffee
A knows that coffee is another warm drink
A knows that coffee, like tea, contains caffeine
A concludes that B probably does not want to drink coffee because

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.26 (1319-1378)
e.
f.
g.
h.
B has had enough warm drinks for the moment
B does not want to consume more caffeine now
B is reasonable
B is ascetic
In other words, B communicates a mere literal proposition (a) that has the
potential of triggering a number of implicated premises (b–c) and implicated
conclusions (d–h), (Sperber & Wilson 1986: 195). This implied information
enhances the relevance of B’s explicit proposition because it unfolds a great
number of additional contextual effects:
The notion of a contextual effect is essential to a characterization of relevance.
We want to argue that having contextual effects is a necessary condition for
relevance, and that other things being equal, the greater the contextual effects,
the greater the relevance.
(Sperber & Wilson 1986: 119)
In other words, the coding of a target conceptualisation is geared to create
its optimal relevance, i.e. to trigger a great number of contextual effects by
consuming as little processing energy as possible. Thus,
Every act of ostensive communication communicates the presumption of its
own optimal relevance.
It is important to emphasise that not all implicated premises and conclusions
that can be triggered by the receiver must also be intended or controlled by
the producer. Linguistic communication is not a telepathic 1:1 transfer from
one mental container to the other. Rather, the communication of an idea always involves the risky potential to trigger contextual effects that do not belong
to the cognitive environment and the intended target-conceptualisation communicated by the speaker. For instance, the implicated conclusions (g) and
(h) are less likely to be intended by B than the most relevant conclusion (d).
Nevertheless, (g) and (h) are relevant and potential conclusions that can be
drawn from the B’s statement. In other words, the implicatures of an utterance vary in strength. Sperber and Wilson (1986: 199) therefore establish the
distinction between strong implicatures and weak implicatures. Strong implicatures are those cognitive effects (implicated premises and conclusions) for
which the speaker can be claimed to take full responsibility. Strong implicatures
are therefore most likely to reflect the intended meaning and most relevant
information to be shared with the receiver in a mutually-manifest cognitiveenvironment. In contrast, weak implicatures are those conclusions that cannot
be fully controlled by the speaker, but may create additional, indeterminate and
less relevant communicative effects:
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.27 (1378-1421)
A speaker who constrains the interpretation of his utterance so that the hearer
takes very little responsibility in the choice of contextual assumptions and contextual effects is said [...] to be engaging in strong communication. The greater
the responsibility the hearer has in the selection of contextual assumptions
and effects, the weaker the communication.
(Blakemore 1992: 157)
The gradual cline between the explicit content of an utterance and its strong
and weak implicatures reflect the cognitive tightrope walk that has to be performed while tackling the coding problem of finding appropriate linguistic
units to categorise a given target conceptualisation in a communicatively functional way. Thus, language use amounts to the process of activating an array of linguistic units from the large and complex category-network called
grammar. The choice is guided by the communicative urge to make one’s
linguistic contribution maximally relevant for the creation of implicatures
to construct mutually-manifest cognitive environments, i.e. shared targetconceptualisations.
.. From conceptualisations to grammatical constructions – the coding
of conceptual content
Language production amounts to coding complex conceptualisations by means
of elaborate grammatical constructions. This creative linguistic categorisationprocess results in the coordinated activation of a set of linguistic units. The
mental construction of a linguistic activation-set is constructed on the basis of
an intricate interaction between the target conceptualisation and the cognitive
grammar-network.
The transformation of conceptual content into symbolic structure is constrained by the linear nature of language. Hence, the problem of coding can be
pared down to the problem of reducing the complexity in multi-dimensional
conceptualisations by manipulating them cognitively in order to create a maximally relevant grammatical construction. With reference to the following fictive restaurant-kitchen scenario, I will shortly sketch Langacker’s model for the
coding of transitive constructions:
Another busy Saturday evening in the restaurant. The kitchen staff have been
preparing meals since 6:00 pm. The atmosphere is very hectic because a great
number of customers are waiting for their food. In this hustle and bustle one
cook, Pete, eventually gives Frank, the chopper, an unvoluntary push. Frank,
who has been chopping onions for five ours, is so annoyed that he swings
the razor-sharp chopping knife and cuts Pete’s finger. For the pain, Pete cries
out in a loud voice. Blood runs down his arm and stains the kitchen floor.

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.28 (1421-1485)
Hearing this uproar, the restaurant manager enters the kitchen and asks what
happened. One cook answers:
(2) Frank has cut Pete’s finger.
Relative to the complex and rich conceptual scenario, the finite transitive clause
in (2) conveys far more limited information. The speaker therefore has to make
a number of coding decisions. Essentially, he must manipulate his complex cognitive environment along the coding parameters of meaning construal that were
introduced above (see Langacker 1991b: Chapter 9). This involves:
a. the selection of the scope of predication and profiling the specific scene
within this scope
b. defining the level of abstraction and granularity at which the scene is construed
c. determining the relative prominence of the profiled substructures
To define the scope of predication, the complexity of the conceptualisation is
reduced by segmenting it into subevents and selecting one subevent for coding. I will call such a subevent a scene. In (2) the cutting-scene is selected for
coding because it comprises the most relevant information for the manager’s
immediate comprehension of the situation. To code the scene, the transitive
construction [sbj cut obj] is selected; the construction imposes a symbolic
profile on the scene and selects the agent Frank and the patient Pete’s finger
as its substructures (the subject and object, respectively). With the choice of
the construction, the cutting-scene thus emerges as the symbolically designated profile which is set off from the conceptual base of the kitchen-scenario.
Thus, relative to the entire scenario, (2) merely conveys the cutting-event and
its immediate participants.
Variance in coding emerges when the cogniser makes focal adjustments to
increase or reduce the degree of abstraction and granularity of the underlying
conceptualisation. Thus, while (3) is more abstract than (2), (4) describes the
event more elaborately and with increased granularity:
(3) Frank injured Pete.
(4) The edge of the chopping knife went through the skin of Pete’s forefinger
almost to the bone.
Given these parameters of cognitive construal, “the objective properties of a
situation do not mechanically determine the grammatical organization of a
sentence or finite clause describing it” (Langacker 1991b: 214). Rather, the cog-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.29 (1485-1531)
nitive construal of a situation is motivated by communicative needs; it can be
seen as the cognitive correlate of constructing a maximally relevant construction in a context-specific usage event. The most conventional and cognitively
neutral way to construe the situation is to conceive the scene at the basiclevel as in (2). As postulated by Carter (2004: 115–117), this unmarked form
of coding corresponds to the use of unmarked core vocabulary (proper names:
John, Pete and basic words: cut, finger) and core constructions (the transitive
construction).
Thus although the cognitive competence to construe a situation along the
coding parameters supports the creation of alternative and variable grammatical constructions, the creation of a given construction is by no means arbitrary;
rather, it is constrained by the communicative situation as well as the archetypal conceptual relationships perceived in the target-conceptualisation. English
transitive clauses, structured by the constructional schema [sbj verb obj], prototypically encode causal events which involve very basic scenes of physical
experience (cf. Langacker 1991b: 212; Croft 1991, 1998: 79–84).
Langacker proposes “the billiard-ball model” as a heuristics to explain the
conceptual basis that underlies the prototypical coding of transitive clauses (see
Langacker 1991a: 209). Constituting an archetypal blueprint for our cognitive
construction of events, the ontology and structure of the billiard-ball model
is essentially physical in nature: the world is conceived as being populated by
discrete things or objects that interact with other objects and are thus interconnected through force-dynamic relations (Talmy 1988, 2000). These forcedynamic interrelations are cognitively represented by image-schemas such as
impulsion, hindering or containment. More specifically, the billiardball-ICM conveys the basic idea that an agent transmits kinetic energy to an
impacted patient which, as a result, changes its state. The transmission of energy between these semantic role archetypes proceeds along an action chain (cf.
Langacker 1991a: Section 4.2).22 An action chain is a causal sequence of interactions between different objects (Langacker 1991a: 283); minimally, such an
action chain consists of two participant objects: an energy source (the agent,
ag) and an energy sink (the patient, pat). Moreover, the transmission can be
mediated by an instrument (instr). Cognitively, these elements are arranged
according to the energy flow hierarchy ag > instr > pat (Langacker 1991a: 238).
On the basis of this abstract ICM for causal event structure, the cognitive organisation of canonical transitive clauses and the corresponding grammatical
relations can be explained.

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.30 (1531-1583)
The syntactically central notions of subject and (direct) object receive their
prototypical values from the force-dynamics that pattern the schematic action
chain ag > instr > pat:
(5) Pete pushed Frank.
(6) Frank cut Pete’s finger with a chopping knife.
As reflected in (5) and (6), the prototypical transitive subject is the agent,
the volitional energy source (Pete, Frank) that initiates the action chain. The
object is the patient, the energy sink (Frank, Pete’s finger) that undergoes a
change of state. The subject and the object are the most central participants
in the transitive process profiled by a finite clause (Langacker 1991b: 213). All
other participants are less central. Thus, the instrument (like chopping knife in
(6)) is added in terms of a prepositional phrase. In other words, the constructional schema [sbj verb obj] can be attributed the status of an autonomous
symbolic unit with the prototypical meaning [ag changes (physical) state
of pat]. Thus, whenever a target-conceptualisation is construed by imposing the constructional schema [sbj verb obj], the syntactic slots sbj and obj
are typically mapped onto the semantic roles of agent and patient in the
target-conceptualisation (see also Goldberg 1995; Croft 2001).23
.. The creation of novel grammatical constructions
As exemplified by the [sbj verb obj]-schema, a cognitive grammar provides
a myriad of entrenched symbolic patterns that can be directly activated to
code the complex conceptualisations that emerge in usage-events. A speaker
can thus depend on a great number of pre-assembled units that facilitate the
coding process (cf. Langacker 1987: 314). Thus, Cognitive Grammar assumes
linguistic constructions to be far more pre-established than suggested by the
atomistic conception of linguistic knowledge advocated in generative theories
of grammar. Constructional frozenness therefore does not constitute a linguistic anomaly, but it is seen as a fundamental characteristic of how linguistic
knowledge emerges through cognitive routinisation. Thus, Langacker is forced
to explain linguistic creativity and variability rather than frozenness.
Linguistic creativity and variation reflect the communicative necessity of
adapting the linguistic means to the variant conceptualisations of heterogeneous social and physical experience and to the pragmatic need of making one’s
contributions optimally relevant (cf. Langacker 1987: 278; Sperber & Wilson
1986, 1995). Linguistic creativity is the cognitive ability to respond to these be-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.31 (1583-1638)
[THING]
[THING]
[PROCESS]
[PROCESS] [ER]
[SHARPEN]
[PROCESS] [ER]
[THING THING]
[THING]
[PROCESSOR]
[THING PROCESSOR]
[PENCIL]
[SHARPENER]
[PENCIL SHARPENER]
Figure 3.5 Constructional schema for pencil sharpener (based on Langacker 1991b: 18)
havioural demands. The relevant coding of variant experiences leads to creative
modifications of linguistic routine. Being sensitive to changes in social, cultural
and physical experience, the cognitive grammar network thus changes dynamically; it is re-performed as “a constantly evolving set of cognitive routines that
are shaped, maintained, and modified by language use” (Langacker 1987: 57).
Since the cognitive grammar contains many pre-established constructions,
the creative activity of the speaker mainly consists in elaborating or manipulating the nodes of these constructions relative to sanctioned constructional
schemas and by making coding decisions along the coding parameters that
were outlined above.24 Langacker (1987: 439) describes this process as the mental construction or computation of a non-entrenched standard (S) by activating
and coordinating a set of conventional linguistic units [U]. This can be illustrated by the symbolic unit pencil sharpener, which is organised in terms of
the following network of lexical units and constructional schemas (see Figure
3.5). To be able to create a novel expression by manipulating this symbolic
unit, it is essential for a speaker to realise the contribution of the symbolic constituents relative to its overall meaning. Pencil sharpener is patterned by the
immanent constructional schemas: [thing processor] and [thing sharpener]. Imagine the situation of an artist, who uses crayons rather than pencils
to draw his oeuvres. This artist has a invented a new tool that he uses to sharpen
his crayons. This specific thing is the target-conceptualisation (T = tool to

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.32 (1638-1653)
sharpen crayons). To give a precise and relevant linguistic categorisation
of this instrument, the artist might coin the novel expression crayon sharpener. To do this, he is forced to compute the novel, context-specific standard,
(S = (crayon sharpener)). This novel standard does not have to be created
through bottom-up composition. Rather, it can be established by exploiting
the pre-established constructional schema [U = [thing sharpener]] by coordinating it with the lexical unit [U = crayon]. This lexical unit can be inserted
into the schema by analogy with the unit [U = [pencil sharpener]]. In other
words, (S) is largely prefigured by the constructional schemas [thing sharpener] and [pencil sharpener] as well as by the categorisational relationship
[pencil] → [crayon]. In other words, the artist’s linguistic creativity only involves the manipulation of the [thing]-node in the constructional schema,
which is elaborated according to the prefigured taxonomic hierarchy [thing]
→ [pencil] → [crayon]. We can therefore regard the process underlying the
coinage of this novel compound as an elaborative form of analogy. Following
Carter’s view of linguistic creativity, this amount to an elaborate process of
linguistic pattern re-forming (Carter 2004: 102–103).
Ignoring the phonological pole, this process of creative grammatical composition works in two dimensions: a ‘horizontal’ dimension that captures the
syntagmatic integration of the component units and the ‘vertical’ dimension
of composition that defines the relationship between the components and the
composite structure (cf. Langacker 1987: 84, 448):
COMPOSITE
STRUCTURE
C
COMPOSITION
COMPONENT 1
A
COMPONENT 2
B
syntagmatic
INTEGRATION
Figure 3.6 The creation of a composite construction
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.33 (1653-1719)
Novel grammatical constructions such as crayon sharpener thus emerge as
integrated, creative combinations of conventional symbolic units. Langacker
develops a model of linguistic composition that interprets this process as a
complex form of categorisation. Langacker’s view of composition is a rejection of the standard view, which attributes full compositionality to grammatical constructions and their semantic interpretation. In contrast, Langacker
(1987: 448) discusses composition by distinguishing the two related but distinct
concepts of compositionality and analysability (see Figure 3.6).
Compositionality focuses on the derivational predictability or regularity of
the composite structure by adding up the values of its component parts; in
other words, compositionality describes the bottom-up perspective on composition. When scrutinising the predication of crayon sharpener, one notices that
its intended meaning does not emerge as a direct compositional result of the
individual meanings of its components: a crayon sharpener is not simply ‘something or someone that sharpens crayons’ but a specific tool. In other words, this
specialised meaning cannot be constructed bottom-up on the basis of the constructional schemas and the lexical units; rather, it is dependent on the specific
conceptualisation which it profiles (the novel tool invented by the artist). The
non-predictability of the composite meaning suggests that the constructional
schema [thing sharpener] does not work as a generative device:
The meanings of component elements evoke and constrain the meaning of a
composite expression but cannot be said to constitute it. An expression’s actual
semantic value results form an elaborate process of meaning construction that
draws upon all available conceptual and contextual resources [...].
(Langacker 2000b: 37)
The component elements, including the constructional schema, thus fulfil a
categorisational scaffolding function: “The constructional schema is not responsible for assembling the expression, but only for its categorization” (Langacker 1991b: 17–18). As a result, the atomistic bottom-up conception of full
linguistic compositionality can be rejected (cf. Langacker 2000b).
Although the novel compound crayon sharpener has a specialised, noncompositional meaning, the components crayon and sharpener nevertheless
pattern its internal structure and render the compound transparent: it is clear
to us why the words crayon and sharpener are evoked to encode the meaning
‘tool to sharpen crayons.’ Langacker (1987: 457) therefore introduces the supplementary notion of analysability. Analysability refers to a person’s (implicit)
awareness of the contribution of a component structure to the overall compositional value. Analysability thus captures the composition process top-down

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.34 (1719-1766)
and consists of a retrospective or post-compositional ascription of parts of the
overall composite meaning to the immanent components. While grammatical
constructions (even fully regular ones) always underspecify the rich contextual meanings that they refer to and therefore cannot be fully compositional,
the components of these constructions must point to the actual meaning that
they encode. In relevance theoretical terms, linguistic composition is therefore “geared to the maximisation of relevance” (Sperber & Wilson 1995: 260).
Accordingly, the traditional view of linguistic composition can be replaced by
Langacker’s scaffolding view: the composite activation of linguistic components is a way to provide motivated, i.e. maximally relevant, access to a more
complex and rich target conceptualisation.
In lieu of the building-block metaphor, we might adopt the scaffolding
metaphor: component structures are seen as a scaffolding erected for the construction of a complex expression; once the complex structure is in place
(established as a unit), the scaffolding is no longer essential and is eventually
discarded.
(Langacker 1987: 461)
This view directly results from Langacker’s description of composition as categorisation. The components of a grammatical construction function as signposts that coordinate the meaning construction and meaning construal in a
usage-context. While they do not work as building-blocks by means of which
the rich contextual meaning is itself constructed, linguistic expressions can
draw attention to and highlight those aspects that seem to be most relevant
for the functional communication of a specific idea by maximising cognitive
effects (Pilkington 2000: 73). As we will see below, this description of linguistic
composition is highly relevant for the description of the formal and semantic
structure of idioms.
. Summarising overview
In the following chapters, these cognitive foundations are applied to the description of idioms and idiom variation to derive a well-motivated cognitivelinguistic model of idiom representation and variation. Assuming the existence
of an external world, a peripherally-connected perceptual representation of
this world (the sensorimotor data set), and conceptual schemas to organise the sensorimotor data set, the architecture of our mental-representationof-knowledge model thus adopts the form of a multi-layered cognitive network (cf. Indurkhya 1992: 184; Lamb 2000: 181). In this network, peripherally-
JB[v.20020404] Prn:20/02/2006; 14:13
F: HCP1703.tex / p.35 (1766-1766)
COGNITION
SYMBOLIC
REPRESENTATION: GRAMMAR
Designated conceptual
substructures symbolised by
phonological units
Symbolic units
Constructions
Imposition of linguistic profile
on conceptual substructures
CONCEPTUAL
REPRESENTATION
Hybrid models
Conceptual blending
Emblematic models
Emblems
Metonymic models
Conceptual metonymies
Metaphorical models
Conceptual metaphors
Processes of semantic
extension
Cognitive models
Directly emergent concepts
Conceptual archetypes
Preconceptual structure
Bodily and cultural
interaction with
sensorimotor data set
Grouping
Categorisation
Projection
Domains
(Frames and scripts)
Taxonomic models
Image-schematic Basic-level
concepts
concepts
Creation of hybrid
conceptualisations
Attribution of emblematic
value to concept
Conceptual extension through
alternative mental access to
contiguous concept
Extension of basic
concepts and cognitive
models to domains abstract
of experience
(Re) conceptualisation
Creation of concept networks
through schematisation and
categorisation
Basic conceptualisation
PERCEPTUAL
REPRESENTATION
image-schematic ability and
basic-level structure
Preconceptual structuring
of experience
sensorimotor data set
Embodiment of meaning
perceptions
Acquisition of unordered
information
EXTERNAL WORLD
physical stimuli
Cognitive
Processes
Construction
Entrenchment
Coordination
Integration
Figure 3.7 The multi-layered architecture of meaning and language

JB[v.20020404] Prn:20/02/2006; 14:13

F: HCP1703.tex / p.36 (1766-1771)
connected events can be grouped by elaborate and embodied concept networks, including entrenched patterns of semantic extension such as conceptual
metaphor or metonymy systems. Moreover, grammar further adds a layer of
symbolic structuring which interacts with the conceptual system and adds
communicative grouping constraints through linear coding and communicative functionality, i.e. optimal relevance. Figure 3.7 should provide the reader
with a basis for easy reference to these theoretical heuristics.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.1 (47-136)
chapter 
Idiom representation –
a cognitive-linguistic model
Not only every language, but every lexeme of a language, is
an entire world in itself.
(Igor Mel’čuk, quoted from Lamb 1999: 16)
The central hypothesis pursued in this book is that many idioms can be attributed a figurative semantic structure that is motivated and analysable – they
are claimed to be intrinsically creative. This intrinsic creativity is claimed to
support variational creativity, i.e. the variable use of idioms in discourse.
Having established the cognitive-linguistic background for the description
of conceptual constructs in general and linguistic constructions in particular,
we are now in a position to reinterpret the nature of idioms in cognitivelinguistic terms. This chapter establishes a cognitive-linguistic model of the
mental representation of idiomatic constructions in Cognitive Grammar. Taking into account the linguistic and psycholinguistic insights into idiom representation and comprehension, my model is an attempt to capture the intricate
and heterogeneous semantic structures of different idioms. For this purpose, I
will integrate Langacker’s view of grammatical composition with the cognitivelinguistic theory of semantic extension (metaphor, metonymy, blending and
emblems). On the basis of this model, a number of typical figuration patterns
in idioms can be distinguished. These types will then serve as a classificatory basis for the empirical analysis of idiom-variation in actual discourse
(Chapter 7).
. Adapting idioms to the cognitive-linguistic framework
In the introduction to this book, idioms were preliminarily defined as a specific subgroup of conventional linguistic units. They are relatively frozen,
composite multi-word constructions, i.e. phrases or semi-clauses, with a non-
JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.2 (136-185)
compositional meaning and an ideational function. Moreover, idioms are institutionalised, i.e. they are part of the linguistic routine – the cognitive grammars – of the speakers of a given language. Our review of previous linguistic and psycholinguistic research on idiom representation and idiom use has
further revealed that these constructions cannot be merely described as wordlike lexical units. That is, they cannot be accounted for uniformly in terms
of autonomous lexical representation and direct lexical retrieval. Rather, the
research results emphasise the need for heterogeneous or hybrid representation and processing models. To adapt the linguistic phenomenon of idiom to
a cognitive-linguistic model of linguistic knowledge, we now have to explain
these insights according to our cognitive architecture of meaning construction
and grammatical organisation (Chapter 3).
Although Langacker does not explicitly discuss the place of idioms within
his framework, he analyses some characteristics of idioms in passing in order to criticise some central assumptions of the Chomskyan paradigm. For
instance, he questions the view of idioms as opaque and semantically unanalysable units – the view which is defended in the generative framework to
employ them as evidence for the existence of syntactic transformations (see
Nunberg et al.’s argument described in Section 2.2). In this context, Langacker
(1987: 25) provides the following general characterisation of idioms:
To regard an idiom as opaque or as primarily a fixed phrase is [...] simplistic.
It is more accurately seen as a complex of semantic and symbolic relationships
that have become conventionalized and have coalesced into an established
configuration. We can plausibly suppose that this configuration of relationships might be recognizable even when anaphora disrupts the normal shape
of an idiom, or when its component words are split up and used in grammatical constructions such that they do not form a contiguous linear sequence:
an idiom may be recognized as a unit that is to some degree independent of
a specific overt morphemic arrangement, even if one such arrangement is far
more similar and hence more “usual” than the others.
For our present purpose, the following key-passage must be highlighted: “a
complex of semantic and symbolic relationships that have become conventionalised and have coalesced into an established configuration.” This account is
fully compatible with Dobrovol’skij’s cognitive view of idioms as summarised
structures or gestalts:
Zusammenfassend lässt sich sagen, dass nicht nur der Ausdrucksplan von Idiomen, sondern auch ihr Inhaltsplan eine Art summarized structure darstellt.
Memorisiert wird in der Regel nicht ein “normgerechter” kodifizierter Wortlaut und nicht eine dazugehörige Bedeutung in der Form, wie sie in Wörter-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.3 (185-227)
Chapter 4. Idiom representation – a cognitive-linguistic model
büchern expliziert ist, sondern eine “Gestalt”, d.h. eine zusammengefaltete
Struktur, die bei der Realisierung in einer Situation unterschiedlich entfaltet
werden kann. Dieses kognitive Phänomen erklärt die berühmte kontextuelle
“Anpassungsfähigkeit” der Idiome. Zu Elementen dieser zusammengefalteten
Struktur gehören Spuren verschiedenartiger Frames und Scripts, die auf ihre
versprachlichten slots reduziert sind. Die konzeptuelle Basis für das mentale
und sprachliche Management dieser Strukturen bildet in der Regel eine Interaktion von zwei Frame, bzw. Scripttypen: einer kognitiven Ausgangs- und
einer Zielstruktur [...].
(Dobrovol’skji 1995: 56)
My cognitive-linguistic model is an attempt to flesh out Dobrovol’skij’s and
Langacker’s notions of summarised structure and established configuration, respectively. Following Mel’čuk’s introductory statement, I would like to chart
the complex cognitive universe that is encapsulated by idiomatic constructions.
In accordance with Langacker’s conception of grammar as a cognitive network of linguistic units that can be activated for coding, I will reinterpret the
notions of summarised structure and established configuration by using the
term idiomatic activation-set. An idiomatic activation-set is the mental network that can be potentially activated when an idiom is used. In other words,
this mental configuration can be projected onto rich and multidimensional
cognitive stimuli to group and encode them in an idiomatic format. In the
same way as other linguistic units, an idiomatic activation-set comprises various symbolic and semantic (conceptual) substructures that are associated with
the idiom. Put differently, the idiomatic activation-set is a complex mental configuration that consists of several coordinated symbolic and conceptual units
that constitute its immanent substructures. From a cognitive-linguistic perspective, it is the complex nature of idiomatic activation-sets that triggers the
intricate complexities discovered in the linguistic and psycholinguistic analyses of idiomatic expressions. Hence, to become able to describe the mental
status of idioms and to explain their variational behaviour, we are forced to
anatomise the internal semantic and formal structure of these configurations.
More precisely, we have to specify
a. the immanent symbolic and semantic substructures that shape the idiom,
b. the characteristic connections between these structures and
c. the variable ways in which these substructures can be activated in an actual
usage-event.
The following cognitive-linguistic model of idioms as mental activation-sets is
designed to provide a descriptive framework that is powerful enough to capture

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.4 (227-280)
qualitatively different idiomatic expressions and to explain their variable use in
discourse.
In what follows, I will first reinterpret the idiomatic characteristic of
compositeness. Thus, I will first focus on the formal dimension of idiomatic
activation-sets. Second, I will revisit institutionalisation and scrutinise the relative activation-strength of their substructures, which determines the characteristic heterogeneity of idiomaticity. I will then turn to the complex conceptual organisation that shapes the semantic structure of idioms. Finally, the
ideational discursive function of idiomatic configurations will be reconsidered
in this cognitive-linguistic light. The phenomena of frozenness and variation
will be discussed in Chapter 6.
. A cognitive-linguistic account of compositeness
To provide a cognitive-linguistic view of compositeness, we have to focus on
the symbolic network relationships that are associated with the formal pole of
idiomatic activation-sets. Compositeness has been defined as the constructional
complexity of an idiomatic construction (see Table 1.1 in Chapter 1). Idioms
such as grasp the nettle were defined as complex constructions that contain
at least two lexical units (grasp and nettle). Following Langacker’s definition
of grammatical composition, the compositeness of grasp the nettle is the entrenched result of the syntagmatic integration of two (or more) symbolic units
(see Figure 3.6 above). Accordingly, the integration of the verbal schema [sbj
grasp obj] with the [thing]-noun [nettle] results in the entrenched composite
structure [sbj grasp the nettle]. This idiomatic construction constitutes a separate node in the cognitive grammar because it has a specific meaning: it evokes
both the corresponding literal meaning of someone touching this stingy plant
and the idiomatic meaning ‘tackle a difficult problem.’
In Section 2.1.1, we saw that the structural complexity of phrase-idioms
posed a serious problem to early generative models because they were described as lexicalised syntactic patterns that preserve their constructional characteristics (e.g. verb inflection, agreement, etc.). In the Chomskyan paradigm,
this “anomaly” is a direct effect of the strictly assumed dichotomy between the
lexicon and syntax: the lexicon is the stock of syntactically basic units, while
syntax provides the rules to combine these units into complex phrases. As we
have further seen (Section 2.1.3), this building-block conception of linguistic knowledge has promoted the overly simplistic characterisation of idioms as
phrasal lexical items that have the semantic structure of long words. In contrast,
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.5 (280-320)
Langacker’s rejection of this modular dichotomy in favour of a syntax-lexicon
continuum does not allow him to strictly separate different types of entrenched
linguistic units. An idiom such as grasp the nettle just occupies a specific area on
the syntax-lexicon continuum – it is a subtype of symbolic unit that is lexicallyrich, constructionally-complex and semantically-idiosyncratic. Therefore, the
idiom constitutes an independent node in the cognitive grammar-network.
But although idiomatic, grasp the nettle is not isolated in the grammar, but
linked to other constructional schemas, which it instantiates (see also Croft
2001: 25–26). These schemas, which are themselves intertwined with other
nodes in the network, provide for its structural description – they categorise
it – in Langacker’s terms:
[CLAUSE]
[SBJ VERB] = (intransitive clause)
[SBJ VERB OBJ] = (transitive clause)
[SBJ grasp OBJ]
[SBJ swallow OBJ]
[SBJ x OBJ]
[SBJ grasp the nettle]
Figure 4.1 The composite structure of grasp the nettle
Since, by definition, we can still perceive the phrasal complexity of idioms
rather than storing them as holophrastic entities, it is counterintuitive to assume that these linguistic units function like long words on the mental plane.
With regard to the recognition of the constructional complexity of lexicalised
expressions, Langacker (1987: 462) therefore states:
The familiarity of a complex expression does not blind us to its componentiality and render us unable to perceive the contribution of individual components. If this were so, the notion of a complex lexical item would be a
contradiction in terms: the unit status characteristic of lexical items would entail their immediate and automatic loss of analyzability, removing any grounds
for considering them to be complex; all fixed expressions would therefore
constitute single morphemes, regardless of size or any resemblance to other
units. In fact, though, a fixed expression appears capable of retaining some
measure of analyzability almost indefinitely. At any one time, a language has
many thousands of complex symbolic units whose values are enriched by the
recognition of their components. We need not assume that the component

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.6 (320-388)
structures are accessed on every occasion when the composite structure is employed, or that when accessed they are necessarily activated at the same level of
intensity as they are in a novel expression. However, only when the composite
structure loses altogether its capacity to elicit the activation of its components
can it be regarded as fully opaque and unanalyzable.
Constructional analysability thus plays the most crucial role in attributing
internal complexity to a composite structure. An entrenched construction remains analysable if the contribution of its immanent component substructures
can still be recognised by a speaker or hearer. This is certainly true for grasp the
nettle; its compositeness is the reflex of the cogniser’s ability to activate the idiomatic constituents, grasp and nettle and the constructional schema [sbj verb
obj], which underlies their integration. In other words, when this idiomatic
construction-node is activated in the cognitive grammar, its categorising symbolic substructures are coactivated with it. As a result, we perceive and process
this linguistic unit as a complex construction. This cognitive-linguistic reinterpretation of compositeness makes it possible to explain why idioms maintain their phrasal characteristics, although they function as automatised, i.e.
entrenched, linguistic routines.
This cognitive-linguistic account is fully in line with the psycholinguistic configuration hypothesis. As we have seen, the configuration hypothesis
rejects the idioms-as-long-words view by emphasising the importance of the
constituent meanings for the activation of the idiomatic meaning (see Section 2.1.2). The psycholinguistic prominence of the constituent words speaks
for constructional analysis in idiom processing. The configuration hypothesis thus provides compelling psycholinguistic evidence for the fact that the
status of idioms as entrenched linguistic units, does not prevent them from
unfolding their structural complexity in actual processing. In contrast, constructional complexity underlies their recognition and processing in actual
usage-events. Interestingly, Cacciari and Tabossi (1988: 680) interpret their notion of idiomatic configuration in terms of distributed representation: “there is
a distributed representation of the idiom in the mental lexicon.” This characterisation is compatible with our description of idioms as complex activation-sets.
Put differently, Cacciari and Tabossi’s term ‘configuration’ can be interpreted as
an entrenched mental network that has the status of a complex linguistic-unit.
The whole idiomatic activation-set is more than the sum of its parts (grasp the
nettle is an independent unit because of its specific meaning), but for the composite configuration to be cognitively unfolded, its immanent parts – the keys –
must first be recognised and activated.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.7 (388-419)
. Institutionalisation and lexicalisation – the cognitive entrenchment
of an idiom
The sociolinguistic concept of institutionalisation describes the process “by
which a standard construction becomes more current in a speech community
and is reproduced as a memorized sequence because of its specific meaning
or pragmatic function” (Barkema 1996: 135). Langacker’s dynamic and usagebased conception of the competence-performance distinction makes it possible
to reinterpret institutionalisation as an effect of cognitive entrenchment: a symbolic event can be regarded as belonging to the grammar of a language if it
does not only occur transitorily, but becomes automatised through a gradual
process of cognitive routinisation. However since institutionalisation is a sociolinguistic rather than a cognitive parameter, cognitive entrenchment is not a
sufficient condition. Rather than being confined to the cognitive grammar of
an individual, institutionalisation is the result of collective cognitive entrenchment spreading over the mental lexicons of the members of a given speech
community in a process of conventionalisation. Thus, institutionalisation involves the simultaneous interplay of the cognitive process of routinisation with
the socio-pragmatic process of the language users’ interactive accommodation
of their cognitive grammars (for a detailed model of this process see Croft 2000:
Chapter 4).25
Institutionalisation must be further distinguished from lexicalisation,
which Barkema (1996: 135) defines as follows (cf. also Bauer 1983; Moon
1998: Section 2.3):
[The] process of lexicalisation [...] is linguistic in nature: the application of
general grammar rules to the expression becomes more and more subject to
restrictions, with as a result limited flexibility as well as limited collocability
and/or compositionality.
Thus, lexicalisation directly correlates with the grammatical behaviour of a
construction becoming more and more idiosyncratic (idiom principle!). It describes the further qualitative stage in the process of cognitive entrenchment,
the stage at which an entrenched construction loses its regularity and, when being (re)produced becomes more and more subject to direct stipulation, similar
to the direct retrieval of a word.
Lexicalised constructions must be institutionalised by definition, because
their production is no longer guided by general principles of linguistic composition. As a result, their lexicogrammatical behaviour becomes restricted
(frozenness!). In cognitive-linguistic terms these limitations can be readily ex-

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.8 (419-465)
 Idiomatic Creativity
plained as an effect of the reduced scaffolding function of the component
substructures in a composite unit. Due to lexicalisation, these structures do
not necessarily have to be activated for the linguistic unit to work as a standard
in the coding process. As a result, the components lose their communicative
value and are pruned from the activation-set.
(1) this is a red herring
(2) this is a red rose
*this is a very red herring
this is a very red rose
For instance, red herring in (1) is lexicalised, while red rose in (2) is not. With
the opaque idiom, the adjective red does not make a salient contribution to the
actual meaning ‘a deception.’ Therefore, it does not serve a semantic scaffolding
function in the communication process. This, obviously, is not true for red in
(2). Here, the adjective qualifies the referent of rose. For the two constructions
to be communicatively functional, the meaning red must therefore be strongly
activated with (2), whereas it is only weakly relevant for the comprehension
of the lexicalised idiom. These differences in activation strength have an effect
on the variability of the two constructions. While it is not possible to intensify
the adjective in red herring, this is fully grammatical for red rose. The reduced
scaffolding function, i.e. the reduced communicative functionality of the components of the lexicalised idiom, therefore constrains its lexicogrammatical
behaviour. Thus, lexicalisation correlates with the weakening or pruning of
the mental activation-set that is potentially triggered by a linguistic unit. Red
herring is still perceived as a phrase. Therefore, the structural activation-set
that underlies our recognition of its compositeness is still salient. However, the
semantic structure of this idiom is fairly opaque. Accordingly, the semantic
network that we evoke with this expression is only poorly structured (cf. also
Lamb 1999: 163–170). This leads us to the next question. How are the semantic
activation-sets that we associate with different idioms shaped, and how is this
shape related to their institutionalisation and lexicalisation?
With regard to their semantic structure, idioms have been defined as patterns of figuration. This view implies that, in their diachronic development, idioms first emerged as creative, i.e. non-conventional metaphors, metonymies,
blends or emblems and then became entrenched as linguistic units through
institutionalisation and lexicalisation. How then can this process of idiomatic
entrenchment be understood? To answer this question, I would like to proceed by analogy. I will first model the lexicalisation of the mouse-metaphor, i.e.
the use of mouse to denote to a computer pointing-device (also cf. Langacker
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.9 (465-549)
Chapter 4. Idiom representation – a cognitive-linguistic model 
2000a: 17–18). On the basis of this one-word metaphor, we can then establish
an analogical cognitive-linguistic account of idiomatic entrenchment.
.. Lexicalising the mouse1 → mouse2 -metaphor
The original process of semantic extension that underlay this metaphor was a
highly creative act; it involved a complex comparison performed by the speaker
to derive the similarities that motivate the semantic association between the
two differing conceptualisations, i.e. the mouse-schema, one the one hand,
and the actual computer device, on the other. Therefore, the following conceptual and phonolocial structures must have been strongly activated for this
innovative coding event:
Table 4.1 The mouse-metaphor
the symbolic activation-set for (mouse1 → mouse2 )
The vehicle/standard –
the symbolic unit
[mouse/mouse] = mouse1
a) semantic pole (predication)
the designated conceptual substructure:
the profile [mouse]
the conceptual base for this profile: the shape-domain
b) phonological pole
the phonological pole associated with this predication:
[mouse]
The topic/target
the target conceptualisation: a rich representation of the
concrete computer pointing-device.
The link
the schematic link capturing the similarities shared between
the vehicle and the target.
The symbolisation of the
novel metaphorical transfer
the newly computed symbolisation linking the phonological structure ([mouse]) with the new conceptualisation
(computer pointing-device)
Having become conventionalised, this complex categorisation event did not
remain transitory, but was automatised by a great number of speakers in terms
of the lexical unit: mouse2 . This lexical unit can be used to refer to any concrete
computer-mouse.
The process of lexicalising mouse2 is accompanied by a reduction of the
activation-set triggered by the original metaphor. The asymptotic state of this
reduction process is reached once the activation network is pruned to such an
extent that the original motivation for the metaphorical transfer can no longer
be perceived. This process can be called homonymisation. Homonymisation
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.10 (549-600)
 Idiomatic Creativity
implies that the transparent association between the vehicle and the topic turns
increasingly opaque: the literal (mouse1 ) and the figurative meaning (mouse2 )
become homonymous. Let us imagine a fictive future state of English when
the mouse-metaphor will be fully homonymous. In this case only the phonological unit [mouse], the semantic unit [computer device] and the symbolic
connection between the two would remain as entrenched substructures of the
original activation-set; the original meaning [mouse (as rodent)] and the motivating link (the similar shape), however, would no longer be evoked by the
language users. In other words, mouse2 would exist alongside mouse1 but the
two would have the status of semantically unrelated words merely linked by
the same phonological structure.
As this fictive scenario illustrates, the development of a creative metaphor
to a pattern of homonymy is primarily accompanied by the loss of the ground –
the recognition of the linking similarities. The institutionalisation of the
metaphor is the primary reason for this loss: once the word mouse is conventionally used to denote a computer device, any language user can learn the word
mouse2 without having to reproduce the metaphorical association from which
this secondary symbolisation was originally derived. Therefore, the metaphorical ground is no longer needed as a scaffolding to motivate the usage: losing its
categorisational importance, the link is reduced in cognitive salience and eventually pruned from the activation-set of the unit mouse2 . The same process can
be assumed for the opaque idiom red herring.
However, from a synchronic perspective, mouse1 and mouse2 cannot yet be
regarded as fully homonymous;26 rather, many speakers of English still seem to
be able to recognise the original metaphorical motivation underlying the use of
mouse2 .27 In cognitive-linguistic terms, they can still activate a link between the
vehicle and the target. As a result, the rich semantic activation-set that underlay the original mouse-metaphor cannot yet be fully pruned in the cognitive
grammars of these speakers. Certainly, by comparison with the original creative categorisation event, the motivating conceptual substructures have lost in
activation probability and strength. This, however, is not to say that they cannot
be activated. Using Goatly’s words these substructures are sleeping rather than
dead (Goatly 1997: 32), i.e. they can be optionally activated. As a result, mouse1
and mouse2 can still be perceived as metaphorically related.
Assuming the existence of ‘sleeping’ metaphorical substructures, it becomes possible to circumvent a shortcoming inherent in the pragmatic theory
of metaphor (cf. Searle 1993). According to the pragmatic view, conventionalised metaphors must be attributed the same status as literal language, because, due to their lexicalisation, these expressions can be directly processed
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.11 (600-644)
Chapter 4. Idiom representation – a cognitive-linguistic model 
by activating their conventionalised semantic values. Accordingly, words such
as mouse2 are regarded as dead metaphors because they have become so worn
out that speakers no longer regard them as metaphors. Live metaphors, on the
other hand, belong to the sphere of pragmatics. They are recognised as deviant uses of language and their meaning is context-dependent. Consequently,
the pragmatic theory correlates the institutionalisation of a metaphor with a
clear-cut change of its semantic quality: from metaphorical to literal. This assumption, however, obscures the fact that entrenched metaphors like mouse2
may still be vivid.
The dead metaphor theory – implied in the equation of conventionality
with literalisation – has been repeatedly criticised by cognitive linguists:
The mistake derives from a basic confusion: it assumes that those things in
our cognition that are most alive and most active are those that are conscious.
On the contrary, those that are most alive and most deeply entrenched, efficient, and powerful are those that are so automatic as to be unconscious and
effortless.
(Lakoff & Turner 1989: 129)
The distinction between dead and live metaphors ignores the intermediate
alternative that patterns of figuration can become entrenched to work as
metaphorical cognitive routines. In contrast, a cognitive-linguistic view of lexicalisation, which accounts for the gradual reduction of metaphorical strength,
does not allow for such an all-or-nothing position. Instead, institutionalisation
and lexicalisation lead to a gradual rather than clear-cut loss of the conceptual
substructures that can be co-activated with an entrenched metaphor.
.. Idiomatic activation-sets – a preliminary overview
Proceeding from the mouse1 → mouse2 example, we are now in a position to
transfer the notions of institutionalisation and lexicalisation to our cognitivelinguistic analysis of idiom structure. More precisely, we have to provide a
model that can explain what conceptual and symbolic substructures of a given
idiomatic configuration (e.g., grasp the nettle) are likely to be entrenched in the
cognitive grammars of native speakers. Moreover, we will have to show how
the mental representation of these substructures affects the use of the idiom
in actual discourse. We have argued that idioms do not constitute homogeneous linguistic categories. To account for the potential range of difference in
idiomatic activation-sets, it is thus necessary to provide an explanatory model
that is open enough to account for this structural heterogeneity. I will therefore
start by designing the maximum activation potential for an idiomatic construc-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.12 (644-738)
 Idiomatic Creativity
tion and then relate this potential to an opposite, minimal asymptotic state –
the minimal state for a construction to count as an idiom still.
Idiomatic constructions differ from one-word metaphors in terms of their
composite structure. This increases the complexity of idiomatic activation-sets:
Table 4.2 Substructures of the idiomatic activation-set for grasp the nettle
idiomatic activation-set (e.g., grasp the nettle)
(for simplicity’s sake, only two constituents are specified)
The vehicle/
standard
–
the literal
meaning
a) the semantic pole (predication)
i) the composite literal meaning, the literal-scene profiled
by the idiomatic constituents: [grasp the nettle]
ii) the conceptual base/domain in which the literal-scene is embedded:
[conceptual scenario of tackling the plant]
iii) the profiled conceptual substructure and base for the first
constituent (grasp): [agent grasp patient]
iv) the profiled conceptual substructure and base of the second
constituent (nettle): [nettle]
v) the constructional schema underlying the integration of the two
constituents: [sbj verb obj]
b) phonological pole
i) the phonological pole associated with the composite literal
meaning: (grasp the nettle)
ii) the phonological poles associated with the first constituent: (grasp)
iii) the phonological pole associated with the second constituent:
(nettle)
The lexicalised
topic/target
–
the idiomatic
meaning
i) the composite idiomatic meaning, the profiled target
conceptualisation: [tackle the problem]
ii) the conceptual domain/scenario in which the idiomatic meaning is
embedded: [conceptual scenario of dealing with problem]
iii) salient conceptual substructure or components of the idiomatic
meaning: [tackle], [problem]
The motivating
links
the correspondences/similarities shared between the literal and the
idiomatic meaning
The idiomatic
configuration as
a symbolic unit
the lexical association linking the phonological structure with the
idiomatic meaning
To be able work as a linguistic unit, the minimal substructures that must be
cognitively entrenched with a given idiom are its composite phonological pole
and the idiomatic meaning. However, if an idiom featured these substructures only, it would have the mental status of a holophrastic form-meaning
association, i.e. it would function like a long word.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.13 (738-787)
Chapter 4. Idiom representation – a cognitive-linguistic model 
In the previous subchapter on compositeness, we have emphasised that idioms are recognised and processed as composite constructions. In accord with
the configuration hypothesis, it must therefore be assumed that the lexical constituents and the underlying constructional schema are co-activated as salient
components (see Figure 4.1 above). In association with the idiomatic meaning, these substructures define the minimum, asymptotic cognitive-structure
for idioms. In what follows, we can therefore ignore the phonological pole.
Turning to the maximum state, it cannot be assumed that for all idioms the
idiomatic meaning is “semantically simplex” (Cruse 1986: 37). Thus, the meaning of grasp the nettle is complex: ‘tackle the problem.’ Given the existence of
complex idiomatic meanings, it is reasonable to assume that specific components of this predication can be activated and manipulated separately as well.
For instance, the idiomatic meaning of grasp the nettle can be segmented into
the components [tackle] and [problem]. This renders the idiomatic meaning
complex and analysable.
Referring to the metaphorical motivation, the status of the linking ground
is open. Since idioms are defined as institutionalised symbolic units, the conventionally reproduced association of the literal constituents with the figurative
idiomatic-meaning can be directly acquired by any speaker who learns a given
idiomatic expression. In this sense, learning an idiom is similar to learning the
word mouse2 . The original process of semantic extension underlying this association does not have to (and often cannot) be re-established. Nevertheless, the
primary claim of this study is that for different idioms the relationship between
their idiomatic and their literal meanings can have different qualities. Idioms
reflect different semantic configurations that range between the two poles of
a) fully conventionalised but transparent patterns of semantic extension (grasp
the nettle), and b) homonymous patterns for which the links between the literal and the idiomatic meaning have become fully opaque (red herring) (cf.
also Cutler 1982). Cognitively, these differences are reflected in the quality and
richness of the activation-set that can be evoked with a given, specific idiom.
To anatomise the specific quality of an idiomatic activation-set, we must
scrutinise the specific quality of the mental substructures that shape it and constitute the cognitive network that can be potentially unfolded when the idiom
is used. Concentrating on the semantic structure, we must therefore focus on
the following global analytical dimensions:
1. Internal structuring: the connections between the substructures of the idiomatic activation-set; e.g. links between the constituents and components
of the idiomatic meaning: grasp → tackle.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.14 (787-844)
 Idiomatic Creativity
2. Transparency: the ways in which the motivating links between these conceptual substructures can be recognised by speakers.
3. Conceptual backing: the conceptual patterns that support the recognition
of a motivated and transparent idiom structure.
(1) refers to the distinction between different types of semantic structuring: compositionality, motivation and analysability; (2) captures the alternative cognitive routes along which links between idiomatic substructures can
be established; finally, (3) characterises the patterns of semantic extension
(metaphor, metonymy, blending, emblems) that make transparency possible.
To establish a workable model of semantic extension in idioms, all of these
analytical dimensions must be dissected.
. A cognitive-linguistic anatomy of the internal semantic structure
of idioms
.. Idioms as complex scenes
In Section 3.4.1, the semantic pole of a symbolic unit was defined as its predication. The predication constitutes a conceptual base with a profile that designates a specific substructure in this domain. But how can this definition be applied to idioms? Since idioms are linguistic standards that are based on patterns
of figuration, they do not offer direct access to the target-conceptualisation that
they code in a given usage-event. Rather, the conceptual base that underlies
their predication is complex: the semantic pole of an idiom constitutes a complex scene (Langacker 1987: 141). A complex scene is a cognitive representation
that does not reduce to a single, consistent conceptual configuration:
Cognitive grammar does not assume that an expression’s meaning necessarily
reduces to a single, consistent configuration. It claims that that our ability to
construe one conceived situation in relation to another, regardless of their degree of compatibility, is fundamental to linguistic semantics. Hence the meaning of a complex expression is defined as including not only its composite semantic structure but also the compositional path through which that structure
is assembled [...]. [...] Recognizing the metaphorical nature of an expression
like take the bull by the horns requires the co-construal of two very different
conceptions, its literal sense and its figurative value. (Langacker 1991a: 133)
The creation of complex scenes is involved in all patterns of semantic extension described in Section 3.3. For these complex forms of categorisation, the
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.15 (844-911)
Chapter 4. Idiom representation – a cognitive-linguistic model 
standard and the target structures cannot be absorbed in a single, consistent
conceptualisation. Rather, they result in a bipartite conceptualisation including
a literal and a figurative conceptualisation (Langacker 1987: 92). Although the
literal and the figurative conceptualisations are not consistent, nothing prevents their categorisational integration on the basis of a linking schema or
ground. More specifically, the literal scene works as the mental background – a
conceptual standard – against which the target conceptualisation is construed:
Internally many predications have the character of a complex scene, in which
one conception is construed against the background of another from which it
diverges in some respect [...]. Their relationship manifests the inherent asymmetry of a categorizing relationship: the background conception serves as
the standard of comparison, while the foregrounded target of categorization
is the active structure; i.e. the one employed for the purposes of coding or
higher-level composition.
(Langacker 1991a: 133)
Thus, a clear-cut line must be drawn between the background and the foreground of a complex scene. In addition, it is necessary to specify how the
complex scene is evoked to establish reference to an actual target conceptualisation – in Langacker’s terms: “how the complex scene is employed for the
purposes of coding or higher level composition.” Consider, for instance, the
idiom take the bull by the horns. The literal scene conveyed by this construction
refers to the action of grabbing a dangerous animal. However, its conventional
idiomatic meaning – tackle a problem with determination – evokes a figurative
scene. The literal scene establishes the conceptual background against which
this figurative scene is compared.
(1) Of course, being the determined woman she was, Nora decided to take the
bull by the horns and organise things for herself. (CEY: 135).
When the idiom is employed in actual discourse, such as in (1), the literal scene
is not directly mapped onto the target conceptualisation to establish literal reference. Rather, it is activated as the metaphorical background of a complex
scene. (1) deals with Nora’s determination to overcome some personal problems. Therefore, the idiomatic, rather than the literal meaning, is integrated
into the overall semantic structure of the statement. In relevance theoretical
terms, the contextually embedded idiomatic-meaning constitutes the intended
and conventional strong implication that is communicated by (1) (see Sperber
& Wilson 1986: 236). Thus, the idiomatic meaning works as the foregrounded
figurative-scene that is mapped onto its referential target – Nora’s determined
tackling of her problems – for the purpose of coding this idea. This is what Lan-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.16 (911-951)
 Idiomatic Creativity
gacker means when saying that “the foregrounded target is the active structure
employed for the purposes of coding.” However, since the idiomatic meaning is evoked relative to the background of the literal scene, this conceptual
background is not deactivated or absent in this usage-event. In contrast, it still
works as a metaphorical standard of comparison in relation to which the idiomatic meaning is itself concretised. In other words, the literal scene works
as a conceptual scaffolding against which the idiomatic meaning is conceived.
Thus, although this conceptual background does not work as the primary semantic structure, the literal meaning organises the target-conceptualisation
indirectly by providing a more concrete cognitive model for tackling problems.
Moreover, this literal background provides conceptual entailments that make
it possible to derive additional weak implicatures for (1) (see Sperber & Wilson
1986: 199). For instance, we can infer that the problem was uncomfortable,
difficult to deal with and that Nora had to make some very significant effort
to handle it. In other words, the literal-scene works as a conceptual resource
to enrich, concretise and flesh out the idiomatic meaning although it does not
constitute the actual referential meaning of the idiom. In the same way, we do
not map the literal meaning of mouse when we use this word to refer to the
computer pointing-device. Nevertheless, the literal meaning can still work as a
standard of comparison, a mental model, against which this target is assessed.
It is important to emphasise that psycholinguistic evidence supports this
interpretation of idioms as complex scenes. The results of the configuration hypothesis, the simultaneous processing hypothesis, Titone and Connine’s hybrid
processing model and other psycholinguistic studies reveal that the processing
of the literal meaning is not suppressed after the activation of the idiomatic
meaning (see Chapter 2). Once the idiomatic meaning can be triggered by
the key-constituents, it is brought into the semantic foreground. This, however, does not lead to the deactivation of the literal meaning and its conceptual
impact. Rather, the literal meaning remains cognitively active and accessible
to work as a semantic background, a standard of comparison, for the foregrounded idiomatic meaning (also cf. Glucksberg 1993: 20). Against the pragmatic theory of metaphor, the impossibility to suppress the literal scene thus
indicates that the semantic complexity of an idiom is a salient mental phenomenon rather than a relict of its diachronic development from a creative to
a dead metaphor. As will be argued below, idioms grossly derive their communicative and cognitive functionality from their status as complex scenes.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.17 (951-1011)
Chapter 4. Idiom representation – a cognitive-linguistic model 
.. Geeraerts’s prismatic model of idiom semantics
To model the semantic anatomy of idioms as complex scenes, I will introduce Geeraerts’s (1995) prismatic model of idiom semantics and adapt it to
the present cognitive-linguistic framework. To account for the semantic complexity of idioms, the model is first given a maximal architecture; it includes
all descriptive dimensions that are necessary to analyse the internal semantic
relationships of any idiomatic expression. Given its all inclusive nature, this architecture is too general to describe the internal semantic structure of specific
idiom exemplars precisely. Thus, when dissecting the semantic anatomy of individual idiomatic constructions, this general architecture must be adapted –
the network must be pruned – to provide a precise structural description
of their predications. On the basis of the model, then, typical conceptual
activation-sets that underlie different idioms can be isolated. These patterns
reflect specific subtypes of figuration in idioms. Moreover, they characterise
the diverse stages of lexicalisation that different idioms occupy.
Geeraerts (1995: 61) introduced his prismatic model (see Figure 4.2) to describe the diversity of semantic structuring reflected by different idioms.28 The
model can be discussed in cognitive-linguistic terms if the semantic substructures labelled by Roman type (A, B and C) are interpreted as categorisation
standards, while the ones in Greek type (α, β and γ) work as their targets
in a complex pattern of figuration. Moreover, we have to apply the semantic
processes of semantic composition (including integration) and semantic extension. Both processes can be scrutinised from bottom-up and top-down perspectives (the arrows indicate how these relationships are represented in the
figure):
1. Semantic composition = syntagmatic dimension
a. bottom-up: compositionality
b. top-down: isomorphism/analysability
←
→
2. Semantic extension = paradigmatic dimension
a. bottom-up: projection/grouping
b. top-down: motivation
-------→
←-------
Thus both the paradigmatic and the syntagmatic dimension involve two semantic relationships that pattern the internal semantic anatomy of an idiom
like rock the boat (cf. Geeraerts 1995: 59–61). In accordance with the traditional view (see Section 2.1.1), compositionality refers to the bottom-up view
of semantic composition. In contrast, the notions of motivation and isomor-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.18 (1011-1033)
 Idiomatic Creativity
DOMAINg
PARADIGMATIC DIMENSION
a
g
Ç
b
DOMAINC
A
C
Ç
SYNTAGMATIC
DIMENSION
B
a
SYNTAGMATIC
DIMENSION
A
b
PARADIGMATIC DIMENSION
B
Figure 4.2 Geeraerts’s prismatic model of idiom semantics (based on Geeraerts
1995: 60)
phism/analysability emerge, if the interaction between these two dimensions of
semantic structuring are scrutinised top-down. In what follows, these dimensions are analysed in detail.
The paradigmatic dimension includes the relationship between the composite literal meaning of the idiomatic expression as a whole (A ∩ B = C,
C = rock the boat) and its composite idiomatic meaning (α ∩ β = γ, γ =
spoil/trouble a comfortable situation). Further, the paradigmatic dimension captures the relationship between the literal meanings of the individual
idiom constituents (A = rock) and (B = boat), and the interpretation that
these parts receive within the idiomatic meaning of the expression as a whole
(α = spoil and β = comfortable constellation). Thus, the three paradigmatic connections in the model capture the dimension of semantic extension;
they focus on the mapping relationships between the literal categorisation standards (A, B and C) and their figurative targets (α, β and γ). The standards are
projected onto the targets to group them by extension from their conventional
semantic values. They reflect what I have called a pattern of figuration. Thus,
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.19 (1033-1094)
the literal scene rock the boat is projected onto the target meaning spoil a
satisfactory constellation to categorise it metaphorically: the literal scene
models the more abstract idiomatic meaning in concrete terms.
The syntagmatic dimension focuses on the relationship between the constituents (A and B; α and β) and the interpretation that these components
receive in the composite idiomatic and literal meanings (C and γ). In cognitivelinguistic terms the syntagmatic dimension captures the processes of semantic composition and integration. Concentrating on the literal level, the constituents A and B are integrated according to the valence relations immanent
in their conceptual substructures. Thus, the boat can be integrated with the
verbal schema [sbj rock obj] because it elaborates its obj-slot. This integration
results in the composite structure [sbj rock the boat] that evokes the composite
literal meaning (agent rocks the boat).
Along these dimensions, compositionality, isomorphism and motivation
work as different parameters that shape the transparency of an idiom. Thereby,
isomorphism and motivation correspond to the notions of analysability and
motivation as introduced in Sections 2.2 and 2.3, whereas compositionality is
in agreement with the bottom-up view of semantic derivation. So far, we have
basically introduced and employed these notions as heuristic concepts. This
is in accordance with their traditional usage in the phraseological literature,
where they are not usually explained relative to a specific theoretical framework
(for a good overviews see Burger 1998a: Chapter 3; Burger et al. 1982: 28). Since
cognitive-linguistics postulates direct correspondences between linguistic and
cognitive structuring, we are forced to explain the existence of these semantic
parameters in accordance with the cognitive foundations established in Chapter 3. In other words, a cognitive-linguistic account of the semantic structure of
idioms should be able to answer the question: What are the cognitive processes
and structures that lead to the transparency of idioms?
.. Idiomatic compositionality revisited
Compositionality is the result of adding the semantic values of the constituents
of a construction bottom-up. In the prism this dimension primarily applies
both to the syntagmatic relationships on the literal level and to the bottom-up
derivation of the idiomatic meaning through this integration.
By definition, idioms are non-compositional if analysed from a bottomup perspective (see introduction and Section 2.1). I will therefore employ the
concept of idiomatic compositionality only to describe the direct literal contribution of a constituent to the idiomatic meaning. For instance, glance in

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.20 (1094-1134)
shoot a glance at sb. can be interpreted literally on the level of the idiomatic
meaning. Therefore, this idiom qualifies as a partially compositional idiom. In
the phraseological literature alternative terms have been used to denote corresponding expressions: semi-idioms (Weinreich 1969), restricted collocations
(Cowie et al. 1993: xiii) or phraseologische Verbindungen (Burger et al. 1982: 31–
32; Sabban 1998: 39). By definition, then, partially compositional idioms are
partially analysable and partially motivated. (Note that my terminology deviates slightly from the compositional view of idioms outlined in Section
2.2. As will be outlined below, the latter conception corresponds to my term
figurative-literal isomorphism, which overlaps with the psycholinguistic notion
of decomposition).
When focussing on the literal level only, the literal meaning is generally
compositional if the bottom-up integration of the constituents triggers a meaningful mental representation – a literal scene. This said, it is not possible for all
idioms to attribute a meaningful interpretation to the composite literal meaning. I will call such idioms literally non-compositional idioms. For instance,
the presence of unique constituents that do not appear outside the context
of the idiom can cause literal non-compositionality. Thus, for most speakers the words gaff in blow the gaff or kibosh in put the kibosh on sth. are not
embedded in rich conceptual networks, but only appear in the constructional
context of these idioms. As a consequence, these constituents cannot evoke rich
conceptual content and do not contribute to the mental construction of a literal scene. In analogy to morphology, Moon (1998: 21) appropriately calls such
constructions cranberry collocations.
Moreover, the computation of a rich literal scene is thwarted by idioms
which feature constructional idiosyncrasy. Such expressions can reflect asyntactic patterns (see Cruse 1986: 38) or unconventional internal valency relations
that do not correspond to the general conventions of the language. In the
phraseological literature, these expressions are also called ill-formed or extragrammatical (Moon 1998: 21; Fillmore et al. 1988: 505). For example, idioms
such as trip the light fantastic (dance), shoot the breeze (talk informally, friendly)
or come a cropper (fall, fail) reflect constructional idiosyncrasy. That is, the integration of the lexical units does not proceed in accordance with any constructional schema that underlies other conventional grammatical constructions.
Thus, in come a cropper the non-transitive verb come is used transitively and
combined with the cranberry morph cropper. In shoot the breeze, shoot is ungrammatically collocated with an inanimate noun that lies outside the shooting-frame. Finally, the NP the light fantastic in trip the light fantastic does not
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.21 (1134-1198)
have a head-noun. Since it is impossible to evoke a sensible literal scene for
these literally non-compositional expressions, they are highly opaque.29
Following Langacker’s framework, the dimension of schematicity can be
added as a further parameter to influence the transparency of the literal meaning. For instance, in the idiom put your foot in it, the literal meaning of the
constituent it is semantically empty. (The etymological explanation given by
the Longman Dictionary of Idioms suggests that it originally referred to animal
excrements). Therefore, the literal scene related to the action of putting-yourfoot-in-it becomes highly schematic as well. As a result, one is unable to relate
this action to a concrete conceptual background; this makes the expression
fairly non-transparent.
In a nutshell, compositionality is an important parameter for the semantic
structuring of idioms. While idioms are derivationally non-compositional by
definition, some idiomatic constructions can reflect partial compositionality.
Moreover, a considerable number of idioms are literally non-compositional:
their formal constituent structure does not evoke a rich literal scene.
.. Motivation revisited
The concept of motivation applies to the top-down perspective on the paradigmatic dimension in the prism. This involves two alternative motivation dimensions: global motivation and constituental motivation. An idiom reflects global
motivation if the semantic extension from the literal to the figurative scene is
still transparent. Thus, for rock the boat the relationship between the literal
scene (rock the boat) and its figurative interpretation (spoil a comfortable situation) seems well-motivated: when someone rocks a boat, the boat
is likely to be overturned; this, obviously, spoils the more satisfactory previous
state or situation.
In addition, the concept of motivation can also be applied to external,
i.e. not idiom-bound, figurative senses of the individual constituents. Constituental motivation appears with idioms whose constituents possess lexicalised
figurative senses that also appear outside the phrasal context of the idiom. In the
phraseological literature this phenomenon is called “semantische Autonomie”
(semantic autonomy), (cf. Burger et al. 1982: 28; Dobrovol’skij 1997: 91). For
instance, swallow has the lexicalised figurative sense ‘accept patiently.’ This
meaning is present in idioms like swallow the bitter pill (accept an unpleasant fact) or strain at a gnat and swallow a camel (accept a major wrong thing
while being concerned with a minor one).

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.22 (1198-1242)
Since motivation is defined as a top-down process, it does not imply that
the idiomatic targets can be predicted on the basis of the literal standards. However, an idiom is motivated if it is possible to make sense of these projections.
That is, on the basis of their symbolic and conceptual knowledge, speakers
are able to explain why the literal scene has been chosen to encode the idiomatic meaning. In accordance with the cognitive theory of metaphor and
metonymy, the recognition of this motivated relationship is triggered by the
cognitive ability to re-establish the linking conceptual similarities on which
the semantic transfer from the literal to the extended figurative meaning is
based: the language user can construct conceptual or image-schematic correspondences between the two meanings. For the global motivation of rock the
boat, this process can be scrutinised as follows:
1. The cogniser must first construct a conceptual representation of the literal
scene, e.g. the scene of rocking a boat. Consequently, global motivation is
dependent on the presence of a well-formed and meaningful literal meaning. This is clearly the case for rock the boat.
2. This literal scene must be related to an ICM – a conceptual domain (frame
or script) in which it can be embedded. Thus, the literal scene rock the
boat can be related to a more extensive boat trip-script.
3. The speaker/hearer must know the idiomatic meaning, i.e. he or she must
be able to activate a conceptual representation that is related to the abstract
idiomatic meaning. For rock the boat the idiomatic meaning corresponds
to a troublemaking scene which is embedded in a more general scenario of
activity.
4. On the basis of these representations, the cogniser must extract conceptual or image-schematic similarities or correspondences between the literal
scene and the idiomatic scene by relying on the knowledge evoked by
the respective conceptual domains. These associative links constitute the
motivation for the given idiom. For rock the boat these correspondences
are straightforward. The literal scene rock the boat describes a specific
subtype of troublemaking. Thus, the force-dynamic causality agent troubles situation is already immanent in the more concrete scenario agent
rocks boat. Therefore, the scene of rocking the boat can be recognised as
a well-motivated model to concretise the idiomatic meaning.
It is important to emphasise that the cognitive motivation of an idiom does
not necessarily have to – although it may – correspond to its original etymological motivation. For some idioms speakers are able to remotivate their
semantic structure although they do not know their true etymological basis
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.23 (1242-1292)
(cf. Burger 1998a: 66). Thus, from a cognitive-linguistic perspective, motivation must always be regarded as a synchronic process that is dependent on the
conceptual knowledge of contemporary language users; therefore it cannot be
equated with the reconstruction of an idiom’s diachronic development.
.. Isomorphism/analysability – a cognitive-linguistic explanation
Applying to the syntagmatic dimension in the prismatic model, isomorphism
involves the top-down re-analysis of an idiom’s compositional path. Following
Geeraerts (1995: 60), an expression is isomorphic (or analysable) if “a one-toone correspondence between the parts of the semantic value of a compound
meaning as a whole [C or γ] and the constituent parts [A and B or α and β]
can be detected.” In the context of idiomatic analysability, the primary focus
is on a possible correspondence between the formal structure of the construction and its idiomatic meaning (see Section 2.2.1). In other words, isomorphism captures the degree of analysability between the idiomatic meaning (γ)
and the idiomatic constituents (A and B). In this sense, isomorphism is the
result of the recognition of syntagmatic correspondences along the paradigmatic dimension of semantic extension. I will therefore call this phenomenon
figurative-literal isomorphism. Rock the boat, for instance, is isomorphic because
its idiomatic meaning (spoil a comfortable situation) can be devolved on
the constituents rock and boat. Boat corresponds to the comfortable situation,
while rock matches the idea of spoiling this situation. For this to work, however, the idiomatic meaning must itself be analysable, i.e. it must be recognised
as being complex.
As we have seen in Section 2.2 the status of idiomatic analysability has
not remained uncontroversial. While Nunberg et. al.’s notion of idiomatically
combining expression and the psycholinguistic decomposition hypothesis provide evidence that this phenomenon is psychologically real, Weinreich (1969),
Chafe (1968), Čermák (1988, 1994), Nicolas (1995) or Schenk (1995) strictly
deny the existence of semantic isomorphism. In a more idiosyncratic vein,
Mel’čuk (1995, 1998) regards all semantically analysable idiomatic constructions as collocations rather than idioms. The questionable nature of these
theoretical views has already been discussed in Sections 2.1.3 and 2.2.3. More
specific criticism against semantic analysability is raised by Burger; he argues
that isomorphism is not intrinsic to the semantic structure of an idiom, but
that it works as a semantic potential that is dependent on the way the idiomatic
meaning is paraphrased (cf. also Dobrovol’skij 1997: 93; Sabban 1998: 63):

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.24 (1292-1406)
Die Teilbarkeit des Phraseologismus zeigt sich in der Isomorphie zwischen der
formalen und der semantischen Struktur. Die Annahme einer solchen Isomorphie ist allerdings immer auch mit einer gewissen Willkür behaftet, insofern
alles davon abhängt, wie man die Bedeutungserläuterung formuliert.
(Burger 1998a: 69)
Of course, Burger is fully right in claiming that the meaning of an idiom can
be paraphrased in different ways and that alternative meaning descriptions can
render the isomorphic structure of an idiom more or less explicit. This, however, is no convincing proof that analysability as a phenomenon of semantic
structuring does not exist. (It must be emphasised that Burger does not claim
this, but he is sceptical about its value as a theoretical concept for the systematic
description of idiom semantics).
From a cognitive-linguistic perspective, the meaning of an idiom certainly
cannot be equated with a possible paraphrase. Rather, it must be located in
the conceptual structure that underlies this meaning paraphrase. But on the
basis of what cognitive parameters does isomorphism emerge? Although being
restricted to verbal idioms with V + NP structure, Nunberg (1978: 125) offers
a definition of isomorphism that approximates the cognitive-linguistic view
advocated here:
Let us say that verb phrases ‘refer’ to states and activities, and that transitive
verb phrases normally refer to states and activities that are best identified as
‘open relations’ of the form Rxb, where ‘R’ stands for the relation referred to
by the verb, ‘x’ is a variable for the referent of the sentence subject, and ‘b’
stands for the referent of the object NP.[...] Then we will say that an idiomatic
transitive VP is decomposable just in case it is used to refer to a state or activity such that it would normally be believed that an activity could be identified
as an open relation Rxb, such that the NP of the idiom refers to b, and the
verb to R.
If the open relation ‘Rxb’ is interpreted as corresponding to the imageschematic interactions between two participants in a relational predication,
then isomorphism can be characterised as an effect of congruent conceptual
structuring and cognitive construal rather than linguistic circumscription. Put
differently, isomorphism is the cognitive effect of the cogniser’s ability to perceive immanent conceptual or image-schematic correspondences between the
ontologies and structures of the source and target conceptualisations. This
makes it possible to relate elements of the idiomatic-meaning domain (α and β)
to the conceptual substructures designated by the idiom’s lexical constituents
(A and B). For rock the boat this can be illustrated as follows:
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.25 (1406-1422)
boat trip-ICM:
(source domain)
activity/trouble making-ICM
(target domain)
profiled literal-scene [C]
profiled idiomatic-scene [γ]
↓
↓
rock the boat-ontology
(conceptual nodes profiled by [A] and [B])
troublemaking-ontology
(nodes profiled by [α] and [β]):
[sbj] = agent = rocker
[A] = rock
[B] = affected entity = boat
[sbj] = agent: troublemaker
[α] = spoil
[β] = affected entity =
comfortable situation
------------→
------------→
------------→
rock the boat-structure:
(interconnections)
impulsive force-image-schema
troublemaking-structure
(interconnections)
----------→
causal force-image-schema
Figure 4.3 Ontological correspondences underlying the isomorphism of rock the boat
Thus, rock the boat is analysable because the literal and the idiomatic meaning feature parallel conceptual ontologies and image-schematic structures.
More specifically, these correspondences involve ontologising projections that
map the concrete objects and processes profiled by the idiom’s literal meaning
onto their figurative correspondents in the idiomatic-meaning domain: the literal meaning of rock the boat has a force-dynamic, transitive profile [sbj rock
the boat], with the [sbj]-node being an agent and [boat] being the patient.
The same force-dynamic construal can be applied to the idiomatic meaning
[agent spoils situation]. Since these correspondences are immanent in the
conceptual scene that underlies the idiomatic meaning, they can be made ‘visible’ by paraphrasing the idiomatic meaning with a transitive syntactic profile
that is congruent to the literal meaning. This is not possible for an unanalysable
idiom such as kick the bucket. For this idiom, the literal scene consists of a forcedynamic relationship between an agent (the subject) and a patient (the bucket)
as well. However, the idiomatic meaning is not force-dynamic. In contrast, the
process die is conceived as an experiencer’s non-volitional, inherent change of
state; it is an intransitive process. Therefore, this conceptualisation does not
overlap with the ontology and structure of the literal scene. Consequently, the
idiom remains unanalysable.
Obviously, analysability is most straightforward for idioms that feature
constituental motivation; via the lexicalised figurative senses direct correspondences can be established between the literal and the idiomatic meaning. Thus,

JB[v.20020404] Prn:20/02/2006; 14:20

F: HCP1704.tex / p.26 (1422-1490)
swallow the bitter pill is fully analysable because its constituents also mean
‘accept’ and ‘unpleasant fact,’ respectively, outside the context of the idiom.
This form of semantic autonomy must be distinguished from the relative semantic autonomy of the constituents of rock the boat. Neither rock nor boat
feature their figurative interpretations ‘spoil’ and ‘comfortable situation’ outside the confines of the idiomatic construction. Therefore, they only convey phrase-induced figurative senses, alternatively this phenomenon is termed
“phrase-induced polysemy” (Glucksberg 1993: 10), “unique lexeme” (Mel’čuk
1995: 211) or “latentes Semem” (Dobrovol’skij 1997: 98).
As these examples indicate, idiom analysability is directly dependent on
the recognition of motivated metaphorical projections of the concepts designated by the constituents onto elements profiled by the idiomatic meaning. If
it is possible to establish such direct correspondences, the idiom is analysable,
otherwise, the idiom remains unanalysable.
.. Are motivation and isomorphism two cognitively
distinct phenomena?
In the phraseological literature, isomorphism and motivation are generally
regarded as two distinct semantic phenomena (see, e.g., Geeraerts 1995; Nunberg et al. 1994: 496–7; Keil 1997). Deviating from this descriptive tradition,
my cognitive-linguistic account conceives both motivation and isomorphism
as the result of cognitive top-down processes that exploit recognisable conceptual or image-schematic correspondences between the idiom’s literal scene
and the idiomatic meaning. In other words, the two transparency dimensions
seem to be cognitively related. This view is shared by Dobrovol’skij (2000),
who also claims that isomorphism is a side-effect of cognitive motivation and
therefore constitutes a subtype of this phenomenon. Given this position, isomorphism without motivation cannot exist, whereas the opposite is possible.
To substantiate this cognitive-linguistic argument, let us proceed from the following statement by Nunberg et. al. (1994: 497), which they propose to justify
the separation of isomorphism and motivation:
When we hear spill the beans used to mean ‘divulge the information,’ for example, we can assume that spill denotes the relation of divulging and beans
the information that is divulged, even if we cannot say why beans should have
been used in this expression rather than succotash.
While the authors are right in claiming that, due to the constituent beans, it
is impossible to find a fully transparent global motivation for this expression,
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.27 (1490-1515)
clearly their analysis of spill the beans still speaks for the possibility of attributing a schematic semantic motivation to this constituent. As we have seen in
Section 2.3.2, this schematic motivation emerges on the basis of metaphorical modelling: the conceptual metaphors ideas are objects, the mind is
a container map little more than image-schematic structure (containers
and objects) from the literal spill the beans-scenario to the target reveal
the secret. In other words, spill the beans is motivated and its isomorphic
structure is an effect of this motivation. However, the expression is not fully
motivated, because it is not possible to find rich conceptual similarities between
beans and secrets. In this sense, the motivation of spill the beans differs from
the rich motivation of rock the boat. While bean-spilling can only be related
to secret-revelation on the highly schematic level of image-schematic structuring, the idea of spoiling a comfortable situation is conceptually immanent
in our stereotypical understanding of rocking a boat. Accordingly, the association of rock the boat with its idiomatic meaning can be motivated on the
basis of conceptual similarities, whereas spill the beans is only schematically
motivated through image-schematic structure. Consequently, the distinction
between isomorphism and motivation can only be maintained, if motivation
is restrictively equated with full conceptual motivation. If, however, one interprets the term more widely and allows it to include more abstract imageschematic correspondences, then isomorphism can itself be regarded as an
effect of motivation.
While isomorphism cannot exist without some degree of motivation, the
opposite is very common. For illustration, spill the beans and spin one’s wheels
can be compared. While spill the beans is analysable but only schematically motivated, spin one’s wheels is fully motivated but not isomorphic. Spin one’s wheels
(fail to achieve anything satisfactory) can also be motivated on the basis of conceptual metaphors: purposeful activity is a car journey and success is
moving forward. These metaphors model a stagnating person in terms of a
stuck car: a car that is spinning its wheels does not move forward and therefore
does not reach its destination; accordingly, a person who does not progress in
life or at work does not achieve his/her life or career goals. Although these conceptual associations render the idiom well-motivated, spin one’s wheels is not
isomorphic because it is not possible to establish direct, analysable correspondences between the constituents (spin, wheels) and conceptual correspondents
in the idiomatic meaning. Rather, it is the whole process of spinning one’s
wheels that is related to the idiomatic meaning ‘stagnate’ through a causefor-effect metonymy: the wheel spinning-process points to the vehicle’s

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.28 (1515-1571)
 Idiomatic Creativity
inability to move; this metonymic association is then mapped in accordance
with the conceptual metaphors to stand for a person’s stagnation.
In a nutshell, isomorphic idioms establish ontological correspondences between the constituents and conceptual correspondents in the idiomatic meaning. As a result they possess an analysable semantic structure. Unanalysable
idioms can often be motivated. However, their constituents cannot be directly
mapped onto target-domain correspondents. One reason for this is the presence of metonymic shifts. By establishing deferred foci on conceptual relationships, metonymies can prevent direct ontological correspondences between
constituents and possible correspondents in the idiomatic meaning. The constituental structure only provides indirect access to these correspondences via
metonymic contiguity relationships.
With this cognitive-linguistic re-interpretation of Geerearts’s prismatic
model in hand, it becomes possible to account for the structural and semantic heterogeneity of different idiomatic constructions: alternative idiomatic
activation-sets can be distinguished with regard to how many nodes and links
in the prism they contain as cognitively salient substructures. But before turning to such a cognitively-motivated typology of idiomatic constructions, I will
first describe the conceptual resources that can be potentially mobilised by a
cogniser to (re)motivate the internal semantic structure of a given idiom.
. Conceptual patterns shaping the internal semantic structure
of idioms
In order to explain the cognitive backing for motivation and isomorphism, it
must now be further elaborated how the abstract categorisation relationships
in the prism emerge. To do this, we must fill the skeletal model with conceptual content by further specifying the conceptual foundations that underlie the
potential recognition of the structural correspondences in the model.
For many idioms it is possible to structure the association of the literal
scene with the idiomatic scene on the basis of the complex conceptual patterns of semantic extension that were described in Section 3.3. These patterns
can be recruited to motivate the link between the two scenes and thus work as
conceptual backgrounds against which transparency can be assessed. The following systematic patterns of semantic extension can be expected to underlie
individual idioms: systems of conceptual metaphor, conceptual metonymies,
emblems, blending and complex combinations of these.30
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.29 (1571-1621)
.. Idiom-transparency and conceptual metaphor
In Section 2.3.2, Lakoff ’s cognitive-linguistic model of idiom motivation was
introduced. It was also shown that psycholinguistic experiments prove conceptual metaphors to be one central cognitive parameter for the motivation of
semantic regularities in idioms. Conceptual metaphors can thus be assumed
to work as extensive and coherent conceptual backgrounds underlying and
shaping the internal semantic structure of idioms.
As has been discussed in Section 3.3.1, conceptual metaphors form extensive conceptual networks to establish coherent metaphorical ICMs. Therefore,
idioms that refer to the same subparts of these networks can be assumed to be
linked associatively.
(2) the ace in your hand, have an ace in the hole, play your ace, hold all the
aces, hold/keep one’s cards close to one’s chest/vest, pass the buck, the
buck stops here, have a card up your sleeve, lay your cards on the table,
stack the cards
For instance, the idioms in (2) cannot be regarded as isolated expressions from
a cognitive-linguistic point of view. Rather, they appear as instances of the
more general metaphorical model goal-oriented human (inter)actions
are card games. The conceptual metaphor constitutes a superordinate
categorisation-pattern on the basis of which these individual idioms are conceptually connected. As will be shown in Chapter 5, this conceptual metaphor
is not an isolated model; rather it is linked to other metaphorical concepts for
success-oriented purposeful activity in agreement with Lakoff and Johnson’s
notion of metaphorical coherence (Lakoff & Johnson 1980: Chapter 17). As will
be revealed, this coherence is directly reflected in the system of idioms that refer
to purposeful activities in terms of denoting success, progress and failure.
The very systematicity reflected in the organisation of idiomatic expressions by conceptual metaphors lays bare the conceptualising power of these
cognitive structures. Arguing from the perspective of cognitive functionality,
one can therefore claim that it is the conceptualising function of conceptual
metaphors that gives rise to the motivated internal semantic structure of a great
number idiomatic constructions.
Idioms also reflect the process of metaphorical composition (see Section
3.3.1). This process leads to what I have termed multiple metaphor motivation
(Langlotz unpublished 1998: 95). For instance, the following idioms denoting
insanity illustrate that more than one metaphorical ICM may underlie the
semantic structure of a single idiomatic construction:

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.30 (1621-1672)
 Idiomatic Creativity
(3)
#have a maggot in the brain, #have rats in the attic/upper-storey, have bats
in the belfry, have a bee in the/one’s bonnet, have kangaroos in your top
paddock (Australian English).
One the one hand, these idioms are motivated by the conduit-metaphor
(ideas are objects, the mind is a container) with its corresponding submetaphors (e.g., the mind is a the upper part of a building derived from
the body is a building). On the other hand, they feature a basic metaphorical
model to refer to insanity: insanity is being obsessed by another creature/force. The systematic nature of the lexical variations in these synonyms
and quasi-synonyms shows that the composite metaphorical conceptualisation
shapes their semantic structure internally and thus motivates them.
While conceptual metaphor plays a decisive role for idiom motivation, it
is important to emphasise that although many idioms can be related to such
comprehensive and interacting metaphorical systems, some metaphorical idioms cannot. Rather, these idioms merely function as image-based one-shot
metaphors (see Lakoff & Turner 1989: 89–91; Lakoff 1993a: 229–231). For instance, image-metaphors underlie the following idioms: a couch potato, a bag of
bones, a sad sack, a pretty kettle of fish. The association of the idiomatic meaning
with the literal meaning in these idioms is grounded in the images associated
with the metaphorical sources (potato, bag, sack, kettle of fish) and the
corresponding (schematic) images associated with the target conceptualisations (physical posture of person, chaotic situation).
.. Metonymic motivation
Metonymic models play a very important function for the conceptual patterning of idioms. Thus, the following idioms belonging to the word-field of stupidity are motivated on the basis of the head for mental faculty-metonymy:
(4) to be soft in the head, to have a shallow brain, to be a numskull, to be
addle-pated, to be dead from the neck up.
As these examples indicate, the head for mental faculty-metonymy provides conceptual access the domain of cognitive ability by highlighting concepts related to the head anatomy-frame.31 The contiguity relationship motivating this link is obvious: the respective organs constitute the parts of the body
where our mental activities are supposed to take place (head) or by which
they are performed (brain). In the above expressions, the concept of stu-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.31 (1672-1726)
Chapter 4. Idiom representation – a cognitive-linguistic model 
pidity is therefore metonymically characterised as an effect of the ‘abnormal’
constitution of these organs (cf. also Feyaerts 1999).
Very often metonymic motivations are not based on physical contiguity relationships, but rather occur as more abstract cause for effect, effect for cause, or part for whole mappings in action-based scripts (cf.
Nordén 1994). For example, the expressions put up one’s sleeves (work hard)
and put one’s head down (work hard, concentrate) derive their idiomatic meanings by highlighting salient physical actions (the parts) within scripts for
physical and intellectual work (the wholes). The actions profiled by the idioms’ literal-meanings are thus linked to their idiomatic-meanings in terms of
conspicuous/symptomatic action for activity-metonymies. For instance,
the putting-up-one’s-sleeves scene describes a preparatory action before hard
physical work. The idiom’s literal scene thus works as a metonymic reference
point relative to which the idiomatic meaning can be motivated (see also the
analysis of spin one’s wheels above).
A specific subgroup of action-based metonymic idioms are what Burger
et al. (1982: 52–60) describe as Kinegramme (kinetic idioms), (see also Burger
1998a: 44–45, 61–62).
(5) grit one’s teeth, bite one’s upper lip, pull/tear your hair out, with open
arms, hold someone’s hand
Kinetic idioms like (5) refer to conventionalised non-verbal behaviour. These
specific actions or states often highlight conventionalised physical gestures,
activities or symptoms to express underlying emotional states: for example,
hold someone’s hand stands for support. Thus, kinetic idioms follow the
metonymic model non-verbal behaviour for conventionally associated emotional cause. When kinetic idioms are used to describe constellations where the concrete non-verbal behaviour is not actually performed, e.g.
The scientists accepted the new model with open arms, they become metaphorical. Thus, they reflect the principle of ‘metaphor from metonymy’ as described
by Goossens (1990: 336), (see Section 3.3.5).
Interaction between metaphor and metonymy can also be observed with
the idiom have a screw loose. The literal meaning of this expression metonymically highlights a specific cause, the loose screw, for its effect, the malfunctioning of a machine. This metonymic cause for effect relationship is then
metaphorised to express insanity via the mind is a machine-metaphor. The
metaphorical transfer does not directly apply to the state described by the literal
meaning of the idiom, but to the associated entailments of that state. These entailments are presupposed in the metonymically-highlighted literal-scene. As
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.32 (1726-1779)
 Idiomatic Creativity
we have seen above, the presence of such metonymic focus-shifts can have an
negative influence on idiom analysabilty.
Since metonymic motivation is strongly dependent on conceptual contiguity, i.e. the activation of frames and scripts with their corresponding ontologies
and structures, metonymy-based idioms can only be transparent as long as the
relevant knowledge that underlies the metonymic shift remains accessible to
the language users. The loss of the linking domain in metonymy-based idioms
can therefore be regarded as one of the primary factors leading to opacity.
.. Emblematic motivation
A considerable number of idioms are internally structured on the basis of
emblems (cf. Dobrovol’skij 1997: Section 2.4.2.4). Emblems transport stereotypical cultural information which can serve as a direct basis for idiom motivation.32 The more salient an emblem is, the more motivated an idiom becomes.
The degree of transparency of an emblem-based idiom therefore depends on
the degree of cognitive entrenchment and transparency of this conceptual substructure.
A well-entrenched and transparent emblem can be found in expressions
such as a wolf in sheep’s clothing, cry wolf, throw somebody to the wolves or keep
the wolf from the door. In these expressions, wolf represents a cultural stereotype for danger, hunger and evil and can therefore be attributed the status of
a salient emblem. Differences in the transparency of emblems can be seen with
the expressions (be) in seventh heaven and at the bottom of your heart, a broken
heart, cry your heart out, a heart of gold. While the number emblem seven has
a very weak degree of cultural entrenchment, heart is a very strong emblem
known by all speakers of English (and other languages). This directly affects the
transparency of these idiomatic constructions. The symbolic value of seven
‘very much, very good’ probably derives from its evaluation as a sacred number. It is certainly fair to argue that this emblematic value of the number is not
very well-entrenched in the minds of most present-day speakers of English. In
contrast, heart must be seen as one of the strongest emblems at all. Its emblematic value as the centre of human emotions is deeply anchored in a great
number of cultures and can possibly be attributed universal distribution. Very
likely, the strength of this emblem is due to its metonymic foundation: the heart
rate reacts to and correlates with emotional states.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.33 (1779-1822)
.. Hybrid ‘literal’ scenes – conceptual blending
A considerable number of idioms reflect the cognitive process of conceptual blending. Blended metaphorical idioms such as (6) do not possess an
experience-based literal meaning, but evoke hybrid images the are mentally
constructed by integrating conceptual substructures from distinct ICMs (see
Langlotz 2004a: 267; also cf. Mena Martínez 2002).
(6) rack one’s brains, like greased lightning, have butterflies in one’s stomach
Blended metaphorical idioms are well-motivated in most cases. For instance,
the adverbial idiom like greased lightning (very quickly, suddenly) blends the
lightning-frame with the machine-frame (greased) to create a hybrid but
concrete model for extreme velocity (see Langlotz 2004a: 265–266). Similarly,
a wolf in sheep’s clothing blends our conceptual knowledge about disguise and
deception, which is inherent in the conceptual metaphor you are what you
wear. The blend that underlies this idiom fills the human actor and dressnodes in the disguise-scenario with the emblems wolf, ‘evil character,’ and
sheep, ‘innocent character.’ The motivation to blend these emblems into the
source domain comes from the figurative target ‘an evil person who behaves as
if innocent.’ In other words, the two emblems are recruited because they match
the abstract target in a stereotypical way. The emergent literal scene, verbalised
by the lexical structure of blended metaphorical idiom, serves the purpose
of providing an imaginative explanatory model to concretise the experience
covered by the idiomatic meaning.
In conclusion, idioms can be motivated on the basis of several dimensions
of conceptual structuring. These underlying patterns of semantic extension
are often combined to create composite conceptualisations. The variety of alternative conceptual processes and backgrounds that underlie the semantic
structure of different idioms as well as the degree to which such conceptual
bases can be mobilised by different speakers triggers the semantic heterogeneity
of different idioms.
. Typical patterns of figuration reflected by idioms
After this cognitive-linguistic investigation into idiom semantics, we can now
propose the following analytical grid to scrutinise the organisation of idiomatic
predications: idiom transparency can be maximally described in terms of the

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.34 (1822-1863)
 Idiomatic Creativity
following categorisational relationships that are motivated by corresponding
conceptual bases:
Dimensions of idiom transparency
1. General patterns of motivation (paradigmatic dimension, top-down):
a. Global motivation:
The idiom can be motivated as a whole by relating the literal scene to the
idiomatic meaning
b. Constituental motivation:
The constituents are semantically autonomous because they have lexicalised figurative senses.
Effected by:
i. Metonymic motivation:
Recognition of contiguity relationship of constituent structure to idiomatic
meaning through conceptual metonymy.
ii. Metaphorical motivation:
Recognition of conceptual or image-schematic links between literal-scene
and idiomatic meaning through conceptual metaphor (systems), conceptual
blending or one-shot metaphors
iii. Emblematic motivation Recognition of emblems in constituent structure
iv. Mixed motivation through interaction of patterns of semantic extension
2. General patterns of figurative-literal isomorphism/analysability (syntagmatic
dimension, top-down):
a. Lexicalised figurative senses:
cf. constituental motivation
b. Phrase-induced figurative senses:
The constituents can be attributed an autonomous figurative sense in the
context of the idiomatic construction.
Effected by:
The recognition of direct conceptual or image-schematic correspondences
between the constituents and figurative correspondents in the idiomatic
meaning on the basis of conceptual metonymy, conceptual metaphor, conceptual blending or emblems.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.35 (1863-1922)
Chapter 4. Idiom representation – a cognitive-linguistic model 
3. Compositionality (syntagmatic dimension, bottom-up):
a. Partial compositionality:
The literal meaning of a constituent contributes directly to the idiomatic
meaning.
b. Literal compositionality:
The constituents can be integrated to construct a meaningful literal-scene.
c. Literal non-compositionality (literal opacity):
The constituents cannot be integrated to construct a meaningful literalscene.
Effected by:
i. Cranberry constituents
ii. Constructional idiosyncrasy
iii. Schematic constituents
With this analytical apparatus in hand, it becomes possible to arrive at a basic cognitive-linguistic classification of different types of figuration in idioms
(see also Geeraerts 1995: 61–73). In cognitive-linguistic terms, the types correspond to the different conceptual node-activation patterns that are reflected by
different idiomatic activation-sets.
At this point, an important qualification must be made. As I have claimed,
idiomatic entrenchment through institutionalisation and lexicalisation is a
gradual phenomenon that is potentially different from idiom to idiom (see discussion in Section 4.3 above). Moreover, its recognition can be different from
speaker to speaker. Since cognitive linguistics stresses the fundamental role of
the cognising language user for linguistic structuring, it must account for the
influence of individual differences on the mental representation of linguistic
units. With regard to the cognitive representation and use of idiomatic constructions, differences between the cognitive grammars of different language
users seem to be particularly influential. Being dependent on complex conceptual associations, the internal semantic structure of idioms is potentially
variable from speaker to speaker: while an idiom may be motivated for speaker
A, it is potentially opaque for another speaker B. This problem has been repeatedly emphasised (see, e.g., Burger et al. 1982: 28; Schenk 1995: 256–257; Moon
1998: 23). The idiom-activation-set model must therefore be understood as
reflecting a dynamic cognitive structure with degrees of complexity and nodeactivation that are different from idiom to idiom, from speaker to speaker, and
from speech-context to speech-context.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.36 (1922-1979)
 Idiomatic Creativity
Nevertheless, the basic argument of this study is that the variation
potential of idioms can only be explained if different types of potential
node-activation are considered. It is therefore necessary to distinguish different idiom-transparency types. These types reflect the different idiomatic
activation-patterns that can be evoked in actual discourse. It is the differences
of idiomatic transparency that trigger different forms of idiomatic usage and
variability. The more lexicalised an idiom is, the less variable it becomes. Cognitively speaking, the poorer an idiomatic activation-set can be conceptually
structured, the more restricted its variation potential becomes.
The following overview focuses on the most salient patterns of figuration
reflected by idioms. The classification is exemplified on the basis of some English verbal idioms with V + NP (+PP) structure. The classification constitutes
the heuristic basis for the analysis of the lexicogrammatical variability of idioms
in the analytical Chapter 7:
1. Core types: idioms with an inconspicuous, fully compositional literal
meaning
a. Idioms with literal compositionality, (global) motivation and figurativeliteral isomorphism, e.g.: grasp the nettle, take the bull by the horns, spill
the beans, burst/prick the bubble, upset the applecart, etc.
This class corresponds to Nunberg et al.’s class of idiomatically combining
expressions. Among many other expressions, grasp the nettle also belongs to
this transparency class; it can be motivated on the basis of the conceptual
metaphors dealing with an issue is grasping it, problems are dangerous objects (see Section 2.4). These metaphors can be directly applied to the
constituents. We know that nettles are dangerous plants and to grasp them
means to be courageous and determined. Thus the constituents grasp and nettle
can be attributed the phrase-induced meanings ‘approach determinedly’ and
‘problem, danger.’ These meanings appear only in the context of the idiom.
As has been discussed above, this type is not restricted to fully transparent
expressions, but also includes idioms that can mainly be grounded in imageschematic correspondences (e.g., spill the beans).
b. Idioms with literal compositionality, global motivation, constituental
motivation, and figurative-literal isomorphism, e.g.: swallow the bitter
pill, jump on the bandwagon, make headway, etc.
Even more transparent than (1a) are all those idioms whose constituents possess lexicalised figurative senses that can also appear outside the idiomatic
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.37 (1979-2031)
Chapter 4. Idiom representation – a cognitive-linguistic model 
context (constituental motivation). As we have argued, the constituents swallow and bitter pill in the expression swallow the bitter pill have the lexicalised
figurative meanings ‘accept/bear’ and ‘unpleasant fact.’ These meaning are
themselves motivated by conceptual metaphors: dealing with an issue is
consumption, coping with an issue is swallowing/digesting it, quality
of issue is quality of consumed substance. Given the lexicalised subsenses
of the constituents, the idiom is motivated and isomorphic ‘by nature.’ Thus,
with Geeraerts (1995: 67) one can argue that for idioms of this type the semantic prism is “fully realized.” In cognitive-linguistic terms, their conceptual
activation-set is very rich and clearly structured.
But how did the subsenses of swallow and bitter pill become lexicalised as
autonomous lexemes? In my view, the independent meaning of bitter pill represents an example of the word formation process that Gläser (1986: 22) calls
dephraseological derivation (“dephraseologische Derivation”). The fact that idiomatic constituents like bitter pill can break away from the linguistic context
of the idiom and nevertheless preserve their figurative meaning is a very strong
indicator for the existence of isomorphism and its evocation by underlying
conceptual patterns of semantic extension.
c. Idioms with literal compositionality and global motivation, e.g.: grit
one’s teeth, spin one’s wheels, turn the page, go round in circles, etc.
With this type, the metaphorical transfer from the literal to the idiomatic
meaning is still vivid, but the constituental structure is not semantically
analysable. Semantic extension in these idioms often involves metaphtonymy,
i.e. metaphorical extension based on metonymy. The metonymy highlights an
entailment in the literal-meaning domain. Thus, go round in circles entails stagnation. This entailment is then metaphorised to the more abstract idiomatic
meaning: go round in circles denotes stagnation in any type of target activity.
Since this process of figuration projects a metonymic association-pattern onto
the target-domain, this idiom – as well as the other idioms of this class – does
not reflect direct constituental mapping as reflected by (1a) and (1b). This corresponds to Dobrovol’skij’s (1988: 51) distinction between global-situative vs.
konstituentenbasierte Motiviertheit.
d. Idioms with literal compositionality but with neither motivation nor
isomorphism, e.g.: kick the bucket, pop your clogs, bite the dust, buy the
farm, carry the can, trail your coat, etc.
With this group, we reach the class of opaque idioms. The idiomatic meanings
of these expressions must be regarded as homonymous to their literal ones
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.38 (2031-2082)
 Idiomatic Creativity
(cf. red herring). In other words, it is not possible to activate a link that motivates the association between the two scenes. Since this book is an attempt to
propose an explanatory model for idiom-transparency and its effects on variation, we are forced to explain why, for these idioms, it is impossible or unlikely
for native speakers to find a motivated relationship between the literal and idiomatic meaning. Following the logic of the cognitive-linguistic model, it can
be claimed that these idioms are opaque for one of the following reasons:
i.
The idiomatic meaning cannot be understood relative to the literal meaning because it is impossible to activate entrenched metaphors, metonymies
or emblems relative to which this link could be established.
ii. It is impossible to evoke a literal-meaning domain on the basis of
which conceptual correspondences to the idiomatic meaning could be established.
iii. The constituents do not have independent figurative meanings that can be
activated to link the two meanings in the absence of salient conceptual or
image-schematic correspondences.
Consider buy the farm for instance. Although it is possible to relate this action to a rich commercial transaction-script, there is no conceptual ground
which makes it possible to map this knowledge onto the target-process of dying. Moreover, none of the idiomatic constituents evokes a figurative subsense
that could be related to the idiomatic meaning. As a result, the expression
remains opaque.
For pop your clogs and trail your coat it is difficult to activate concrete domains to which these actions could be related. In the absence of an ICM which
could work as a conceptual base for these actions, the constituents pop and
clogs as well as trail and coat, respectively, cannot be related to the meanings
‘die’ and ‘provoke.’ The same is true for bite the dust. As the Longman Dictionary of English Idioms indicates, this idiom has its etymological origin in a
military context; it captures the image of a soldier being killed in battle: the
soldier falls to the ground with his face in the dust. In other words, the idiom’s
literal meaning is based on an image-metaphor. The position of a slain man
lying on the ground is compared to a person biting (into) the dust. Since conventional images of dying people do not correspond to this characterisation,
this image-metaphor can only be activated within the specific military context. However, since the constituents bite and dust do not provide any clues to
the military scenario, the idiom can only be motivated for those speakers who
know the idiom’s etymological basis.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.39 (2082-2131)
Slightly departing from these core idiom-activation-sets are a number of
idioms that have a conspicuous or non-compositional literal meaning. The existence of these additional subtypes underscores the fact that idioms do not
reflect uniform linguistic and semantic structures. Moreover, they illustrate the
heterogeneous nature of idiom-transparency.
2. Marginal types: idioms with conspicuous or non-compositional literal
meanings
e. Idioms with a compositional but experientially unrealistic literal meaning: blended metaphorical idioms, e.g.: cook the books, rack one’s brains,
like greased lightening, etc.
This group of idioms is characterised by having a literal meaning that is experientially unrealistic, unlikely, hyperbolic or even impossible. Traditionally,
idioms of this type have often been classified as being semantically ill-formed
(see, e.g., Botelho da Silva & Cutler 1993: 129).33 Opposing this view from
a cognitive-semantic perspective, these constructions are not defective, but
rather exemplify the constructive nature of conceptual modelling. In a nutshell, they are blended metaphorical idioms as defined above. The unrealistic
scenes conveyed by these expressions are imaginative, but precise metaphorical sources to concretise the idiomatic meaning.34 Unlike other peculiarities on the level of the literal meaning (see subsequent sections), conceptual
blending does not usually block global motivation. As has been shown above,
blended metaphorical idioms can often be globally motivated and may also be
isomorphic.
f. Partially compositional idioms, e.g.: know one’s onion, look daggers at sb.,
rain cats and dogs, etc.
In accordance with our definition of partially compositional idioms, this type
involves the direct literal contribution of one constituent to the idiomatic
meaning. Thus, in rain cats and dogs or look daggers at sb. the constituents rain
and look, respectively are to be understood literally. As these examples further
illustrate, a considerable number of semi-idioms are semantically and syntactically ill-formed. Look daggers at sb. violates the valence of the intransitive verb
look: one cannot look an object at somebody. Similarly, rain is conventionally
intransitive. In my opinion, these ‘violations’ can best be interpreted as the result of conceptual blending. Thus, with look daggers at sb. the action look at
another person is blended with the scene throw a dagger at one’s enemy.

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.40 (2131-2188)
 Idiomatic Creativity
The blend constructs the violent form of looking in terms of throwing daggers
at one’s opponent.
g. Literally non-compositional, constructionally idiosyncratic idioms, e.g.:
trip the light fantastic, come up roses, shoot the breeze, go places, etc.
As has been discussed above, constructionally idiosyncratic idioms such as trip
the light fantastic, shoot the breeze or come a cropper are opaque because it is
impossible to construct a meaningful literal-scene from the formal structure.
Nevertheless, these idioms can be recognised as complex constructions rather
than holophrastic sequences. One can therefore claim that for these expressions, the literal-scene only exists as a highly schematic mental representation:
shoot the breeze describes some form of shooting, trip the light fantastic is a
form of tripping and come a cropper is a form of coming.
h. Literally non-compositional idioms with cranberry morphs, e.g.: the
whole kit and caboodle, put the kibosh on sth., cock a snook, blow the
gaff, etc.
As with constructionally idiosyncratic idioms, the literal scenes for these expressions are highly schematic because their cranberry constituents cannot
contribute to the evocation of rich conceptual scenarios. As a result, most of
these idioms are opaque. Motivation for such expressions can only be constituental. For instance, with blow the gaff (tell people something that was
supposed to be kept secret), it is difficult to construct a sensible literal meaning
due to the presence of gaff. Still, its idiomatic meaning can be partially remotivated with the help of the lexicalised figurative sense of blow sth. (reveal sth.
as in blow the myth). Given this lexicalised figurative sense, it becomes possible to remotivate gaff to mean ‘secret.’ As a result, the idiom reflects potential
figurative-literal isomorphism. Variation data support this argumentation:
(7) Once the gaff about Whisky not being suitable for the kitchen was blown,
the [...]. (EFU: 525).
(8) [...] that will blow Rainbow’s gaff, and send the interloper sloping off.
(HGN: 2363).
The premodification and postmodification of gaff in these examples suggests
that this constituent may be attributed the phrase-induced meaning ‘secret’
through remotivation via blow (Langlotz 2001b: 299).
i. Idioms with absent literal compositionality due to the presence of highly
specialised word-meanings and garden-path constituents, e.g.: hide your
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.41 (2188-2239)
light under a bushel, have had one’s chips, across the board, chomp at the
bit, kick the bucket, etc.
This type of blocked literal compositionality is similar to the previous type.
Idioms of this type contain a lexical element with a very specialised meaning that is not generally known by many speakers. This, for instance, is true
for bushel (unit of volume equal to eight gallons) in hide your light under a
bushel (not show one’s true talents). While, from a strict, theoretical point of
view, this lexical unit cannot be attributed the status of a cranberry morph because it still appears as a separate entry in contemporary English dictionaries,
it can be regarded as cranberry-like from a cognitive point of view: for many
speakers this lexical unit does not exist as an independent symbolic unit in
their mental lexicons. Note, however, that the inability to activate a concrete
meaning for bushel does not necessarily render this particular expression unmotivated, although the literal scene does not trigger a rich conceptualisation.
(The literal scene can be rendered as hiding light under some cover.) However, the idiomatic meaning can be remotivated with the help of the following
conceptual bases. First, hide preserves its literal meaning; second, the conceptual metaphors knowledge is light and ideas are light sources make it
possible to ascribe the phrase-induced figurative sense skills to light; third
the PP under a bushel can be schematically motivated by the image-schematic
metaphor being secret is being underneath a cover. Relative to these
underlying conceptual bases the idiom becomes motivated and isomorphic
although the concrete derivational basis for the metaphor remains opaque.
The degree of motivation of a given idiomatic expression is further influenced by the presence of what I would like to call garden-path constituents. A
garden-path constituent is a word that possesses a number of polysemous or
homonymous meanings. If, for this constituent, the language user is unable
to activate the specific subsense by which the idiomatic meaning can be motivated, global motivation is blocked. For instance, champ/chomp at the bit and
get the bit between your teeth contain the homonymous constituent bit. The
prototypical meaning of this word is ‘a very small part of something.’ In the
present idioms, however, bit refers to the highly specialised homonym ‘piece of
metal held in a horse’s mouth by the reins and used to control the horse when
riding.’ Geeraerts (1995: 69) therefore calls such constituents “hidden cranberry morphs.” From a cognitive-linguistic point of view, constituents like bit
work as garden-path constituents: they are likely to mislead a language user in
the process of (re)motivating correponding idioms. Assuming that the speaker
does not know the etymological basis for champ/chomp at the bit and get the bit

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.42 (2239-2296)
 Idiomatic Creativity
between your teeth, he or she is very likely to evoke the prototypical meaning
of bit (small amount of sth.) to make sense of these expressions. However, in
doing so, one is led up a semantic garden path and, as a result, becomes unable
to motivate the expression (see also Langlotz 2001a: 115–116).
Relying on the notion of garden-path constituent, it is also possible to explain the semantic opacity of kick the bucket, which is commonly listed as a
prototypical, opaque English idiom. Indeed, it is difficult to explain why this
construction means what it does. Some people relate the action of kicking the
bucket to a suicidal hanging-scenario: the agent, who wants to kill himself,
stands on a bucket to put his head into the noose; at the moment he kicks
the bucket away from below his feet, the weight of the body pulls him down.
As a consequence, he is strangled and dies. Although being related to death,
this explanation seems unrealistic. The idiomatic meaning of kick the bucket
denotes the process of dying rather than an act of suicide. How then did the
literal meaning kick the bucket become a homonym of the idiomatic meaning
‘die’? The Oxford English Dictionary and the Longman Dictionary of Idioms provide a more realistic etymological explanation. According to these references,
the idiom originates from an ancient method for slaughtering pigs. Before being killed, the pigs’ feet were tied to a beam – the bucket – to prevent them
from kicking out and to hang them up by the heels after being killed. At the
moment the pigs were killed, they kicked the bucket for a last time. Therefore, this movement was seen as an indicator of their death. In other words,
the action kick the bucket stands in a metonymic relationship to the idiomatic
meaning: it is the effect of the underlying cause – death (see also Ortony et
al. 1978: 476). On the basis of this etymological explanation, we can now provide a cognitive-linguistic explanation for its cognitive opacity. The constituent
bucket can be regarded as a ‘garden-path constituent.’ Being a highly specialised
term, the meaning ‘beam used in butchering’ is not part of the mental lexicon of ordinary speakers of English. Therefore, they are unable to activate the
relevant encyclopedic knowledge which serves as the ground to motivate the
metonymic shift from the literal to the idiomatic meaning. Since the specialist term is homonymous to bucket, in the sense of ‘pail,’ people rather activate
the scenario of overthrowing a bucket, which, however, does not make sense
relative to the idiomatic meaning. Moreover, because the idiom is metonymically motivated, the ability to find a link between the literal and the idiomatic
meaning is dependent on the specific source of specialist knowledge relative to
which the contiguity relationship can be established. Therefore, any attempt to
interpret kick the bucket as a metaphor for death is condemned to failure: kick
the bucket is not a ‘dead metaphor,’ it is a worn out metaphtonymy.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.43 (2296-2339)
In conclusion, different idiomatic constructions can be claimed to reflect
different typical activation-sets. These activation-sets characterise alternative
ways in which idioms can be stored in the minds of speakers. It must be hypothesised that the qualitative differences reflected by these types have a direct
influence on how idioms are processed. To some extent this is reflected by
psycholinguistic experiments. However, the present cognitive-linguistic classification of idioms is much more detailed than the classifications commonly
employed for psycholinguistic testing (cf. Glucksberg 1993: 16–19). Hence, it
would be highly desirable for psycholinguistic models of idiom comprehension and production to take the more fine-grained distinctions between these
different idiom types into account.
. The cognitive functionality of idioms
Man’s ideas do not chart a steady progression towards some ultimate philosophical or scientific truth. Mostly, they go round in circles, reflecting only the
idiom of the day.
(Davidson 1991)
In Chapter 1, idioms were defined as having an ideational function: idioms
are evoked to refer to constellations (things, states, situations, ideas or processes) appearing and occurring in our world of experience. On the basis of
the cognitive-linguistic model of idioms as complex mental activation-sets,
their functionality can now be reconsidered. Above, we have defined the semantic poles of idioms as complex scenes. In other words, they constitute
conceptual routines that are evoked to group a target-scene relative to an alternative source-scene. But what is the cognitive value of these indirect conceptual
associations?
.. Idioms as cognitive micro-models
As sketched in Section 2.3.1, Nunberg et al. (1994) try to explain the phenomenon of idiomaticity with regard to its discursive functionality, which
resides in what they call proverbiality:
Proverbiality: Idioms are typically used to describe – and implicitly, to explain – a recurrent situation of particular social interest (becoming restless,
talking informally, divulging a secret, or whatever) in virtue of its resemblance
or relation to a scenario involving homey, concrete things and relations –
climbing walls, chewing fat, spilling beans.

JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.44 (2339-2383)
 Idiomatic Creativity
Nunberg et al. thus attribute an explicit descriptive and an implicit explanatory
function to idioms. According to them, these functions are characteristically
mirrored in the tendency of idioms to associate abstract with more concrete
situations. When considering the general cognitive functionality of semantic
extension, it becomes possible to incorporate this view of proverbiality into
our cognitive-linguistic framework.
The cognitive function of metaphor and blending involves three macrofunctions: conceptualisation, concretisation and reconceptualisation. These
dimensions involve the subfunctions of creating conceptual similarities between two experiential domains, providing easier cognitive access to the target
realm and altering the cognitive perspective on an given experience. As has
been shown, these functions are characteristically fulfilled by mapping conceptual entities from an empirically-concrete to an empirically-abstract domain
of experience. The same qualitative functionality also underlies metonymy:
metonymies select highly salient conceptual reference-points to access contiguous conceptual entities that are more abstract or harder to name. Similarly,
emblems also work as concrete conceptual trigger points to activate more abstract evaluative associations (e.g., gold symbolises richness, importance,
value, luxury).
Since a great number of idioms can be related to institutionalised metaphors, blends, metonymies and emblems, they perform the very same cognitive
functions. From a cognitive-linguistic point of view, proverbiality must therefore be understood as a process of conceptual modelling. The literal meaning
of the idiom highlights a concrete scene, which is evoked to work as a model for
a conceptually more abstract constellation. This observation is further underscored by the fact that the constituent-structure of idioms typically encodes
basic-level concepts (see Section 3.2.2) that are interconnected in terms of
basic force-dynamic (image-schematic) relationships like compulsion, containment, obstruction, manipulation, etc. Idiom constituents are often
related to body parts (head, hand, foot, leg, arm, knee, etc.), (domestic)
animals (dog, cat, pig, horse, mouse, etc.), food (nut, apple, beans, etc.),
tools/vehicles (hammer, screw, wheel, ship, cart, etc.) and related processes
(cook, hold, eat, catch, etc.), (for a comprehensive analysis of basic idiom
constituents see Hegedüs-Lambert 1995). Carter (2004: 132) further highlights
that “idiom uses do not simply describe but comment in positive and (more
usually) negative ways on events, processes and persons.” Thus, by profiling
empirically-concrete everyday frames and scripts and by projecting them onto
more abstract conceptual relationships, idiomatic expressions function as pre-
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.45 (2383-2432)
Chapter 4. Idiom representation – a cognitive-linguistic model 
coded cognitive micro-models to concretise and evaluate a great number of
abstract target-constellations.
The specific functionality of idioms as cognitive micro-models can be exemplified by upset the applecart. The lexical constituents of this idiom trigger a concrete scene in which a stable configuration is brought out of balance. The image of a cart loaded with apples evokes rich image-schematic
and conceptual associations that make the abstract idea of causing instability/disequilibrium comprehensible: being round objects, apples are notoriously difficult to maintain in a balanced position when they are piled up on a
cart. The concept applecart thus iconically represents the idea of a balanced
situation that fulfils its purpose satisfactorily. To upset the applecart means
to spoil this balanced configuration. This abstract implication constitutes the
expression’s idiomatic meaning. The literal scene with all its image-schematic
and conceptual entailments thus works as a rich and accessible micro-model
that can be projected onto more abstract target-events. Thus, the encyclopedic knowledge associated with the lexical constituents makes it possible to
concretise, describe and explain a more abstract target scenario of induced disequilibrium. In this way, the idiom’s literal scene works as a mental background
against which the conceptual relationships in the target conceptualisation can
be assessed.
A very concrete and nice illustration of the modelling power of idioms is
hidden in the quotation that introduces this subchapter. The author makes a
statement about the development of human knowledge, which is portrayed as
not advancing, but as changing circularly without progression: “Man’s ideas do
not chart a steady progression [...], they go round in circles.” Funnily enough,
the author formulates his claim by recurring to an idiomatic expression, go
round in circles. The literal scene of this construction evokes concrete knowledge about physical movement in space: to reach one’s desired destination, one
has to move forward, whereas going round in circles implies losing a great
deal of energy without any progress. This knowledge underlies the conceptual
metaphor progress is forward movement. In other words, the author develops his own idea with reference to this conventionalised metaphorical model
that is permeated in the symbolic structure of the idiomatic construction. In
doing so, he indirectly proves the validity of his statement: even his own idea
only reflects an idiom of the day.
In a nutshell, idioms fixate complex conceptual micro-models in the mode
of institutionalised and lexically-rich figurative constructions. In this way, idiomatic language perpetuates conventionalised and generally accepted models
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.46 (2432-2500)
 Idiomatic Creativity
of the world (see Dobrovol’skji 1995: 61–70). Hence, idioms play an important
role in conveying a pre-configured system of conceptual knowledge.
.. The different qualities of specific idiomatic micro-models
It was shown that different idiomatic constructions reflect different degrees and
types of transparency and opacity, respectively. These differences must have a
great influence on the cognitive functionality of idioms. Therefore, the notion
of idiomatic micro-model must now be qualified, i.e. the cognitive function
of different idiomatic configurations must be expected to vary depending on
the quality of the conceptual correspondences that shape their semantic structure. More specifically, if idioms work as cognitive micro-models that map a
concrete literal-scene onto a more abstract target-scene, then this function can
only be fulfilled if the idiom’s literal meaning is concrete indeed. Moreover,
one cannot expect that opaque idioms can fulfil the same cognitive purpose as
transparent ones. The role of idioms as micro-models is therefore subject to
two qualitative parameters that trigger functional variation:
a. The semantic quality of the literal scene.
b. The quality of the connection between the literal scene and the target scene.
Given these parameters, specific idioms can range between the two opposite
poles of zero-models and full-scale explanatory models. In what follows, the position of different types of idiomatic activation-sets along this continuum will
be outlined.
At the zero pole one can find idioms like spick-and-span or the whole caboodle whose literal scene is (almost) fully opaque. Since it is impossible to
evoke a meaningful mental background against which the idiomatic meaning
of these idioms can be assessed, they cannot fulfil the function of cognitive
modelling. In this sense, these constructions function like words: their meaning
is directly evoked on the basis of a specific formal structure. I will therefore call
the function performed by these idioms alternative symbolisation. While such
idioms do not perform a cognitive modelling function, they can nevertheless
be regarded as communicatively functional from a socio-pragmatic perspective. Adapting an argument that was proposed by Gibbs (1989) in the context
of metaphorical functions, such expressions can be claimed to trigger intimacy
effects. More specifically, the ability to share, use and understand such opaque
and idiosyncratic means of expression “often presupposes and reinforces an intimacy between speaker and listener, the cultivation of which is, perhaps, the
primary function of such language” (Gibbs 1989: 250).
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.47 (2500-2541)
Chapter 4. Idiom representation – a cognitive-linguistic model 
Above, we have further noticed that it is often impossible to activate a
meaningful literal scene for constructionally idiosyncratic idioms. For instance,
the highly schematic literal meanings of blow the gaff, trip the light fantastic or
shoot the breeze do not evoke concrete conceptual knowledge. Therefore, their
modelling power is highly restricted. This, however, is not to say that these
constructions have the status of unanalysable words. Rather, they constitute
complex scenes with very abstract literal standards. Although abstract, these
standards can still be recognised as processes: trip the light fantastic is a subtype
of tripping, and blow the gaff is a kind of blowing. In other words, the cognitive
ability to recognise the semantic contribution of the verbs causes them to be
perceived and processed as constructions rather than lexical items. I will therefore use the label schematic reformulations to refer to the restricted modelling
function performed by such expressions. Again, it is very likely that such expressions create intimacy effects to support acts of social bonding. Knowing
and using these linguistic routines asserts “a sense of intimacy or complicity”
(Gibbs 1989: 250).
The role of idioms as cognitive micro-models is not restricted to fully
transparent expressions, but can also be fulfilled by opaque ones. Consider
kick the bucket. Although this idiom is not motivated, it provides an alternative description of the event of death in terms of the action of kicking a bucket.
Therefore, the expression fulfills a reconceptualising function. This function
is not dependent on the idiom being semantically motivated, but it is fulfilled
once the literal meaning of the expression is understood. Consequently, kick
the bucket establishes a micro-model for dying that is institutionalised but unmotivated and thus highly idiosyncratic. With reference to this idiom, one can
assume that idioms receive their stylistic status to a great extent from the tension between of the literal scene and the idiomatic meaning. As we have argued
in Section 4.4.1, the literal scene of an idiom makes it possible to derive a great
number of weak implicatures that could not be triggered by non-figurative language (cf. Sperber & Wilson 1986: 217–224; Pilkington 2000: Chapter 7). These
added dimensions of conceptual and emotional structuring create a further
level of meaning that must be integrated with the target conceptualisation. Expressions like kick the bucket or bite the dust are stylistically marked as ‘slang,’
‘colloquial,‘ ‘informal’ or even ‘offensive.’ To a great extent, the domain of death
is under taboo. The reconceptualisation of death in terms of an alternative
scene, thus works as an avoidance strategy. Following Anders (1995: Chapter
III.4) the socially-sanctioned demand “never say die” can be implemented by
euphemisms (pass away) as well as disphemisms (kick the bucket, bite the dust).
While euphemisms try to describe death in positive terms, disphemisms are
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.48 (2541-2583)
 Idiomatic Creativity
used to create a shock effect or to turn the phenomenon into ridicule. Both
euphemisms and disphemisms depend on the evocation of idioms as complex
scenes. The activation of the literal scene triggers conceptual associations that
are either experienced as decent or as offensive relative to their conceptual target. The weak implicatures of complex idiomatic scenes thus emerge through
the cogniser’s integration of the literal scene and the idiomatic meaning. The
literal scene works as a rich resource for weak implicatures because it encodes
directly accessible basic conceptual and preconceptual experiences that are rich
in both conceptual, sensory and emotional content (for a good overview of the
cognition of emotion see Pilkington 2000: Chapter 6).
Comparing opaque idioms to transparent ones like go round in circles or
upset the applecart, it is possible to describe the following potential range of
variation with regard to the function of idioms as cognitive micro-models:
1. Explanatory micro-models:
With explanatory idiomatic micro-models the literal scene is evoked to explain the idiomatic meaning. Explanatory idiomatic micro-models must
therefore be transparent. I regard both metaphor-based and metonymybased transparent idioms as explanatory. However, their cognitive functionality is slightly different, one must thus distinguish:
a. Symptomatic models; cause-effect models:
In metonymy-based idioms, e.g. grit one’s teeth or grease the wheels, the
physical symptom or cause described by the literal meaning is evoked
to access the related emotional or physical state (e.g., endure very unpleasant situation, progress faster). In other words, the existence of the
associated emotion or physical effect is indicated by the metonymic
standard.
b. Ontologising models:
Metaphor-based models (including blends) concretise the conceptual
relationships in the target domain by mapping the concrete ontology of
the source onto the target. The expression take the bull by the horns explains the meaning ‘tackling a difficult problem by dealing with its most
dangerous part’ by providing a specific framework to ontologise these
abstract relationships.
As we have seen above, metonymy and metaphor may interact in shaping the internal semantic structure of idioms. Symptomatic models thus
often occur in combination with ontologising ones.
JB[v.20020404] Prn:20/02/2006; 14:20
F: HCP1704.tex / p.49 (2583-2636)
2. Descriptive micro-models:
Descriptive idiomatic micro-models merely represent the idiomatic meaning in terms of another conceptual relationship. Since the association between the literal scene and the idiomatic meaning is not motivated, they
do not fulfil any other cognitive function than to reconceptualise the target
as something else: buy the farm, kick the bucket, chew the fat, carry the can,
have a chip on one’s shoulder.
Since the distinction between idiomatic opacity and transparency is gradual, the descriptive and explanatory status of idioms must also be seen as
being subject to gradual variation: the more opaque an idiom is, the more
descriptive and less explanatory its function becomes.
3. Schematic reformulations:
Here, the literal standard is highly schematic. Therefore it cannot conceptualise or reconceptualise the target. Rather, it provides a schematic
reformulation of the idiomatic scene that cannot be motivated: come a
cropper, shoot the breeze, go places, trip the light fantastic.
4. Alternative symbolisations:
This final functional class of idioms reflects the zero-state of cognitive
modelling. For these idioms, the literal scene is fully opaque, it cannot
reorganise the target in any way, but merely works as a alternative symbolisation for it: the whole caboodle, spick-and-span.
It is important to emphasise that these classes do not constitute clear-cut taxonomic subcategories. Rather, they must be regarded as heuristic tools, to assess
the potential ideational functions that different idioms can fulfil in discourse.
In the following chapter, the systematicity of idiomatic modelling relative to
entrenched patterns of semantic extension is substantiated empirically.

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.1 (47-115)
chapter 
The conceptual motivation of idioms
denoting success, progress and failure
In this first analytical chapter, the idiom activation-set model is subjected to
empirical scrutiny. The chapter tries to reveal the extent to which idioms can be
motivated and structured by well-entrenched patterns of semantic extension,
particularly conceptual metaphors. With the help of the cognitive-linguistic
model of idiom representation, this analysis thus aims to clarify the controversy centred about the role of conceptual metaphor for idiom motivation (as
outlined in Section 2.3.3). To uncover how conceptual backgrounds pattern
the semantic organisation of idioms, a distinct set of idioms will be modelled by means of the cognitive-linguistic architecture of conceptual knowledge
outlined in Chapter 3.
The data for analysis is derived from approximately 600 standard British
English idioms that denote success, progress or failure (spf). The full list
of idioms is given in the electronic appendix. This particular word-field was
chosen because I assumed the evaluation of the success, progress or failure
of different types of activities (political and social events, sports events, personal experiences and so forth) to constitute a very basic and frequent topic
in discourse. Hence, I expected idioms of this word-field to appear relatively
often and thus to provide a representative sample of the idiomatic phrasestock. Moreover, success, progress and failure are abstract concepts; therefore a great number of these idioms can be expected to reflect underlying
metaphorical models.
To extract the set of spf-idioms, I read through the Collins Cobuild Dictionary of Idioms (CCDI) (1995) and chose all expressions whose meaning
referred to success, achievement, progress, failure, stagnation, regression and
circumstances that lead to either success or failure (problems, etc.). I chose
the CCDI as a resource for spf-idioms because it only includes constructions
that are fully compatible with the definition of idiom advocated in this book.
Containing approximately 4400 entries, the CCDI offers a broad collection of
contemporary British and American English idioms. Its entries, compiled on
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.2 (115-193)
 Idiomatic Creativity
the basis of the Bank of English-corpus, are specified for their distribution in
these varieties. Since my study is based on British English data, the CCDI was
perfect to exclude idiomatic constructions that do not occur in this variety.
As discussed above, the view of idiom motivation advocated in this study
is broad: it makes reference to different processes of semantic extension
(metaphor, metonymy, blending and emblems), it is based on cognitive models and it includes image-schematic motivation. With a view to the description
of the semantic structure of idioms, such a broad definition of motivation
has various advantages over traditional linguistic accounts. While many traditional phraseological approaches acknowledge that motivation is a salient
phenomenon of idiom semantics, they generally do not attempt to explain it
as an effect of more general conceptual regularities. This leads to a methodological shortcoming: when trying to describe the motivation of individual idioms, the constructions are generally analysed in descriptive isolation. In other
words, the traditional phraseological analysis of idiom semantics is negatively
influenced by what Lakoff and Turner (1989: 128) have called a “Case-by-case
Methodology” in the context of metaphor theory. In contrast, the following
analysis scrutinises the motivation of spf-idioms relative to extensive and systematic bases of conceptual knowledge. It attempts to outline the coherent,
underlying conceptual patterns of semantic extension that can be activated to
structure this subsystem semantically.
. What is success, progress and failure? – A cognitive-model
To model the motivating bases for spf-idioms, the conceptual target-domain
must first be delimited. success, progress and failure are very common
evaluative concepts that we use for the comprehension and interpretation of
everyday events and activities: a bankrupt company has failed; an Olympic
champion has achieved success; science is expected to bring about social and
cultural progress, and so forth. In other words, we hardly ever perceive actions,
activities, events or experiences without evaluating the quality of their development or their results. But how can success, progress and failure be appropriately characterised and what other concepts belong to the abstract cognitive
model established by these notions? What is the ontology and structure of this
target domain?
Every activity or life event can be attributed an inherent aim. For instance,
the aim of a career is to get a good job and to earn a lot of money, the aim
of a competition is to win, while a love relationship is expected to make the
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.3 (193-226)
Chapter 5. The conceptual motivation of SPF-idioms 
lovers happy. To achieve this aim people use particular means or methods.
For instance, to win a football match, a team uses a particular tactics including
a specific line-up. Moreover, to achieve or approach the aim, the activity must
be performed. Depending on the adequacy of the means and methods and the
quality of the performance, the activity can either become successful or not.
The success and progress of the activity can be further influenced by additional
positive or negative factors such as problems and interferences or help and
support. In addition, positive or negative external circumstances affect the
achievement of the given aim. If, for instance, a football team is forced to play
a match without its best players, it is less probable for them to win the match.
Thus, the external conditions in which an activity takes place affect its chances
of success respectively failure. More precisely, success, progress and failure
can therefore be said to evaluate the result, state, performance or external circumstances of an activity relative to its inherent aim. Thereby the concepts
establish different evaluative foci on the given constellation. While success
and failure evaluate the result of the given activity or the status achieved
through it, progress and its (quasi-)antonyms stagnation and regression
rather constitute evaluations of the temporal development of the activity. Success and failure come about when the aim of the activity is achieved or not
achieved, respectively. Alternatively, a constellation is said to progress if it approaches its inherent aim, whereas it stagnates or regresses when the aim is
not approached. The two evaluative foci (result/status vs. development) can
be further linked. Since a progressing activity is likely to lead to a positive result, progress is conventionally associated with success, whereas stagnation and
regression imply failure.
The following general domain (Figure 5.1) thus works as an abstract conceptual base relative to which the concepts of success and progress vs. failure, stagnation and regression derive their semantic values. The concepts
can be understood as abstract conceptual profiles that are set off against the
background of this complex ICM for the comprehension and interpretation of
purposeful human activities or life events.
. The conceptual source domains for spf-idioms
The literal scenes of most spf-idioms can be related to six general source
domains: competition, struggle, life, eating, physical movement and
machine. The specific ontologies and structures of these conceptual bases are
mapped onto the spf target-domain to structure, conceptualise and concretise
stagnation
not
approach
not
achieve
problem
interference
ACTIVITY
performance
means
plan/method
Aim
(e.g., task, career, competition, process,
friendship)
Figure 5.1 The conceptual ontology and structure of the spf-domain
regression
development
of activity
evaluation of
result of activity
failure
bad
situation
help
support
approach
achieve
good
situation
progress
development
of activity
evaluation of
result of activity
success
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.4 (226-226)
 Idiomatic Creativity
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.5 (226-290)
Chapter 5. The conceptual motivation of SPF-idioms 
it. For instance, the literal scene in set the ball rolling refers to a salient stage
in a ball game-scenario. The scene profiles the starting point of the game.
Following the conceptual metaphor purposeful activities are ball games,
this idiom therefore denotes the event of starting a more abstract purposeful
activity such as a scientific project.
The global source-domains underlying the motivation of spf-idioms are
not isolated chunks of knowledge, but they have closely overlapping areas. This
is in full accordance with the view that the conceptual organisation of knowledge is not characterised by strict boundaries between semantic domains (see
Langacker 1987: Chapter 4). For example, the conceptual metaphor purposeful activities are battles (as implied in fight to the death) can be interpreted
in relation to the source-domains of competition, struggle or life. On the
one hand, battles are competitive struggles between two armies and are thus
similar to other forms of competition such as football matches, races or card
games; on the other, they imply a struggle for life and death and therefore
overlap with other forms of struggle such as hunts or single combats.
In addition to the basic source-domains, spf-idioms profile a set of
archetypal image-schematic concepts such as containment, balance, integrity, force, etc. These conceptual archetypes underlie the conceptual
organisation of most of the basic source-domains and add further metaphorical coherence to spf-idioms. The underlying conceptual sources profiled by
spf-idioms are illustrated in Figure 5.2.
These source-domains constitute concrete activity and development
scenarios that are mapped onto more abstract target-scenarios of activity
and development according to the generic is specific-metaphor (Lakoff
& Turner 1989: 162–64; Lakoff 1993a: 233–235). Thus, spf-idioms reflect the
following general metaphorical models:
1.
2.
3.
4.
5.
6.
purposeful activity is a competition
purposeful activity is a struggle for life
development is life
dealing with an issue is eating
development is physical movement
development is the processing of a machine
Each of these metaphorical models constitutes a complex system, i.e. a conceptual network that consists of more specific subordinated metaphors that instantiate and elaborate it. The network further involves the low-level mappings
established between the ontologies and structures of the sources and targets of
the individual metaphors. Thus, each metaphor-system constitutes a network
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.6 (290-303)
 Idiomatic Creativity
Basic SPF-Source-Domains
COMPETITION
PHYSICAL
MOVEMENT IN
SPACE
STRUGGLE
LIFE
MACHINE
EATING
UNDERLYING
ARCHETYPEMETAPHORS
Figure 5.2 Overview of source-domains for spf-idioms
with conceptual connections between source and target domains and between
individual conceptual metaphors. Moreover, the motivation of specific spfidioms relative to these conceptual backgrounds is shaped by metonymic links,
emblematic associations, blending and complex interactions between them.
In what follows, these conceptual backgrounds will be sketched individually and their role for the semantic structuring of spf-idioms will be discussed relative to some selected examples. In the electronic appendix to this
book (www.idiomatic-creativity.ch) the conceptual motivation of 120 out of
the total set of 600 spf-idioms is analysed in detail. Moreover, these idioms
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.7 (303-353)
Chapter 5. The conceptual motivation of SPF-idioms 
are categorised according to the core idiom-transparency classes outlined in
Section 4.6.
.. competition-models
A great number of spf-idioms profile literal-scenes related to competition
scenarios such as races, football matches, games, combat sports (boxing and
wrestling) and so forth. The models are projected onto more abstract activities or states such as taking political decisions, dealing efficiently with a mental
task, being in a successful position, acting cleverly, etc. The motivation for these
source domains to conceptualise success and failure in more abstract scenarios
is obvious: ICMs for games, races, combat sports, etc. contain success and
failure as salient, immanent conceptual nodes. Therefore, they can be evoked
to provide easier cognitive access to and to structure the less prototypical and
less salient target-domains.
Game-related frames (ball sports, card games, gambling, board
games or golf) are powerful conceptual source domains to concretise successful or awkward actions in purposeful activities. Games are ultimately goaloriented: the players’ major objective is to win. Following the purposeful
activity is a game-metaphor, success and failure can be easily ontologised
by referring to those stages in a game where the players’ success or failure is
imminent. Moreover, the performance of games is characterised by strategic
considerations, skilful moves and decisive actions on the part of the players.
Consequently, the fine-grained knowledge about specific game-related skills
and strategies as well as advantageous or disadvantageous constellations can
be exploited to reflect similar forms of knowledge and behaviour in the target
activities.
For instance, the card game-script works as one subordinate sourcedomain in the inheritance hierarchy of the competition-model. To become
successful, each player has to make intelligent use of the cards at his or her
disposal (1). In the target-domain of purposeful activity, the cards (aces,
trump cards), which are in the hands of each player, thus correspond to the
figurative means that are controlled by different agents (2). Hence, the constituents card and ace have the autonomous semantic values idea, plan or
means in these idiomatic constructions. This renders them isomorphic. Accordingly, playing the cards corresponds to the agents’ skilful, secretive and
strategic use of the means at their disposal (3). Correspondingly, risky or dangerous activities with an insecure outcome can be concretised with reference
to gambling or betting as in (4). The ultimate goal of a card game and bets
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.8 (353-406)
 Idiomatic Creativity
is to win. With reference to this knowledge, the idioms in (5) denote win, be
successful and fail by referring to the winning or losing stage in a card game
or in gambling.
(1) play your cards right
(2) have an ace/card up your sleeve, hold all the aces
(3) play/keep your cards close to your chest, lay/put your cards on the table,
play your ace/trump card
(4) a good/safe bet, hedge your bets, have an eye for the main chance, raise/up
the ante
(5) sweep the board, hit the jackpot, have had your chips
Note that idioms such as raise the ante, have had one’s chips or sweep the board
can only be motivated if the speaker knows that the literal scene refers to respective scenarios. This said, the constructions in (1)–(5) illustrate that the
card game-frame structures a great number of idioms that concretise the
conscious performance of decisive actions to achieve success.
A considerable number of idioms from the spf word-field are based on
source domains that refer to ball games. As in (6), the ball game-script is
metaphorised to concretise joint activities that centre around a concrete issue.
Following the logic of this metaphorical model, this issue is ontologised by the
ball-concept; hence, ball has a corresponding phrase-induced figurative sense
for the idioms in (7). Accordingly, competent and successful acting is the effect
of dealing with the ball efficiently. The idioms in (8) highlight this knowledge
metaphtonymically in terms of state of being on the ball for the effect
of controlling it successfully and vision is control. In contrast, the
scene of losing the ball points to a stupid mistake that is a sign of incompetence
and looming failure.
(6) a new/different ball game
(7) set the ball rolling, take the ball and run with it, on the ball, juggle the
balls in the air
(8) keep your eye on the ball, take your eye off the ball, drop the ball
After having played a match, sportspeople usually take a bath or shower. However, due to injury, expulsion or substitution athletes sometimes have to quit
playing before the game is over. This knowledge is recruited by the idioms in
(9), which metaphtonymically correlate the event of having to go the shower
room with the idiomatic meaning ‘be forced to quit an activity’:
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.9 (406-466)
Chapter 5. The conceptual motivation of SPF-idioms
(9) take an early bath, send someone to the showers
(10) (bat) on a sticky wicket, (not) get to first base, have a good innings, below
par, par for the course
The ability to motivate these idioms depends on the user’s ability to relate the
profiled literal-scenes to the ball game-domain. If he or she cannot do so,
these idioms remain opaque. Similarly, the expressions in (10) are based on
specialist knowledge about cricket, baseball and golf. Therefore, the expressions are also likely to remain opaque for a considerable number of speakers.
Nevertheless, the ball game-ICM provides a strong motivating basis for a
considerable number of spf-idioms.
Like games, races are strongly institutionalised forms of competition.
Therefore, following the purposeful activity is a race-metaphor, they provide highly salient models to be mapped onto more abstract purposeful activities that involve some competitive performance under time pressure.
(11) can’t stand the pace, get up to speed, lead the field, be ahead of the pack,
be streets ahead, show a clean pair of heels
(12) be in pole position, get off the (starting) blocks, quick/first/slow off the
mark, have the inside track
(13) the home straight, first past the post, pip NP at the post
The purpose of a race is to measure whom of a group of competitors can cover
a given distance in the shortest time possible. In the activity target-domain,
the competence of an agent can therefore be ontologised in terms of a racer’s
speed or his position in the field. This conceptual knowledge underlies the motivation of the idioms in (11). The skeletal structure of the race-script further
involves a start, a finish line and the competitors’ contest along a path between
these two spatial reference points. Given its spatial basis, the metaphorical
model purposeful activity is a race interacts closely with the metaphor
system progress is forward movement in space (see below). The two models overlap to constitute the coherent composite metaphor fast progress in
a purposeful activity is fast forward movement in a race. spf-idioms
refer to this composite metaphor by profiling the different stages in the race.
For instance, the constructions in (12) concretise the successful start of an activity, while in (13) the final and decisive stages of the race correspond to its
final stages. Finally, (14) nicely illustrates how the race-ICM can be manipulated to create a precise, blended cognitive micro-model. In this idiom the
general metaphor purposeful activity is a race is intensified by replacing
the racer-slot by the emblem rat, which stands for disloyalty. The resulting

JB[v.20020404] Prn:20/02/2006; 14:56

F: HCP1705.tex / p.10 (466-515)
blended metaphorical conceptualisation of life as a rat race concretises the
folk view of modern lifestyle as a competitive but unfair and immoral struggle.
(14) get out of the rat race
The metaphor purposeful activity is boxing exemplifies the close association between aggressive competition in the daily struggle for life and its cultural transformation into an institutionalised sport. spf-idioms are motivated
by this metaphorical background in terms of the following correspondences:
before the fight, boxers try to impress their rivals by making fighting gestures.
The idiom in (15) adopts this image to denote determination. Further, the literal meanings of the constructions in (16) refer to scenes of in-fighting and
the boxers’ physical position or posture. Relying on the conceptual entailments
of these scenes, these idioms denote favourable or disadvantageous situations
figuratively.
(15) come out swinging/fighting
(16) box clever, in/out of a (tight) corner, on the ropes, paint/box NP into a
corner, keep your chin up
If critical periods cannot be overcome, the ringing of the bell, which terminates
each round in the match, can help the boxer to get out of this difficult situation
and to regain energy before the next round; this idea is highlighted by (17).
However, the boxer can also be knocked out. Metaphorically speaking, as in
(18), this means failure or giving up.
(17) saved by the bell
(18) down for the count, down and out, throw in the towel/sponge
.. struggle-for-life-models
A considerable number of spf-idioms are patterned by source domains that
refer to manifest aggression: hunts, fights, single combats and battles. These
source domains are closely connected to the competition-model: hunts and
fights are primordial forms of competition in a biological sense. In terms of this
survival of the fittest, the winner of this biological competition is successful. In
contrast, suffering from a defeat in the struggle for life often leads to death,
that is, to obvious failure. Following this line of reasoning, the struggle-forlife-model overlaps with the competition-models as well as the metaphor
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.11 (515-570)
development is flourishing life with its entailments failure is death and
success is life (see below).
A number of spf-idioms profile the metaphorical model purposeful activity is a hunt systematically. To obtain food, predatory animals – including
human beings – prey upon other animals. From the perspective of the predator,
success is obtained when the prey is caught, whereas, from the victim’s point of
view, a possible escape from the aggressor must be seen as a positive achievement. Corresponding scenes are profiled by the expressions in (19). Evoking
the specific fishing-frame (20) highlights the fact that the chances of achievement are reduced if an activity takes places in a unfavourable context. Finally,
(21) connects the purposeful activity is a chase/hunt-metaphor with the
model stagnation is circular movement (see below) to create a coherent
image of ineffective activity.
(19) the early bird catches the worm, kill two birds with one stone, scent/taste
blood, get off the hook
(20) fish in troubled waters
(21) chase your own tail
The primordial struggle-for-life between a predator and its prey is directly
related to the domain fight and its corresponding subframes.
Fights constitute a primitive form of achieving one’s aims at the cost of
another competitor. In single combats, wars and battles two opponents or opposed political or social groups face one another to settle their conflict on a
violent basis. spf-idioms exploit the knowledge associated with the congruent
battle and combat-scripts to model success and failure in activities such as
political conflicts or arguments. Thus, they are motivated by the purposeful
activity is a fight/battle-metaphor. A combat starts when the fighter confronts his opponent. Within the figurative target-activity, the opponent often
corresponds to a problem that has to be solved in order to become successful;
thus, the object-slot in (22) corresponds to a figurative problem. Following this
logic, the scenes of defeating the opponent in (23) are evoked to correspond to
tackling a problem or thwarting a competitor. Moreover, being on top of the
opponent brings the fighter close to success, whereas being dominated by the
opponent leads to failure. With the constructions in (24), this knowledge about
fights is recruited in combination with the metaphors difficulty is a force
(from above) and failure is down. Following the same metaphorical logic,
a lost battle corresponds to the failure of achieving one’s aims (25).
(22) come face to face with NP

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.12 (570-626)
 Idiomatic Creativity
(23) bring NP to its knees, to break the back of NP, beat the living daylights out
of NP, strike a blow for/against NP, a body blow, a death blow
(24) you can’t keep a good man/woman down, be on top of NP, have your back
to the wall
(25) fight a losing battle; win the battle, lose the war; lose the battle, win the
war
Within the battle-frame, weapons constitute the soldiers’ major means
for achieving success. In (26)–(30) this motivated association is exploited;
weapons are mapped onto the idiomatic target-domains to concretise the
means or advantages that support the agent’s efforts to fulfil a task. Alternatively, when the fighter uses his strength or weapons ineffectively (27), inappropriately (28) or if he risks too much (29), he is likely to fail. Moreover, when
the fighter cannot mobilise his full strength, he suffers from a disadvantage
which reduces his chances for success (30). Following these entailments, these
spf-idioms denote instances of appropriate or awkward behaviour or limited
freedom of acting:
(26) give NP your best shot, another string to your bow/many strings to your
bow, have NP on your side
(27) have shot your bolt, have only one shot in your locker
(28) hit and miss, blow up/explode in your face, shoot yourself in the foot,
make a rod for your own back, cut your own throat
(29) chance your arm, cost an arm and a leg, give your right arm
(30) with your hands tied behind your back, a chink in NP’s armour, a long
shot
Note that these idioms vary in analysability, depending on whether and
how the weapon-node is explicitly encoded in the corresponding constituent
structures.
In battle, success or failure is decided in the front line: through territorial wins the front line can be pushed forward. The party that is able to break
through the front line and the defense of their enemies is likely to succeed. A
last ditch may therefore constitute the only chance left to escape from failure.
Feeding on this knowledge, the idioms in (31), (32) and (33) describe deep
involvement in the most decisive actions of an activity, progress, complete success or failure (with waterloo being an emblem for total annihilation and
complete failure).
(31) in the front line, gain/lose ground, make up lost ground
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.13 (626-679)
(32) be in at the kill, move in for the kill, take somewhere by storm, meet your
Waterloo, be shot to pieces
(33) a last ditch (attempt, effort)
Being a matter of life and death these battle and combat models are directly
linked to the following life-models.
.. life-models
The development of an organism constitutes a very basic metaphorical source
to understand the temporal development of abstract activities such as policies, careers or economic parameters. The metaphorical model development
is flourishing captures the progressing of a given activity by analogy with
an organism’s growth: while vitality ontologises success, failure corresponds to
the death of an organism. These correspondences are exploited by the idioms in
(34) and (35). The life-model thus interacts with the purposeful activity is
a struggle for life-metaphor; in both models success is vitality or survival, while failure is death. Following the logic of the metaphorical model,
the idioms in (36) characterise taking too much risk in an activity in terms of
causing one’s own death. Some idioms, as in (37), more specifically describe the
process of failing as drowning. This image is motivated on the basis of two additional conceptual metaphors: failure is down and instability/insecurity
is water.
(34) be alive and kicking, come back/rise from the dead
(35) a dead duck, flog a dead horse, someone’s days are numbered, the death
knell sounds for NP
(36) dig your own grave, put/stick your head in a noose, put your head/neck
on the block
(37) a drowning man will clutch at a straw, keep your head above water, sink
or swim, be dead in the water
The development of a configuration is life-metaphor entails the
conceptualisation of progress as growth. As Hundt (1995) illustrates, this
metaphor is deeply entrenched in economic terminology. As in (38), idioms
feed on the model by profiling various scenes that refer to the growth of
plants, animals or human beings. In accordance with this submodel, reward
is conceived in terms of the metaphor reward is harvest, which underlies all
constructions in (39). Exploiting this metaphorical model, the proverb in (40)

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.14 (679-733)
 Idiomatic Creativity
feeds on knowledge about the invisible growth of a chicken in the egg to convey
the wisdom that the success of an activity is not guaranteed before its outcome
is certain.
(38) growing pains, teething problems/troubles, sow the seeds of NP, put down
roots, nip NP in the bud
(39) reap the harvest, bear fruit, earn your corn
(40) don’t count your chickens before they are hatched
.. consumption-models
The overarching survival-scenario constitutes the motivating basis for another model within this complex conceptual network: dealing with an issue
(activity) is consumption. This metaphor exploits the experiential fact that
an organism’s survival is dependent on nourishment. Following this metaphorical ICM, an agent’s involvement with an activity or issue is concretised by the
amount of food being consumed by the eater (41):
(41) (want) a piece/slice of the action, bite off more than one can chew,
get/have a second bite at the cherry
In contrast, one’s chances of dealing with an issue or activity successfully
are considerably restricted when one’s resources for action or the issues to be
dealt with are negatively affected. In the source-domain this corresponds to
one’s food being spoiled or wasted. These correspondences are profiled by the
idioms in (42). Note that kill the goose that lays the golden eggs and kill the golden
goose contain the emblem gold for value/profit.
(42) pull NP’s chestnuts out of the fire, your goose is cooked, kill the goose that
lays the golden eggs/kill the golden goose, too many cooks spoil the broth
While good food is favourable to the organism, bad or poisonous substances threaten its existence or they are unpleasant to consume. With the
idioms in (43), these entailments correspond to dealing with dangerous or
unpleasant situations or issues on the idiomatic level; correspondingly, the
metaphor good/bad food is a positive/negative issue to deal with renders these expressions analysable. Moreover, if good food is no longer obtainable, the organism’s life quality diminishes and thus forces it to move to a more
favourable and rewarding environment. Metaphorically speaking, this means
moving to a better situation or dealing with more pleasant issues (44).
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.15 (733-787)
(43) a poisoned chalice, swallow a bitter pill
(44) seek greener pastures, move to pastures new
In sum, the food-model provides metaphorical coherence to a considerable number of idiomatic constructions. The model is connected with the life
and struggle-models to constitute a coherent world-view: purposeful activities are seen as a form of natural struggle for life. In this struggle, the agent is
forced to guarantee his own development and well-being by obtaining food and
power. By enforcing their will-to-life, agents enter into competition with other
subjects who are driven by the same interest. According to this world-view, this
often results in conflict situations whose positive or negative outcome determines the agent’s success or failure. In its most extreme form, the struggle for
successful survival may lead to death. Conversely, the successful achievement
of one’s life aims is regarded as the basis for flourishing life. In other words,
spf-idioms that are motivated by the previously discussed models constitute
a coherent semantic system, rather than an unstructured assembly of opaque
and isolated constructions.
.. progress as forward movement in space
In accordance with Lakoff ’s analysis of the event-structure metaphor (Lakoff
1993a: 220–222), the idea of progress is essentially construed as forward
movement in space. Expressions such as the company has gone fast forward
or since the last tournament, he has been stepping backwards indicate that this
model plays a fundamental role in our evaluation of the development of activities and events. Indeed, it is virtually impossible to conceptualise development or change without reference to spatial orientation. This is also reflected in my abstract definition of progress above, which depends crucially
on phrases such as approach an aim or develop towards an aim. Many spfidioms encapsulate and convey this metaphorical perspective. In the following
sketch, I will, however, not analyse every metaphorical subsystem individually
(e.g. development is a journey by car vs. development is a voyage vs.
development is swimming vs. development is walking, etc.), but merely
classify the constructions in terms of how they feed on the overall system.
To become successful, the development of a configuration must first be
started. This involves leaving the starting point as in (45). However, the start
of an activity can be difficult and lead to failure because the chosen method
or direction of development is wrong or too ambitious. Therefore, it is often

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.16 (787-847)
 Idiomatic Creativity
advisable to start slowly. These conceptual relationships are concretised by the
idioms in (46) and (47), respectively.
(45) off/out of the (starting) blocks, get the show on the road
(46) in at the deep end, get off on the wrong foot
(47) dip your toe in the water, try your wings
The success of an activity further depends on the path that is chosen to
reach one’s destination. The source-concept path structures two distinct dimensions in the target-domain of development: a) the direction of development and b) the quality of the circumstances in which a given configuration develops. Thus, a path that leads to the desired destination is favourable
to the successful development of the configuration, otherwise inefficient or
counterproductive as highlighted by (48). Following the logic of this metaphorisation, the idioms under (49) evoke the agent’s straying from the right path
to concretise a cause for failure, while staying on the path implies successful development. Alternatively, a new path must be found to increase one’s chances
of favourable development (50).
(48) be on course for NP, on/along the right lines, on the right/wrong track, in
a rut, lead to a dead end, a blind alley
(49) jump off the rails, fall by the wayside, stay on the rails, stay the course
(50) turn the corner, break a path, blaze a trail
To concretise the positive or adverse circumstances in which an activity takes
place, the quality of the path is highlighted. In the metaphorical world, the
smooth constitution of the path stands for a context or situation which is
favourable to progress and success. In contrast, the idioms in (51) convey the
idea that development is impeded and failure likely to occur if the path is
difficult, dangerous or unpleasant to pass. Consequently, it is sometimes necessary to make preparations to guarantee success. In the source-domain this
corresponds to enhancing the quality of an existing path as pointed out by (52).
(51) up the creek (without a paddle), up shit creek, skate on thin ice, a
rough/bumpy ride, walk a tightrope, be on a knife-edge
(52) pave the way, prepare the ground
Movement along the path towards the destination implies approaching
one’s goal. Accordingly, the expressions in (53) portray progress by referring
to the idea that the more distance is covered, the closer success becomes even
if special efforts have to be made. If movers are very fast and competitive (see
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.17 (847-908)
Chapter 5. The conceptual motivation of SPF-idioms 
purposeful activity is a race), they can achieve their aims long before others. Conversely, development stagnates or regresses if the agent is not moving
forward as in (54). Ineffective development is also expressed by the idioms
under (55), which evoke images of circular movement.
(53) make headway, gain ground, make up lost ground, go the (full) distance,
go the extra mile
(54) a hard act to follow, be streets ahead, tread water, (not) get to first base,
lose ground, stop NP dead in its tracks
(55) go around in circles, chase your own tail, back to square one
For human beings walking is the most natural form of movement. A considerable number of somatic idioms make reference to this source domain by
profiling concepts of the leg-domain and its subdomains (leg, foot, heel).
More specifically, these idioms are based on the metonymy leg/foot for
ability to walk. In combination with the progress is walking forwardmetaphor, idioms that incorporate the resulting metaphtonymy conceptualise
an agent’s ability to progress and be successful by describing the constitution of the limbs (56) or coordinated movement (57). Accordingly, failing is
understood as stumbling, falling or uncoordinated movement (58).
(56) have legs, not have a leg to stand on, shoot yourself in the foot
(57) start off on the right/wrong foot, (not) put a foot wrong, hit your stride
(58) put NP off their stride, fall flat (on your face), drag your feet/heels, run
before you can walk
A traveller depends on footgear or can make use of a vehicle to support his or her getting on. Some idioms that encapsulate the development
is movement-metaphor make explicit reference to vehicle-domains such as
car, ship, cart or bus, depending on the concrete journey-frame underlying
a given idiomatic construction. The vehicle-slot is mapped onto conceptual elements in the target-domain that are conceived as supporting successful progressing. Such elements include efficient means or methods, help or
favourable situations. The resulting metaphor supporting basis of development is a vehicle thus concretises the idea that the development of a
configuration becomes more efficient if it is aided by favourable external influences. In terms of this metaphorical link, boarding or not boarding the vehicle,
respectively, correlates with taking or not taking an opportunity as in (59).
Once the vehicle is in motion, plenty of distance can be covered in a short
time and thus progress is guaranteed, whereas successful development is im-
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.18 (908-980)
 Idiomatic Creativity
peded when the vehicle stops moving properly or effectively as highlighted by
the idioms in (60). Therefore, it is necessary to adapt the vehicle to the external
conditions, rather than to spoil one’s chances of success by taking inappropriate or inefficient measures (61). Accordingly, unfavourable vehicles concretise
inefficient means or methods, problems or bad developmental situations. The
vehicle must be abandoned if it does not seem to lead to success (62).
(59) miss the boat/bus, catch the wave, on the crest of a wave, jump on the
bandwagon, in the same boat, step into NP’s shoes/boots
(60) set the wheels in motion, on a roll, the wheels are turning, on an even keel,
spin your wheels, run on empty
(61) trim your sails, rock the boat, upset the applecart
(62) jump ship, abandon a sinking ship, like a rat leaving a sinking ship, hang
up your boots
Apart from the bad constitution of the path, uncoordinated movement or
the inefficient use of one’s vehicle, the success of a journey can also be thwarted
by obstacles, obstructions and dangers that block the direct way to one’s destination. Obstacle concepts are therefore mapped onto the target domain to
ontologise problems. The idiomatic constructions under (63) profile such obstacles in their literal scenes. Following the mapping problems are obstacles,
many of these expressions therefore become isomorphic.
(63) a stumbling block, have a mountain to climb, a hard row to hoe, be over
the hump, hit the wall, hit the buffers, bang your head against a brick wall
Finally, voyages are notoriously hazardous journeys. On the path to its destination, a ship can be affected by various dangers like storms, reefs or shallows.
Further, the conceptual identification of water with instability/insecurity
turns the crossing of an ocean into a risky endeavour. This knowledge is encapsulated by the following idioms, which characterise problems or difficult
situations as potential dangers occurring on a sea trip:
(64) on the rocks, weather/ride out the storm, the tip of the iceberg, sail close
to the wind, in the doldrums
In sum, the analysis of the metaphor successful development/progress
is forward movement has revealed a highly systematic network of conceptual
correspondences between source-domain scenarios and the conceptual target
domain success, progress and failure. As a consequence, idioms profiling
this metaphor system are highly motivated in most cases.
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.19 (980-1020)
.. machine-metaphors
machinery serves as a further, but less extensive source domain to characterise
spf-related concepts. When working normally, machines are important tools to
facilitate work. This implies a direct correlation between the efficient processing of a machine and the increased production of a given product. This causal
relationship serves as the motivating ground for the metaphorical model development of a configuration is the smooth processing of a machine.
Following this metaphor, starting the development of a configuration corresponds to starting the machine as in (65). Accordingly, a target-configuration
is seen as progressing if the machine works optimally (66). In contrast, the
idioms in (67) model stagnation or failure in terms of the breakdown of the
apparatus or its ineffective processing due to the lack of fuel or other factors.
(65) set the wheels in motion, get into gear, pick up steam
(66) go/be in overdrive, up and running, the wheels are turning
(67) run out of steam, spin your wheels, go up in smoke
When the product is being processed, this corresponds to coming closer to the
achieved aim as in (68). Effective processing can be supported or enhanced by
careful servicing and maintenance. With the idioms under (69) this relationship is mapped onto the target domain to characterise supporting influences
on development. In contrast, the expressions in (70) highlight that efficient
processing is impeded when the machine is stopped by some obstruction:
(68) in the works, in the pipeline
(69) prime the pump, bring NP up to speed, oil/grease the wheels
(70) put a dampener on NP, throw/put a spanner in the works, put a spoke in
NP’s wheel
Note that the machine-model overlaps with the previous model in terms of
evoking machine-powered vehicles as models for progress.
.. Archetype-metaphor models
The metaphorical models discussed in the previous paragraphs are framebased and script-based extensions of rich conceptual knowledge. However,
in accordance with the global architecture of cognitive structures described
in Section 3.2, very basic, image-schematic archetype-concepts can also serve
as a meanigful resource to make structurally poor target-domains concrete

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.20 (1020-1070)
 Idiomatic Creativity
(cf. Lakoff & Johnson 1980: Chapter 4). I will call such fundamental imageschematic metaphors archetype metaphors. Indeed, archetype metaphors create most of the structure that underlies our metaphorical understanding of
success, progress and failure.
A number of archetype metaphors construct evaluative concepts such as
good/bad or positive/negative in terms of basic bodily experiences like
spatial orientation and bodily equilibrium. These experiences are structured
by fundamental image-schematic concepts such as verticality, balance or
brightness. These conceptual archetypes underlie and shape the motivation
of a great number of spf-idioms.
The evaluation of processes, things and actions with reference to up/down
orientation applies to spf-idioms in terms of two metaphorical models: success is up, failure is down and high status is up, low status is down.
Following the principle of metaphor-composition, these archetype metaphors
are combined with all the metaphorical models discussed above. For instance,
the combination of success is up and high status is up with the metaphor
development is movement along a path towards a destination leads to
the understanding of successful development in terms upward movement; the
idioms in (71) encapsulate this composite metaphorical conceptualisation. Following this complex conceptualisation, the constructions in (72) coherently
characterise ambitious aims and high expectations as very high landmarks that
may not even be reached.
(71) move up in the world, move up the greasy pole, get in on the ground floor,
be on the up (and up), up and coming
(72) the top of the tree, hitch your wagon to a star, reach for the stars/sky/
moon, the sky’s the limit
However, as the idioms in (73) indicate, upward movement also involves danger, particularly if the landmark is not stable. Thus, the conceptualisation of
success as being high-up intrinsically entails the idea of latently lurking failure:
if one is up, one can easily fall down. Following this experiential logic, failing
is metaphorically equated with moving or falling down (74). As a result, the
developing configuration is in a bad situation and thus finds itself at a very low
position relative to others (75).
(73) be up a gum tree, the dizzy/dizzying heights
(74) come down in the world, down the drain, come down to earth (with a
bump), be heading for a fall, fall flat (on your face), go into freefall
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.21 (1070-1136)
Chapter 5. The conceptual motivation of SPF-idioms 
(75) bump along the bottom, plumb the depths, between the devil and the deep
blue sea, hit/reach rock bottom
In bodily experience, the sensorimotor sensation of equilibrium and disequilibrium is naturally correlated with the success or failure of a given activity.
Given its high relevance, the archetypal image-schema balance serves as a fundamental experiential basis for the archetype metaphor positive/negative
development is balance/imbalance. This model is encapsulated by the
idioms in (76).
(76) (hang/ remain) in the balance, throw off balance, walk a tightrope, upset
the applecart, rock the boat
The balance-metaphor is very closely linked to the development is movement-model. When walking, one’s physical balance is directly influenced by
the constitution of the terrain on which one moves. This sensorimotor correlation (partially) motivates the following idiomatic constructions in terms of
the metaphorical mappings: security/stability is solid ground (77); insecurity/instability is water (78); instability/uncertainty is air (79).
(77) keep your feet on the ground, skate on thin ice
(78) keep your head above water, dip your toe in the water, be in deep water, in
at the deep end, a drowning man will clutch at a straw
(79) keep/juggle the balls in the air, have your head in the clouds
Finally, the evaluative correlation of stability with positive constitution
or positive development is also experienced in the interaction with physical objects or substances. Objects that are easily broken cannot be handled,
manipulated and controlled and are therefore evaluated as unfavourable. This
experiential correlation motivates the metaphor negative development is
disintegration, which underlies the idioms in (80).
(80) make bricks without straw, burst/prick the bubble, that’s the way the
cookie crumbles, come/fall apart at the seams, fray at the edges
Starting from very early childhood, human beings have a great number
of experiences with basic objects and substances. This primordial knowledge
underlies a number of ontological metaphors that comprise knowledge about
containers and forces. To concretise abstract conceptual categories such as
state, situation or problem, a number of archetype metaphors feed on the
image-schematic structures we derive from these direct physical experiences.
For instance, container-schemas provide very basic cognitive models to un-
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.22 (1136-1186)
 Idiomatic Creativity
derstand abstract states and situations (cf. Johnson 1987: 22). Following the
metaphor involvement in a state or situation is being in a container,
acting in positive or negative circumstances is conceived as being ‘contained’
in an unfavourable or dangerous container or location; this idea underlies the
idioms in (81). The deeper one is confined within the container, the more difficult one’s situation becomes (82). In Christian culture, hell is the epitome of
a bad place or situation. To be caught in this container is therefore emblematic
for great difficulties or complete failure; this conceptualisation motivates the
idioms under (83).
(81) out of/in the doldrums, in the dumps, in a hole/out of a hole, walk into
the lion’s den
(82) be in over your head, in the shit, in deep shit
(83) go through hell, the road to hell is paved with good intentions
As the container-schema, our interactions with different types of forces
or energy states serve as directly meaningful source domains to make empiricallyabstract difficulties, problems and abstract forms of control comprehensible.
Consider, for instance, the metaphor intensity of activity is heat. Following this metaphor, the amount of heat attributed to a process stands for its
degree of activity and positive development. Something that is cold is inactive and thus unsuccessful, as encapsulated by (84), but once it regains energy
and becomes hotter, success is coming closer again as in (85). However, everyday experience also tells us that fire is not only an energy resource, but a
destructive force that should be handled with care. The concept of fire therefore ontologises very intensive difficulties and problems or negative situations.
The analysable idioms under (86) evoke the fire-concept in this sense:
(84) off the boil, on the back burner, be left out in the cold
(85) on the boil, keep the pot boiling, come in from the cold, have a lot of irons
in the fire
(86) pull someone’s chestnuts out of the fire, fan the flames, out of the frying
pan into the fire, add fuel to the fire/flames
Experience with mechanical forces provides a further background for the characterisation of difficulties and problems. For instance, problems and difficulties
are conceived as weights that press the experiencer to the ground or a lurking danger that hangs over the agent’s head. This is in accordance with the
evaluative archetype metaphor bad is down. The idioms in (87) are rendered
analysable by this problems/impediments are oppressive forces-metaphor.
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.23 (1186-1243)
Chapter 5. The conceptual motivation of SPF-idioms 
Alternatively, the force is seen as a form of pressure that is crushing the experiencer (88).
(87) a cross to bear, the straw that breaks the camels back, a millstone around
your neck, an axe hanging over NP
(88) go through the mill, go through the wringer
Establishing counter-images to the conceptualisation of problems as obstacles,
weights and pressure, the idioms in (89) point to efforts involved in the process
of achieving one’s aims, i.e. to overcome these forces:
(89) move heaven and earth, leave no stone unturned, get NP off the ground,
pull out all the stops
The archetypal metaphorisation of problems as forces is coherently connected
to the global success and failure scenarios outlined above. Thus, in the journey-frame problems and difficulties appear as obstacles, obstructions and
storms (90). In the machine-scenario, problems are equivalently seen as a form
of blockage that stops the processing (91). Moreover, in the life, hunt and
battle-scenarios, problems and difficulties are embodied in terms of dangerous animals or plants, enemies and weapons (92) that threaten one’s life, i.e.
success.
(90) a stumbling block, the tip of the iceberg, weather/ride out the storm, have
a mountain to climb
(91) throw/put a spanner in the works, put a spoke in NP’s wheel
(92) take the bull by the horns, stir up a hornet’s nest, grasp the nettle, come
face to face with NP, sit on a powder keg
In contrast, advantages, favourable means, help and support are seen as positive forces that help the agents to move on, to defeat their opponents or competitors, or to overcome obstacles. These conceptual correspondences create
isomorphic structures for the respective idioms.
The idea of possessing tools or weapons is closely linked to the metaphor
control is physical grasp, as implied in (93). These idioms refer to the
control of advantages or favourable means, or they characterise the tackling
of problems through controlled physical manipulation.
(93) the ace in your hand, hold all the aces, grasp the nettle, take the bull by the
horns
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.24 (1243-1294)
 Idiomatic Creativity
On balance, many spf-idioms are structured by very basic archetypal
metaphors that have generally been ignored in the semantic analysis of these
expressions. (See our discussion of the Keysar and Bly study on metaphorical
motivation in Section 2.3.3). With a view to the great semantic systematicity
of the idioms that are motivated by these conceptual structures, this is a severe shortcoming, which demands for further linguistic and psycholinguistic
investigation.
Not all motivated spf-idioms can be related to extensive metaphor-systems
to motivate their internal semantic structure. As has been illustrated in Section
4.5.1, idioms may also be motivated by one-shot and image-metaphors. The
idioms in (94) should suffice to illustrate this for spf-idioms. Both expressions
are based on the concept of drama.
(94) the curtain comes down on NP, the plot thickens
The first idiom refers to the end of an activity in terms of the end of a theatre
performance by highlighting the moment when the stage curtain comes down.
Since this event marks the end of a very lively activity, the idiom can denote
a person’s failure as well as death. The second construction can be related to
the development is movement-model with reference to the problems are
obstacles-metaphor. In accordance with this idea, an image of density or impenetrability is evoked by the verb thickens. However, the image created by this
idiom is drama-specific; as a result, it is only distantly related to the more
general metaphor system of progress as forward movement.
. Metonymic and metaphtonymyic motivation
Metonymy also plays an important role for the motivation of spf-idioms.
Above (Section 5.2.5), we have already highlighted the role of leg-related
body-part metonymies in connection with the progress is forward movement-metaphor. Thus, metonymy appears predominantly in combination
with conceptual metaphor to create metaphtonymic motivation in terms
of ‘metonymy within metaphor’ (see Section 3.3.5). In addition, part for
whole-shifts appear with idioms such as (95):
(95) set the wheels in motion, the wheels are turning
In these constructions, the concept wheel(s) is highlighted as a part to refer
to the whole: the vehicle. This part for whole-shift appears in combination with the metaphor development/progress is forward movement
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.25 (1294-1347)
Chapter 5. The conceptual motivation of SPF-idioms 
and applies to the correspondence the developing configuration is a
mover (here the vehicle). Thus, the event of setting the wheels in motion
metonymically points to the fact that the vehicle starts to move. In terms of
the metaphtonymic projection this corresponds to starting the development of
some configuration or activity. Accordingly, when the wheels are turning, the
configuration or activity is developing.
Further, a great number of idioms with the syntactic structure of locative
PPs such as (96) refer to developmental states figuratively.
(96) (get the show) on the road, in the pipeline
Above, we have classified most of these constructions according to the
metaphor states and situations are containers/locations. However, for
these specific idioms, the underlying conceptual motivation is more appropriately explained in terms of metaphtonymy. The locations profiled by the PPs
either describe the mover’s state of being on the path or they point to specific
locations or stages on the route. Thus, these idioms metonymically highlight
the state of being on the path for the state of being in movement. In
combination with the metaphor development is movement the PPs on the
road and in the pipeline thus describe a given configuration as being in a state
of progress or development.
Moreover, a great number of spf-idioms convey metonymically highlighted patterns of causation. These idioms usually profile a physical cause that
leads to a predictable effect:
(97) hit the wall, hit the buffers, grease/oil the wheels, throw a spanner in the
works, cut your own throat
These cause for effect-shifts work in combination with the conceptual
metaphors outlined above. For instance, when a vehicle or train hits the wall
or the buffers, respectively, it is caused to stop. In contrast, when the vehicle’s
wheels are greased it is prepared to move faster and more efficiently. In combination with the conceptual metaphor progress is forward movement, these
cause for effect-patterns are projected onto the spf-target domain: to be
stopped in one’s movement means to stagnate and fail, whereas being supported to move more efficiently means to progress successfully. As has been
discussed in Section 4.4.6, the presence of such metonymic shifts exerts a strong
influence on idiom analysability. Thus, although the wall and the buffers in
(97) can be readily interpreted as obstructions that stop the agent’s forward
movement, these profiled conceptual nodes do not receive a corresponding figurative interpretation in the idiomatic meaning. Rather, the underlying cause
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.26 (1347-1403)
 Idiomatic Creativity
for effect-metonymy only maps the implied effect of the literal scene onto
the idiomatic meaning. Since hit the buffers metonymically points to the holistic effect stop, the metaphorisation of this effect according to the progress is
moving forward-metaphor leads to the idiomatic meaning fail, rather than
to the analysable structure face a problem/interference. As a consequence,
the idiom has the status of a motivated but non-isomorphic semantic unit.
This provides further evidence for the influence of metonymic shifts on the
semantic structure of an idiom.
. Motivation by emblems
In spf-idioms emblems play an important role for the evaluation of purposeful
activities or the quality of a situation. A number of spf-idioms contain animal
emblems. Above, we have already discussed the use of rat to refer to disloyalty or deception. Moreover, the emblem dog appears in the constructions
under (98). Following Dobrovol’skij and Piirainen (1997: Ch. 4.1.5.1), this emblem can be attributed the semantic values aggressivity and inferiority.
Accordingly, the first of these idioms can be motivated in terms of the struggle-model, while the second and third involve the association of failure with
low social status.
(98) dog-eat-dog, every dog has its day, go to the dogs
wolf conveys the semantic value evil. In (99) this emblem interacts with the
metaphor being in a positive situation is being sheltered by a building
to create an image of defence from a lurking threat, i.e. from failure.
(99) keep the wolf from the door
Bad qualities are further alluded to by distasteful substances. Thus, shit in
(100) emblematically highlights the negativity of the target situation, which is
enforced by the disgusting scenes that these idioms convey:
(100) the shit hits the fan, in the shit/deep shit, up shit creek
Further, the concepts of devil and hell are strong cultural emblems to ontologise evil. In spf-idioms, these emblems are evoked in combination with the
metaphor being in a bad situation is being in an unfavourable location
to characterise highly unfavourable constellations as in (101).
(101) between the devil and the deep blue sea, go through hell, hell on earth
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.27 (1403-1453)
Chapter 5. The conceptual motivation of SPF-idioms 
In contrast, idioms pointing to success and positive situations contain highly
positive emblems such as rose or gold. The weak cultural emblem rose in
(102) signifies positive hopes and success. Finally, gold is a very strong
cultural emblem to symbolise great wealth and value (103):
(102) come up smelling of roses; coming up roses
(103) kill the goose that lays the golden eggs, not be the pot of gold at the end of
the rainbow, strike gold
. Opaque and constructionally-idiosyncratic spf-idioms
There is a considerable number of spf-idioms that cannot be motivated because their literal scenes cannot be related to a domain of knowledge that
would make sense relative to the idiomatic meaning. These constructions, as
the ones in (104), belong to the class of opaque idioms with a compositional
literal meaning:
(104) bite the dust, take a bath, drop your bundle, carry all before you, give up
the ghost, do a job on NP, get the wooden spoon, be up the spout, not cut
the mustard, leave NP high and dry, have NP over a barrel
A specific subgroup of opaque spf-idioms contains names that originate
from Greek mythology. Since average speakers of English cannot be expected
to know these myths, these idioms must be regarded as opaque (105):
(105) open a Pandora’s box, halcyon days
Along similar lines, the idioms in (106) make reference to Greek and Christian myths. However, unlike the former constructions, these idioms cannot
be regarded as completely unmotivated. While the term Gordian knot cannot be motivated without the mythological background, the scene of cutting
a knot makes sense relative to the idiomatic meaning ‘overcome a problem’:
the scene can be interpreted in terms of the conceptual metaphor solving a
problem is overcoming an obstruction by the help of a tool. In a similar vein, the nominal sword of Damocles is non-transparent, but the lurking
failure alluded to by the idiomatic meaning can be readily related to the image of a sword hanging over one’s head in terms of the conceptual metaphor
difficulty is an oppressing force from above. Finally, it is fairly obvious
how the concept lion’s den can be mapped onto the idiomatic meaning dangerous/difficult situation, although it may not be obvious what role the
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.28 (1453-1511)
 Idiomatic Creativity
character of Daniel plays in this constellation. Again, this idiom must be seen
as partially motivated.
(106) cut the Gordian knot, have the sword of Damocles hanging over one’s
head, Daniel in the lion’s den
These idioms thus nicely illustrate that, in many cases, idiom motivation may
be partial. From a cognitive-linguistic perspective, such gradual differences in
motivation can be easily handled in terms of the number of conceptual correspondences that a speaker is able to establish between the literal and the
idiomatic meaning.
Discussing the class of opaque idioms, we must also reconsider a number
of spf-idioms that we have classified as motivated by conceptual metaphor.
For instance, the constructions in (107) have been accounted for in terms of
the model purposeful activity is boxing. Although these idioms can indeed be motivated by this metaphor, their degree of transparency is primarily
dependent on a given speaker’s ability to relate the profiled literal scenes to the
boxing-frame. If a given speaker cannot contextualise the literal scene, the conceptual metaphor cannot be activated, which renders these expressions opaque.
The same is true for the expressions in (108) which are based on the football
match and race-frames.
(107) throw in the towel/sponge, saved by the bell, on the ropes
(108) take an early bath, send sb. to the showers, pip someone at the post
In addition, the activation of the motivating metaphor-system can be
blocked by the presence of garden-path constituents (see Section 4.6). For instance, the spf-idioms in (109) can only be motivated if speakers activate those
subsenses of the nouns colours, board and chips which allow them to evoke the
motivating frames that underlie the conceptual metaphors discussed above:
colours = battle flag, board = game board and chips = gambling counter.
(109) with flying colours, sweep the board; have had one’s chips, when the chips
are down
Consequently, it must be assumed that for a couple of idioms that we have
semantically classified in terms of conceptual metaphor, speakers are not generality able to retrieve the corresponding source-domains on the basis of the idioms’ constituent-structure. If this is the case, the expressions remain opaque.
Thus, it cannot be assumed that all speakers can equally penetrate all idioms
relative to the system of conceptual metaphors. This is in accordance with the
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.29 (1511-1558)
cognitive-linguistic assumption that the activation-set triggered by a given idiomatic construction may be slightly different from idiom to idiom and from
speaker to speaker. This, however, is not to say that the metaphor systems
outlined above remain generally blocked to the mental grasp of speakers. In
contrast, the great amount of semantic coherence reflected by these systems
suggests that most corresponding idioms can be motivated by them – otherwise, it would not be possible to uncover or comprehend their systematicity in
the absence of etymological explanations.
All types of constructional idiosyncrasy described in Section 4.6 can also be
observed with spf-idioms. Some spf-idioms contain cranberry morphs, which
render their literal meaning schematic as in (110). Consequently, the semantic
structure of these idioms becomes non-transparent:
(110) put the kibosh on NP, put the mockers on NP, go for broke, come a
cropper
Another subset of spf-idioms reflects syntactic idiosyncracies (111). Note that
most of these constructions could potentially be related to the basic source
scenarios described above, because they contain constituents such a trumps,
dice, guns, etc. Nevertheless, these idioms must be regarded as opaque, because
they do not establish concrete literal scenes that could be made sense of on
the basis of the metaphor-models. Finally, the idiom in (112) has an odd literal
meaning. The reason for this idiosyncrasy is unclear from a cognitive-linguistic
point of view.
(111) turn/come up trumps, no dice, come within an ace of something, go great
guns, go places
(112) a Catch 22
. Implications for the psycholinguistic controversy about
metaphorical motivation
In conclusion, the results of this analysis can be revisited with a view to the
controversy about the influence of conceptual metaphor on idiom-motivation
(Section 2.3.3). The analysis of spf-idioms in terms of conceptual metaphor
and other patterns of semantic extension has revealed that a great number of
them can be classified according to six global metaphor systems and a set of underlying archetype-metaphors. The metaphor systems select the global source

JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.30 (1558-1605)
 Idiomatic Creativity
scenarios competition, struggle, life, eating, movement and machine.
These source domains are projected onto the abstract target domains purposeful activity and development to concretise, conceptualise and structure
them. The scenarios are not isolated but show considerable overlap. Therefore, they can be claimed to convey a coherent, global world-view of success,
progress and failure. It is from this metaphorical world-view that many spfidioms derive their cognitive function as pre-coded micro-models.
The metaphorical world-view is not created on the basis of only one
consistent ‘image.’ Rather, the metaphor systems are coherently patterned by
a number of different subframes. For instance, the development is movement-metaphor is instantiated by elaborations that establish mappings from
concrete frames and scripts, such as walking, car trip, swimming, voyage,
onto abstract activities and forms of development such as learning, making
a career, living a relationship, etc.
Moreover, the specific ICMs encoded by spf-idioms show considerable
variation along the dimensions of culture-specific experience vs. universal
experience. For instance, the source-domains baseball and cricket are restricted to an Anglo-American cultural background. The domain voyage is
less culture-specific. However, the literal scenes conveyed by the corresponding
idioms mainly point to sailing-scripts that are historically rooted in the age of
18th and 19th century seafaring. In contrast, experience related to the domain
walking is universal and experientially basic. Similarly, the struggle-model
both highlights relatively general experiences related to fights, single combats
and hunts but also includes more specific scenarios based on 19th and 20th
century warfare.
The relationship between frame-specific motivation and motivation on
the basis of general bodily experience is particularly striking with archetypemetaphors. A great number of idioms that are structured by archetype
metaphors select relatively isolated and specific scenes. On the level of the concrete conceptual knowledge associated with these scenes, these idioms seem
to be highly distinct. For instance, the idioms go down the drain, sink or
swim and fall off your perch have relatively little in common on their conceptual surface. However, the underlying image-schematic structure, which
highlights non-intentional downward movement, involves highly general
sensorimotor and force-dynamic representations. The same is true for different images of disintegration fall apart at the seams, make bricks without straw
or that’s the way the cookie crumbles that involve the failure is disintegration-metaphor. In other words, on the level of primordial image-schematic
structure (development as movement, stability as balance, etc.) these id-
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.31 (1605-1664)
Chapter 5. The conceptual motivation of SPF-idioms 
ioms are highly coherent and motivated by the same archetypal experiences,
whereas they are relatively inconsistent on the level of the specific scenes that
they encode.
With a view to Burger’s point of criticism that Lakoff ’s conceptualmetaphor-approach to idioms often violates the historical and cultural specificity of the idiomatic data and, in addition, neglects the specificity of the
individual metaphorical transfers established by each single idiom, one can
thus react as follows: if viewed in isolation, many individual spf-idioms establish relatively specific metaphorical transfers. This is true for both the cultural
and the historical basis of the source-domain knowledge. However, these transfers are not unorganised. Rather, they can be related to a more general system
of conceptual metaphors. While I am not claiming that this system is universal,
it can neither be regarded as idiosyncratic; it serves as a well-organised conceptual background against which the meaning of spf-idioms can be structured.
Conceptual metaphors thus play a decisive role for the motivation of idioms.
The very systematicity of spf-idioms indicates that conceptual metaphors play
an important structuring function. This, clearly, also speaks against the Glucksbergian school of metaphor and idiom analysis, which does not see a place for
conceptual metaphor in the analysis of idiomaticity (cf. Glucksberg 2001).
As regards the scope of idiom motivation on the basis of conceptual
metaphor, spf-idioms have revealed the following trends. Clearly, not all spfidioms can be motivated relative to the systems of conceptual metaphor outlined above. A number of idioms are structured by more restricted one-shot
metaphors, and several expressions are opaque. In addition, metaphorical motivation often occurs in combination with metonymy to create various forms of
metaphtonymy. This said, it must be emphasised that the analysis has uncovered a strong trend for the majority of idioms belonging to this word-field to
be at least partially related to well-structured systems of conceptual metaphor.
This substantiates the view of idioms as cognitive micro-models advocated
in this study. Constituting conventionally pre-coded complex-scenes, idioms
transport generally accepted associations of salient and concrete literal scenes
with more abstract conceptual targets. In discourse, these pre-configured conceptual associations can be evoked to concretise or (re)conceptualise complex
conceptualisations in a conventionally-sanctioned figurative format. In analogy to Lakoff and Johnson’s phrasing, idioms thus constitute linguistically
pre-configured cognitive micro-models we live by.
Thus, rather than substantiating the view of idiomatic constructions as
unmotivated, unanalysable and idiosyncratic units, a cognitive-linguistic analysis points to their systematic nature. If one avoids an overly narrow, atom-
JB[v.20020404] Prn:20/02/2006; 14:56
F: HCP1705.tex / p.32 (1664-1667)
 Idiomatic Creativity
istic description of idioms relative to the principle of compositionality, and if
one broadens the semantic perspective to include conceptual domains, imageschematic structures and complex patterns of semantic extension, idiomatic
subsystems turn out to be far more organised than was assumed in their traditional grammatical description. At least this is true for the phraseological
subsystem analysed in this chapter. The extent to which conceptual metaphor,
metonymy, blending and emblems play a role for the lexicogrammatical behaviour and systematic variability of these constructions in discourse will be
analysed in the second analytical part of this book.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.1 (47-111)
chapter 
Idiom variation and variability –
a cognitive-linguistic model
Any theory rich enough to generate the meanings people actually assign to nonfigurative language is rich enough to deal
with figurative language as well. (David E. Rumelhart 1993)
Frozenness was established as one defining characteristic of idioms. Given the
superficial recognition of idiomatic frozenness, the search for lexical and syntactic flexibility in idioms seems a contradiction in terms. Being partially defined as stable linguistic units, idioms appear to be prototypical exemplars of
non-creative language. Contrasting with this view, it has long become a phraseological truism that idioms are often varied in discourse and may also possess
institutionalised variants. For instance, Moon (1998: 120) finds that in her corpus of English phraseological constructions 40% have “lexical variations or
strongly institutionalised transformations, and around 14% have two or more
variations on their canonical forms.” Idiom variation thus constitutes a counterweight that works against the phraseological tendency towards stability. The
seemingly characteristic stability of idioms must therefore be relativised (cf.
also Burger 1998a: 31).
Ultimately, idiom variation is a phenomenon of language in use. Therefore,
the variable nature of an idiom can only be directly observed through actual
usage-data. Idiom-variation data provide strong empirical evidence speaking against a uniform conception of idioms as being similar to long words.
It is the major aim of this book to provide an explanatory framework for
systematic idiom-variation based on a cognitive-linguistic model of idiomrepresentation. So far, we have provided the cognitive-linguistic account of
idiom-representation: idioms are complex linguistic routines which are mentally represented as idiomatic activation-sets. To explain idiom variation on
the basis of this model, we must now show how these configurations are activated in actual discourse; that is, we have to provide a cognitive-linguistic
model of idiom use and variation. Trying to grasp the idiom-variation nettle,
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.2 (111-166)
 Idiomatic Creativity
this chapter will discuss the spectrum of idiom use and variation in cognitivelinguistic terms.
. Frozenness vs. variability – towards a cognitive-linguistic view
In Section 1.1, frozenness was preliminarily defined as the limited lexicogrammatical flexibility of idiomatic constructions. If we analyse their discursive
behaviour from the counter-perspective of idiom variation, however, most idioms possess a certain openness to lexicogrammatical malleability. I call this
degree variability. Put differently, variability defines an idiom’s limited potential for variation in discourse. Frozenness and variability thus constitute two
counterforces that determine the lexicogrammatical behaviour of idioms (cf.
Burger 1998a: 25; Keil 1997: 68).
To define the concepts of frozenness and variation, it is necessary to assume the existence of a neutral form. Following Barkema, each phraseological
construction can be attributed a base form, which is defined as “the simplest
morpho-syntactic form that an expression can take” to activate its specific idiomatic meaning and function (Barkema 1996: 141). Very often, idioms are
used neutrally, in an unobtrusive way:
(1) The latest version really upset the applecart [...]. (CTX: 2201).
(2) We’ve got to grasp the nettle. (HUE: 413).
(3) Sarah hesitated, then seemed to decide to take the bull by the horns.
(CB8: 705).
(4) But now is the time for everyone to roll up their sleeves. (K52: 1571).
The term neutral use can be employed as a heuristic concept to define the standard usage of an idiom according to its base-form. Uses that deviate from this
standard are defined as variations or alterations. Slightly modifying and extending the terminology suggested by Barkema (1996: 141–149), I will use the
following terms to denote the range of phenomena involved in idiom variation. (Note that the significance of some of these distinctions will only become
obvious later in this chapter):
–
–
Variation/alteration: any type of formal and semantic change of the baseform and/or idiomatic meaning of an idiom.
Idiom variant: the idiomatic construction resulting from a given variation/alteration.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.3 (166-219)
Chapter 6. Idiom variation and variability 
–
–
–
–
–
–
–
Institutionalised variant: an idiom variant that has become institutionalised.
Usual variant: a frequently recurring idiom variant.
Occasional variant: a non-recurrent idiom variant that appears in actual
use.
Systematic variant: an idiom variant that can be described as regular and
grammatical.
Pun variant: an idiom variant that exploits the formal and semantic structure of the base-form for the sake of wordplay.
Erroneous variant: a non-intentional departure from the base-form; an
idiomatic slip-of-the-tongue.
Pseudo variant: a departure from the base-form that cannot activate the
idiomatic meaning but can only be interpreted literally.
To approach these phenomena in cognitive-linguistic terms, we are first forced
to re-interpret the notion of base-form. For this purpose, let us reconsider
part of Langacker’s previous statement against an oversimplified treatment
of idioms:
[...] an idiom may be recognized as a unit that is to some degree independent
of a specific overt morphemic arrangement, even if one such arrangement is
far more similar and hence more “usual” than the others.(Langacker 1987: 25)
What Langacker describes as a “specific overt morphemic arrangement” refers
to the potential lexical, syntactic and morpho syntactic alterations that an idiom variant can adopt in discourse. Nevertheless, one of these forms is “more
‘usual’ ” in Langacker’s terms. Following our terminology, this is the base-form.
In accordance with the usage-based view of grammar as a complex inventory
of digested performance, an idiom’s base-form can thus be defined as an idiom’s context-independent default-structure that is destillated from various
usage-events. This constructional standard is directly associated with the idiomatic meaning, which itself emerges as a meaning standard (a schematic
meaning-type) derived from concrete occurrences.
My cognitive-linguistic model of idiom-variation starts from the assumption that both the formal and the semantic structure of a given idiomatic construction is mentally represented in terms of a context-independent standardconfiguration – a complex but structured activation-set as outlined in the
previous chapter. The assumption of such a neutral form of representation
creates a basic methodological problem. Given the empirically-attested and
widespread variability of idiomatic expressions, it seems inconsistent to pro-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.4 (219-270)
 Idiomatic Creativity
mote the existence of such a canonical base-form; this problem is eloquently
addressed by Moon (1998: 121):
All this [evidence for variation] calls into question the whole notion of fixedness, shedding doubt on the viability of the notion of the canonical form. [...],
I shall set out evidence which suggests that the notion should be superseded,
and newer models of FEIs [= fixed expressions and idioms] developed in its
stead. The line I shall take in exploring this issue may seem perverse or illogical, in view of my conclusions, since I propose to start from the assumption
that I myself started from and that studies of FEIs in general start from: the
assumption that FEIs have fixed or canonical forms and that variations are to
some extent derivative or deviant. [...]. With regard to the term ‘FEI’ I would
defend it by saying that even in extreme cases there still remains some kind of
fixedness, symmetry, or integrity: it is just that it is not always lexical fixedness.
Here, this methodological problem is tackled by the cognitive-linguistic conception of idioms. In my model, idioms are not conceived as word-like lexical
units with a fixed and canonical surface-structure; they are complex linguistic and conceptual activation networks, which can be unfolded variably. On
the formal level this complex configuration is conventionally symbolised by a
base-form, i.e. a basic set of specific lexical units that are arranged according to
a typical constructional schema. For instance, [sbj grasp the nettle] constitutes
the base-form of the corresponding idiom. This pre-coded construction constitutes the institutionalised and conventional symbolisation of the idiom’s rich
predication, which itself functions as a cognitive micro-model. It is this cognitive integrity of idioms as complex lingua-conceptual arrangements which
characterises their mental fixedness.
For purely practical reasons, I will equate a given base-form with the idiom’s
citation-form in idiom dictionaries. I take it for granted that lexicographic
practice attempts to record only highly familiar lexicalised constructions belonging to the langue of a given variety (i.e. those units that are entrenched
in the mental lexicons of most speakers). Dictionary citation-forms therefore
approximate the present view of a usage-based default construction. Thus,
the identification of a citation-form with a mentally represented base-form is
methodologically unproblematic as long as the dictionary entry is not interpreted as a mentally-fixed lexical unit that can only be processed through direct
stipulation. In contrast, the citation-form only constitutes the formal surface
of the complex mental activation-set that can be dynamically adapted to the
demands of a given usage event.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.5 (270-310)
Chapter 6. Idiom variation and variability 
In what follows, I will first sketch the potential alterations of the lexicogrammatical surface and the semantic structure. Afterwards, I will address
the central problems that we are faced with when analysing these alterations.
.. The technical description of formal and semantic alterations
in idiom variants
Irrespective of the underlying linguistic motivation for and the communicative effect resulting from a given idiom variant, it is possible to describe and
categorise the formal and semantic changes reflected by any variant. Taking
all possible types and subtypes of formal and semantic variation into account,
one can propose the following scheme for the technical classification of idiomvariation phenomena:
morphosyntactic
formal
syntactic
lexical
idiom variation
polysemy (lexicalised)
semantic
ambiguation (discursive)
meaning adaptations
(intensification,
specification, etc.)
Figure 6.1 Technical classification of idiom alterations
In the following sections these classes and subclasses will be characterised more
explicitly. In doing so, I will only highlight those alterations that are of direct
or indirect interest for the present study. More comprehensive and meticulous
analyses of formal and semantic variation are offered by Moon (1998: Chapter
4), Burger et. al (1982: 28–30), Burger (1998a: Section 1.3) and Sabban (1998).
... Formal variation
a. Morphosyntactic variation:
Morphosyntactic variation covers inflectional variants of one (or several) idiom constitutents, including verb inflection, noun inflection (pluralisation)
and the flexible use of determiners and quantifiers. For instance, the vari-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.6 (310-373)
 Idiomatic Creativity
ant of swallow the bitter pill in (5) features determiner variation (some) and
pluralisation with the NP bitter pill:
(5) The traditionalist camp had to swallow some bitter pills. (A1V: 1025).
b. Syntactic variation:
Syntactic variation involves changes in the constructional organisation of the
base-form. (6), for example, involves idiom extension through postmodification by an of -complement, while (7)–(8) reflect syntactic permutations such
as passivisation and clefting:
(6) All this has upset the applecart of the relation of fertility to prosperity.
(EDK: 1431). (from upset the applecart)
(7) ‘Too late now for being discreet. All the Ardakkean beans have been spilled.’
(G3G: 1532). (from spill the beans)
(8) It was often said that her new-found reputation was a bubble that would
burst once people realized it was just the raunchiness of her work that got
her a wide readership. (AC3: 853). (from burst the bubble)
c. Lexical substitution:
This variation-type refers to general alterations of an idiom’s lexical constituents. While it is possible to classify lexical substitutions according to word
class, it is more interesting to focus on the semantic relationship between the
alternating elements and the meaning adaptations that result from the substitutions (for a precise discussion see Moon 1998: 124–139). (Meaning adaptations will be specifically discussed below). For instance, it is possible to create
synonymous variants on the basis of synonymous (9) and non-synonymous
alternants (10). Or one can trigger antonymous variants on the basis of antonymous alternants (11):
(9) a hard row to hoe vs. a tough row to hoe
(10) bleed NP dry vs. bleed NP white
(11) go up in the world vs. come down in the world
... Semantic variation
In most cases, changes in the formal structure of an idiom are not semantically neutral, but trigger adaptations of the idiom’s meaning. While changes
that lead to ambiguation are quite radical, other alterations create more subtle
meaning adaptations.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.7 (373-420)
Chapter 6. Idiom variation and variability
a. Polysemous idioms:
Polysemous idioms include expressions such as come a cropper, which have a
lexically invariant form but two conventional meanings (‘fall’ and ‘fail’). Idiomatic polysemy reflects the creation of an institutionalised pattern of meaning extension based on a figurative construction. Therefore, Burger (1998a: 72)
calls this semantic shift “sekundäre Metaphorisierung” (secondary metaphorisation). Such lexicalised meaning variants therefore reflect the general potential
for meaning shift in idioms.
b. Ambiguation:
Unlike polysemous idioms, semantic variation through ambiguation is not
conventionalised, but an occasional meaning adaptation. Ambiguation is the
generic term for the creative process of playing with the semantic potential of
idioms (Burger et al. 1982: 29).35 By the process of ambiguation both levels of
meaning – the idiomatic meaning and a potential literal meaning – have their
own reference in the usage-context (see also Langlotz 2001a).36 The advertising
slogan in (12) exemplifies this process:
(12) If you like wine, have a good nose.
(WWW.WEEKENDWINES.COM)
The idiom have a nose for sth. (‘be good at discovering precious and interesting
things’) is ambiguated by activating the literal meaning of have a good nose. A
wine expert, who is good at discovering interesting and high quality products,
must also depend on his or her nose. In other words, in co-activation with
the idiomatic meaning the literal meaning of the idiom is also given its own
referential value: wine connoisseurs can actually distinguish good quality wine
from average wine by means of their olfactory sensibilities. Thus, this conscious
variational process exploits the semantic structure of the idiom, while being
independent of any formal changes.
c. Meaning adaptations:
In many lexicalised and occasional variants, the idiomatic meaning is modified in combination with corresponding lexicogrammatical alterations. Such
meaning adaptations comprise intensifications, specifications, antonymy and
perspectivisations. For instance, the variant of upset the applecart in (13) intensifies the standard meaning by elaborating the underlying image by the
premodifying adjective phrase most ponderous:
(13) [...] this time overturning the most ponderous applecart of all. (B74:1571).

JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.8 (420-472)
 Idiomatic Creativity
Specification narrows down the scope of reference of the idiom variant if compared to the denotation of the base-form. For instance, the meaning of the
postmodified variant in (14) is more specific than the one of its neutral form
(fan the flames):
(14) Preaching hatred, as people like Ian Paisley do, only fans the flames of
vengeance. (K55: 252).
A number of verbal idioms vary their lexical or syntactic structure systematically to focus on the scene or action described by the literal meaning from
an alternative perspective (see Sabban 1998: 80; Dobrovol’skij 1988: 173–184;
Moon 1998: 139–145). For instance, the uses of the idiom in (15) alter between
a declarative and a causative reading:
(15) a. the curtain comes down on sth.
b. bring the curtain down on sth.
With this classificatory scheme in hand, we can now address the analytical
problems and questions that are raised by such alterations.
.. Problems in the analysis of idiom variation
Although idiom variation has long been recognised as an empirical fact, its
precise description and theoretical explanation is far from trivial. Consider the
following data:
(16) It seemed to us that two nettles had to be grasped: the first was to decide
whether divorce law should attempt to remedy the ‘injustice’ meted out to
‘innocent’ spouses who are divorced against their will. (BNK: 1130).
(17) *It seemed to us that two buckets had to be kicked.
The variational use of grasp the nettle in (16) involves two basic syntactic and
morphosyntactic alterations: first, the NP the nettle is pluralised and quantified → two nettles; second, the NP is emphasised by fronting → two nettles had
to be grasped. If the same alterations are applied to the idiom kick the bucket
as in (17), the resulting sequence cannot be interpreted according to its idiomatic meaning but triggers a non-idiomatic literal-reading. Thus, while the
alterations are fully grammatical for the former idiom, they are ungrammatical
for the latter.
A further problem is illustrated by the following pair of variants:
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.9 (472-524)
Chapter 6. Idiom variation and variability 
(18) Each spurt in investment has for a time been halfway successful in boosting harvests and production, but policy to date has failed to grasp the nettles of productivity, variety, distribution and responsible land use. (CMT: 44).
(19) ‘Social imperialism’ suggests that the main beneficiaries of this policy were
British consumers, and indeed one writer has gone so far as to argue a direct link to the Attlee government’s social reforms: ‘The nationalisations,
medical provision and expansion of education so magnanimously legislated by the Labour Ministry were largely achieved because the Bank of
England kept the Sterling Area show on the road.’ (A6G: 902).
In (18) and (19) the NPs the nettle and the show of grasp the nettle and get
the show on the road, respectively, are modified. While (18) involves postmodification by an of-complement (the nettles of productivity ...), (19) features
noun-premodification (the Sterling area show). Thus, (18) and (19) involve
slightly different modification processes. However, the communicative function of the modifiers is the same. They point to the specific referents, i.e. the
topics, of the NPs. The function of topic-indication (see Section 6.4.3 below)
also underlies (20). Here, however, it is implemented by the process of lexical
substitution:
(20) Data Connection, too, is playing its price-list close to its chest. (CNM: 89).
(from: keep one’s card close to one’s chest)
In other words, different lexico-syntactic alterations may underlie one and the
same variational purpose.
The process of lexical substitution can trigger a variety of different communicative effects. Thus, while (20) is grammatical and relatively inconspicuous,
(21) must be described as a pun variant of call a spade a spade because it creates
a striking, humorous effect:
(21) I wondered why Americans always called a spade a manually powered
entrenching instrument. (CCW: 951).
Since it is not possible to predict a direct and precise correlation between the
type of formal and structural change and the semantic and communicative effects triggered by these alterations, idiom variation cannot be simply explained
in terms of a uniform idiom-variation grammar. While for one idiom a specific structural alteration is fairly trivial and hardly noticeable, the very same
change may be unconventional or ungrammatical for another idiom or lead to
a completely different communicative effect. With a view to these facts revealed
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.10 (524-581)
 Idiomatic Creativity
by usage-data, the phenomenon of idiom variation must therefore be analysed
along three distinct dimensions:
a. the technical description of the formal and semantic alterations that appear
at surface
b. the variation processes that underlie these alterations (topic indication,
punning, etc.)
c. the classification and evaluation of these alterations in terms of their
linguistic regularity and communicative motivation (systematic vs. nonsystematic variation, wordplay, etc.).
For a cognitive discussion and explanation of idiom variation, the revelation
of motivated links between (a), (b) and (c) must be the ultimate objective. Put
differently, a cognitive-linguistic explanation of idiom variation should be able
to model an idiom’s lexicogrammatical variation-potential; it should be able
to explain the semantic and communicative effect of a given alteration for a
given idiom in a given context. Moreover, it must find reasons for the different variation potentials of different idiomatic-constructions and explain them.
This amounts to the revelation and description of a cognitive idiom-grammar
which is powerful enough to explain systematic idiom variability. Dobrovol’skij
(1997: 69–70) describes this challenge as follows:
Zu den Aufgaben der Theorie der Phraseologie gehört die Beantwortung
der Frage, warum bestimmte Idiome sehr breitgefächerte Transformationen ihres Konstituentenbestandes zulassen, während andere Idiome in ihrem
Wortlaut überhaupt nicht verändert werden dürfen, ohne dass es als Verletzung der Norm empfunden wird bzw. sogar Verständnisprobleme schafft.
Gibt es in diesem Bereich bestimmte Regeln? Wie kann sich ein Sprecher
diese Regeln merken? Eins ist klar: Der Sprecher kann nicht für jedes Idiom
eine Liste von transformationellen Restriktionen auswendig gelernt haben.
Dies anzunehmen wäre nicht nur vom sprachpsychologischen Standpunkt
aus äusserst unrealistisch. [...]. Man kann annehmen, dass entweder in der
formalen oder in der semantischen Struktur des Idioms bestimmte Besonderheiten enthalten sind, die dem Sprecher signalisieren, welchen Transformationen das betreffende Idiom standardmässig unterliegt. Wenn es gelingen würde, in diesem Bereich bestimmte Regularitäten aufzudecken, so dass
morphologische, syntaktische und lexikalische Modifizierbarkeit der Idiome
prädiziert werden könnte, wäre das ein bedeutender Fortschritt nicht nur im
Sinne der lexikographischen Kodifizierung der Idiome, sondern auch im Sinne
der Theoriebildung. [...].
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.11 (581-632)
Chapter 6. Idiom variation and variability 
The revelation of an idiom-grammar thus amounts to the description of a system of cognitive conditions under which the stability of an idiom is weakened
or enforced. From a cognitive-linguistic perspective these conditions must reside in the specific shape of an idiomatic activation-set as represented in the
cognitive-grammar of a speaker. Therefore, we must try to detect correlations
between different idiom-activation-set types and their variation potential. For
this purpose, I will reinterpret idiom use and variation in terms of Langacker’s
usage-event model.
. Idiomatic creativity
So far, we have only provided a descriptive classification of potential formal
and semantic changes caused by idiom variation. To explain these changes, it is
necessary to model idiom-use in processing terms.
.. Idiomatic usage-events
In Chapter 4, idiomatic configurations were defined as complex linguistic units
that can be activated as mental standards to encode or decode a concrete usageevent. An idiomatic activation-set thus functions as an idiom-type that can be
instantiated as a specific idiom-token in actual discourse. In Figure 6.2, this
process of idiomatic coding is depicted in accordance with Langacker’s usageevent model (see Section 3.4.3, Figure 3.4). In other words, any given idiomatic
activation-set can work as a symbolic standard to encode a context-specific
target-conceptualisation (or to decode a target-vocalisation in the process of
comprehension). This can be illustrated by the following use of grasp the nettle:
(22) The proposed reform of nurse education in the UK, which received approval in principle from the Conservative government in May 1983,
promises a period of tremendous upheaval as the delicate balance of educational and service priorities is carefully weighed and restructured. But
with upheaval comes renewal, and an unrivalled opportunity to grasp the
nettle nursing’s leaders have talked about for years: an opportunity to redefine and reshape the parameters of nursing itself. (Returning to nursing –
a guide for nurses and health visitors, taken from BNC, CHT: 1154).
To produce the idiom-token grasp the nettle in (22), it is necessary for a given
speaker (or writer) to activate and select the corresponding activation-set as a
pre-established linguistic standard from his or her cognitive grammar. The lin-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.12 (632-665)
 Idiomatic Creativity
GRAMMAR
Idiomatic activation-set
USAGE CONTEXT
DOMAINg
a
A
A
C
Ç
g
Ç
b
B
b
a
B
usage event
coding conceptualisation
(to be coded by
idiom)
coding
STANDARD/
SANCTIONING
STRUCTURE
vocalisation
(to be coded by
idiom)
TARGET STRUCTURE
CATEGORISATION
Figure 6.2 Idiomatic usage-events
guistic and conceptual usage-context or domain of (22) suggests that the actual
meaning deals with reform in the area of nursing. More specifically, the paragraph deals with tackling problems in the area of nursing. Let us assume
that this is the cognitively-represented context, or the cognitive environment,
that the speaker had in mind when formulating the passage.
To encode the actual target-conceptualisation, i.e. the strong implicated
meaning of grasp the nettle in (22), the speaker first had to search his or her
grammar for a unit whose semantic pole is able to give expression to this
meaning. Since grasp the nettle has the complex idiomatic-meaning tackle a
problem with determination, this idiom can be activated from the grammar to work as a symbolic schema to give expression to the actual targetconceptualisation; thus, tackling the nursing reform problem in (22) is
categorised as a subtype of tackling a problem. In other words, this targetmeaning works as the contextual referent onto which the entrenched denotation of grasp the nettle is projected.
However, since idioms involve figuration, the abstract target-meaning is
not encoded literally, but on the basis of a non-literal idiomatic construction.
By our cognitive-linguistic definition, idioms convey complex scenes. Thus the
idiomatic meaning must always be encoded by means of a pre-coded construction that encapsulates a literal scene. To the extent that this literal scene is con-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.13 (665-724)
Chapter 6. Idiom variation and variability 
sistent, it works as the semantic scaffolding or background against which the
target-meaning is conceived, modelled, reconceptualised or reframed. Thus, in
(22) the scene of grasping the nettle is mobilised to provide a concrete model
for the more abstract idea of tackling the nursing problems. Thus, the speaker
probably chose the idiom for the sake of concretising the problems talked
about. By the help of the idiom’s conceptual force, problems are conceived as
dangerous, stingy nettles that are unpleasant to approach. The literal meaning triggers a number of weak implicatures that enrich the implicated meaning
about nursing problems. Using the idiom, the speaker thus becomes able to exploit the proverbiality of the expression: he or she can model and evaluate the
target conceptualisation with reference to the concrete literal scene.
With the selection of the complex idiomatic predication, the constructional schema [sbj grasp the nettle], which consists of the lexical units grasp and
nettle and which profiles the literal scene (grasp the nettle), is (automatically) co-selected. To encode the context-specific meaning, this constructional
schema must often be adapted. In other words, the idiomatic construction
must be manipulated to provide for the functional encoding of the target.
However, in (22), the constructional schema can be directly instantiated without any further adaptations. Thus, (22) constitutes an unvaried, canonical evocation of the idiomatic construction according to its neutral base-form. With
the activation of this construction, the phonological (or orthographic) pole is
automatically co-activated. In speech, the phonologial structure is transformed
into an actual vocalisation; in this written example into the actual sequence
of letters.
I regard this informal account of idiom production as being grossly compatible with Chafe’s framework:
[...] we might say roughly that an idiom is a semantic unit which is symbolized
by other semantic units before these go on to be symbolized by phonetic units.
[...] If we reverse this view to begin with semantics rather than syntax, we can
say [...]: that idioms are semantic units like other semantic units, but that they
require conversion into arrangements of other semantic units before they are
further encoded into sound.
(Chafe 1968: 121–122)
On the basis of this general account of idiomatic usage-events, we can now
scrutinise idiom variation and variability in cognitive-linguistic terms: idiomvariation is the creative computation of a novel idiomatic standard for the
purpose of encoding a given target conceptualisation.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.14 (724-753)
 Idiomatic Creativity
.. Creative idiomatic usage-events
Variability is bound to the concept of linguistic creativity – the flexible evocation of linguistic units to create diverse linguistic constructions. To be able
to scrutinise the phenomenon of idiom variation and variability in cognitivelinguistic terms, we must re-examine the cognitive-linguistic conception of
linguistic creativity and apply it to the description of idiom variation. As was
illustrated in Section 3.4.5, Langacker understands linguistic creativity as the
result of the behavioural necessity to adapt the means of symbolic expression
to constantly changing experiences. Moreover, from the perspective of relevance theory, linguistic creativity emerges from the demand to make one’s
contribution maximally relevant, i.e. to maximise its contextual effects. Hence,
linguistic creativity results from the communicative urge to adapt one’s linguistic units – the means of linguistic expression that are entrenched in the
cognitive grammar – to code complex and ever changing conceptualisations in
a contextually appropriate way. This process involves the cognitive computation of a non-entrenched linguistic standard. How, then, can this conception
of linguistic creativity be transferred to the notions of idiomatic frozenness and
variability?
To describe idiom variation and variability as a form of linguistic creativity, we must provide a cognitively motivated account of creative idiomaticcoding. As should have become obvious in the discussion of linguistic and
psycholinguistic research on idiom variation, idiom production is not reducible to a process of direct lexical stipulation. Rather, the lexicogrammatical
changes in idiom variants must be explained in terms of creative cognitive processing (as exemplified by the analysis of crayon sharpener in Section 3.4.5).
Creative idiomatic coding is based on mental computation; it involves the
mental manipulation of the idiomatic activation-set and its integration with
other pre-established symbolic units and constructional schemas. The result
of this computation process is a novel idiom variant – an idiomatic standard
that can encode a target conceptualisation for which the pre-established baseform does not offer a sufficiently appropriate and relevant match. To trigger
contextually-relevant and fine-tuned implicatures, the speaker must manipulate the idiomatic activation-set correspondingly.
To activate and manipulate linguistic units for coding, speakers must construe their predication. That is, they must take quick and unconscious coding
decisions and make corresponding focal adjustment along the following coding
parameters (see analysis of transitive coding in Section 3.4.4):
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.15 (753-819)
Chapter 6. Idiom variation and variability 
1. Selection of a conceptual base (defining the scope of predication of the
clause) and choosing a scene for profiling.
2. Construal of scene in terms of
a. defining the level of abstraction at which the scene is conceptualised.
b. defining the level of granularity/precision with which the scene is conceptualised.
c. defining the relative prominence of the profiled substructures.
Depending on the outcome of these decisions, a specific (set of) linguistic
constructions can be selected, which encapsulate the manipulated conceptualisation most effectively.
However, the base-form of an idiom such as grasp the nettle already constitutes the pre-coded symbolic profile of a literal-scene. Thus, in the formal
structure [sbj grasp the nettle] the construal of the idiom’s literal scene relative
to these coding parameters is already pre-determined by the conventional lexicogrammatical structure of the idiom. More specifically, the construction has:
1. A fixed scope of predication: the base-from evokes a restricted scenario of
physical activity. Within this scope the constituent structure profiles the
scene agent grasps the nettle.
2. A pre-established construal of this scene in terms of
a. a specific level of abstraction at which the scene is conceptualised:
grasp and nettle can be described as basic-level concepts, thus the
specific scene is pre-coded at the basic-level.
b. a conventional level of granularity/precision at which the scene is conceptualised: grasp the nettle does not convey specific details about the
characteristics of the nettle such as grasp the green nettle.
c. a pre-defined degree of relative prominence of the profiled substructures: the profiled force-dynamic relationship, agent grasps the nettle, places the agent into the syntactic foreground in accord with the
schematic meaning and function of transitive clauses (see Section 3.4.4)
In cognitive-linguistic terms, frozenness thus describes the impossibility to
change this pre-coded lexicogrammatical structure along the coding parameters, i.e. to alter the conveyed conceptual content or the construal of the literal
scene. In contrast, variability suggests that idioms are, to some extent, open
to additional coding decisions. Idiomatic variability thus implies that the lexicogrammatical base-form can be altered to modify or break-up this pre-coded
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.16 (819-879)
 Idiomatic Creativity
symbolic construal. Following Carter’s concept, idiom variation thus reflects
an idiomatic form of pattern re-forming (see Section 3.4.5).
Thus, idiom production always involves the mental evocation of an idiom
activation-set for the sake of rendering the target-meaning idiomatically. Coding decisions must be taken relative to the target conceptualisation. However,
since idioms are complex scenes, the choice of an idiom for coding is motivated
by its cognitive and discursive functionality. If speakers encode a target conceptualisation on the basis of an idiom, they choose to render this target conceptualisation figuratively, i.e. by means of a figurative cognitive micro-model. Since
the literal scene is evoked for the sake of providing an alternative, non-literal
conceptualisation of the target, the coding process must be effected against this
conceptual background. Put differently, the coding decisions listed above are
not merely applied to the target, but to the target as it is understood via the
literal scene: any creative adaptations of the base-form, [sbj grasp the nettle],
are the result of coding decisions that apply to the target conceptualisation but
which are mediated via the entrenched idiomatic meaning and effected in relation to the literal scene. Idiomatic coding therefore results in a complex process
of conceptual and linguistic integration relative to three levels of meaning:
a. the actual, discursive meaning that is to be encoded,
b. the entrenched idiomatic meaning and
c. the literal meaning.
Thus, when adapting an idiom for context-specific coding, these three levels of meaning must be mentally coordinated. This abstract account can be
exemplified by the following example.
In (22) the idiom grasp the nettle was directly stipulated. The literal scene
was not further adapted in response to the target conceptualisation because
it provided an appropriate match to code this information. In contrast, (23)
reflects idiom variation:
(23) ‘The Government has shown in the past that it is willing to grasp nettles
that others have shied away no doubt we shall grasp this one if we have to.’
(A2L: 12).
The variational use in (23) involves pluralisation (nettles), corresponding article variation (grasp Ø nettles, i.e. the use of a zero article to mark an indefinite
plural) and postmodification by a restrictive relative clause (that others have
shied away from). In addition, the variant features anaphora (grasp this one).
This occasional variant thus reflects the following changes in the symbolic
construal of the literal scene:
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.17 (879-944)
1. The content of the pre-coded scene is modified. Instead of one specific
nettle, the agent grasps an indefinite number of unspecified nettles.
2. The pre-established construal of the scene is extended by enhancing the
conventional level of precision with which the scene is conceptualised: nettles is postmodified by that others have shied away from. Thus, more specific
information about this nominal head is added.
These variations suggest that, in the process of coding the actual target conceptualisation, the speaker subjected the mentally represented idiom
activation-set to a number of coding decisions to make the expression grasp
the nettle better suit the idea that he or she had in mind when formulating
the statement in (23). Ignoring the postmodification and the use of the indefinite zero article, the cognitive computation processes that underlie this creative
coding-event can be modelled as follows:
a. The idiomatic activation-set for grasp the nettle must be triggered in the
mental lexicon, i.e. the mental network comprising the idiomatic scene
[agent tackles a problem], the base-form [sbj grasp the nettle] and the
literal scene [agent grasps the nettle] must be activated and unfolded.
b. To alter this complex configuration for context-specific coding, the abstract
idiomatic meaning must first be adapted to correspond to the implicated,
actual target-meaning (the government tackles problems that others...). This adaptation process is mediated by the literal scene, [agent
grasps nettle], which works as a cognitive model for the more abstract
target-meaning. Thus, the adaptation process is dependent on the conceptual organisation of the complex scene.
The internal semantic structure of grasp the nettle is motivated and
analysable due to the presence of an underlying set of conceptual metaphors: mental grasp is physical grasp, controlling is grasping,
problems are dangerous physical entities (see Section 5.2.7). This
makes it possible to adapt the idiomatic meaning to the context by manipulating the literal scene: the government tackles problems that
others... corresponds to the government grasps nettles that others. The conceptual metaphors thus mediate the manipulation of the
literal scene and the corresponding adaptation of the lexicogrammatical
structure for the communication of the actual target-meaning.
c. Thus, to give expression to the context-specific semantic adaptation, the
base-form, [sbj grasp the nettle], must itself be modified to code the targetmeaning in a satisfactory way. On the formal level, the adaptation of the
literal scene corresponds to the alteration of the lexical units. In the present

JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.18 (944-1011)
 Idiomatic Creativity
case this involves the co-instantiation of the plural-schema to compute
the specific plural form of nettle. Since grasp the nettle is motivated and
isomorphic, the head-noun nettle can be direct pluralised in response to
its figurative target: problems. Put differently, nettle can be directly pluralised because this constituent possesses the phrase-induced figurative
meaning ‘problem.’ This autonomous semantic value is effected by the motivated and isomorphic conceptual association of the literal scene with the
idiomatic meaning.
This example illustrates the following claims. Grasp the nettle constitutes a
conventional and institutionalised lingua-conceptual routine that conveys a
complex scene. Its pre-coded base-form profiles the literal scene of this bipartite conceptualisation. Hence, any changes in the constituent structure, [sbj
grasp the nettle → sbj grasps nettles], are dependent on how the literal scene
can be manipulated in response to the target conceptualisation via motivating
conceptual links. The two scenes (the literal and the target scene) are functionally related: the literal source works as an (alternative) model for the target.
When the idiomatic configuration is activated, the literal scene is evoked as a
conceptual scaffolding – a standard to categorise the conceptual target in accordance with the conceptual metaphors that motivate this association and render
it analysable. The conceptual metaphors thus pre-structure the adaptation process in (23). In this usage-event, the literal scene is accommodated in response
to the context-specific meaning. Accordingly, the lexicogrammatical structure
is adapted for the sake of maximally relevant coding.
Idiomatic creativity thus amounts to a form of figurative-language competence. It involves the ability to manipulate a linguistically pre-coded, figurative
cognitive micro-model to adapt it to the communicative purposes of a specific
usage-event. Systematic creative adaptations are supported and restricted by
the specific semantic and formal characteristics of this micro-model.
With a view to its psycholinguistic implications, this cognitive-linguistic
account clearly speaks against a direct lexical stipulation model of idiom production. While some form of mental stipulation certainly takes place when the
idiom activation-set is unfolded, the complex coordination of the three levels of meaning and the corresponding manipulation of the lexicogrammatical
structure rather suggests a simultaneous processing view of idiom production.
In order to fine-tune the conceptual construal of the literal scene relative to
the target conceptualisation and in order to effect the corresponding linguistic alterations, conceptual and linguistic processing cannot occur in isolation
from one another but they must interact. By analogy with Gibbs’s simulaneous
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.19 (1011-1039)
Chapter 6. Idiom variation and variability 
processing model and Titone and Connine’s hybrid view of idiom comprehension, I would therefore like to argue that the formal and the conceptual
substructures in the idiomatic activation-set are manipulated in parallel rather
than in a modular or incremental step-by-step fashion (cf. Levelt 1999). With
isomorphic idioms, conceptual adaptations of their idiomatic meaning via
the literal scene can be directly implemented linguistically due to the phraseinduced meanings of their constituents. In other words, for these idioms an
idiomatic form of compositional production is possible. This explains their lexicogrammatical flexibility as suggested by Nunberg et al. and as evidenced by
the decomposition views on idiom processing. For motivated but unanalysable
idioms, the cogniser first has to reassert the precise semantic contribution of
the constituents in relation to the target meaning. That is, he or she first has
to check their context-specific scaffolding function with a view to how the
literal scene models the target conceptualisation. When varied, this can prolong their computation time and restrict their lexicogrammatical malleability.
For opaque and constructionally-idiosycratic idioms a motivated scaffolding
function often cannot be established. This limits their lexicogrammatical productivity. Without being able to provide a precise, psycholinguistic idiomproduction model that can describe these distinctions in technical detail, I
would like to propose that interactive processing models of language production
are best provided to account for these complexities of idiom production (e.g.,
Lamb 1999: Chapter 8).
Given this cognitive-linguistic characterisation of idiomatic variability and
frozenness, the fundamental question for a systematic account of idiom variation thus reads: to what extent are idiomatic constructions open to creative
coding decisions and to what extent are such alternatives blocked? Since the
mental activation-set that can be potentially evoked when using a given idiom
differs from idiomatic construction to idiomatic construction, one cannot expect that all idioms have the same variation potential or that they are subject
to the same variation restrictions. Rather, their variability is dependent on the
specific quality of the mental activation-pattern that can be evoked for a given
idiom. In other words, the potential creation of an idiom variant seems to be
determined by the idiom’s formal and semantic organisation and the way this
complex configuration is adapted to a concrete discursive constellation. This is
the basic claim of this study:
The semantic and the formal structure of a given idiom define its variation potential and constrain the ways of how the idiom’s base-form can be altered in
a concrete context of use. To the extent that the association between the literal
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.20 (1039-1086)
 Idiomatic Creativity
and the idiomatic scene can be motivated and analysed on the basis of underlying conceptual patterns of semantic extension, the cognitive micro-model can
be systematically manipulated. In the absence of motivating conceptual bases,
an idiom’s potential for systematic variation is restricted.
Before this hypothesis can be scrutinised on the basis of empirical data, it is
necessary to characterise different idiom-variation types. This leads us to the
next question: How can the distinction between conventional, usual occasional
variation and wordplay be delineated in cognitive-linguistic terms?
. Subtypes of idiom variation
There are many motivating factors for linguistic creativity. Accordingly, Carter
(2004: 148) states:
The purposes for creative language in common everyday speech are [...] highly
varied and can, [...], include: offering some new way of seeing the content of
the message; making humorous remarks; underlining what is communicated;
expressing a particular attitude, including negative and adversarial attitudes;
making the speaker’s identity more manifest; playing with language form to
entertain others; ending one bit of talk and starting another; or simply oiling
the wheels of the conversation [...].
While these factors point to the socio-pragmatic dimension of conspicuously
creative linguistic practices, linguistic creativity also includes the relatively inconspicuous rule-governed dimension of adapting the conventional means of
linguistic expression in productive ways. In other words, the cline of creative
linguistic practices ranges from the systematic productivity of grammatical
patterning to the conspicuous creativity underlying, for instance, poetry or
linguistic humour (cf. Carter 2004: Chapter 2). A theoretical account of idiomatic creativity must be able to account for a corresponding cline of creative
manipulations in the variational use of idioms.
The aim of this study is to carve out a cognitive idiom-variation grammar that can account for the systematic variability of idiomatic constructions.
Thus, we are now forced to find criteria to define the notion of systematic variability. It is of primary interest to chart the fuzzy area between systematic and
non-systematic idiom-variation because the notion of wordplay is often misused to exclude idiom-variation data from grammatical analysis if they do not
correspond to arguments in favour of one’s theoretical apparatus:
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.21 (1086-1143)
Chapter 6. Idiom variation and variability 
Methodologically, the ability of people to play with words is outside the scope
of a theory of idioms proper; therefore, data involving word games cannot play
a role in a theory of idioms.
(Schenk 1995: 258)
For instance, Schenk (1995: 257) claims that the following example (reproduced from Gazdar et al. 1985: 241) must be regarded as wordplay (see also
Mel’čuk 1995):
(24) The piper wants to be paid.
This variant is derived from the proverb he who pays the piper calls the tune (the
person paying for something has the right to decide how that thing operates or
is organised). The meaning of this proverb is motivated and analysable: pay
the piper corresponds to bearing the cost of an undertaking, while call the tune
refers to the sponsor’s right to make decisions. Accordingly, the variational use
in (24) can be interpreted as: the cost of the undertaking must now be covered. To
evoke this meaning, the literal scene was manipulated and coded in a creative
way. But this manipulation is in full accord with the motivated semantic structure of the idiom. Thus, as far as the metaphorical motivation of the proverb is
concerned, (24) is quite systematic. Schenk (1995: 257), however, regards this
alteration as a form of wordplay because “that sentence [...] is intended to be
funny [...].” However, in the absence of the specific usage-context, I do not see
the strikingly humorous effect created by (24).
In general, wordplay is a fairly vague term for which no clear-cut definition
exists (Veisbergs 2001: 258). In The Oxford Companion to the English Language,
McArthur (1992: 787) defines this notion as: “Any adaptation or use of words
to achieve a humorous, satirical, dramatic, critical, or other effect.” More restrictively, wordplay is characterised as a clever and creative exploitation of the
meaning and form of words; it involves ambiguation and punning. Punning
denotes the conflation of homonymous and homophonous words to produce
a semantic double take resulting in a humorous effect (see Gréciano 1987;
McArthur 1992: 822; Veisbergs 2001: 258). To scrutinise the fragile and gradual
interpretative distinction between systematic idiom-variation and idiom-based
wordplay, I will try to find cognitive criteria to illuminate this problem. In doing so, I am trying to avoid the aforementioned shortcoming of theoretical
discussions of idiom variation.
Without relying on a prejudiced definition of wordplay, I would like to
claim that the notion of systematic variability can be qualitatively delineated
although the alternative phenomena that appear on the gradual continuum of
linguistic creativity cannot always be strictly distinguished and separated. In
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.22 (1143-1197)
 Idiomatic Creativity
other words, idiomatic creativity is defined by a cline between the two poles
of strikingly conspicuous wordplay, on the one hand, and inconspicuous lexicogrammatical adaptations, on the other (cf. Burger 2000). Phenomena of
systematic idiom variability are located close to the latter pole, while nonsystematic adaptations are more clearly influenced by the latter. In other words,
systematic idiom-alterations include those variation processes that weaken an
idiom’s frozenness, but do not create a striking semantic or stylistic effect.
Systematic variation is a grammatical and regular alteration of an idiom’s baseform.
From the perspective of relevance theory, the gradual distinction between
systematic variation and non-systematic wordplay can be charted with reference to the notions of strong and weak communication. With systematic uses
of idioms the speaker primarily engages in the strong communication of the
idiomatic meaning without leaving it to the hearer to generate a series of additional weak implicatures. That is, the primary communicative intention is to
trigger a strongly implicated context-specific use of the idiomatic meaning. In
contrast, idiomatic wordplay creates contextual effects that go well beyond the
evocation of the conventional idiomatic meaning: they trigger the weak communication of additional poetic and stylistic effects. Such effects change the
mutual cognitive environment of speaker and hearer on the emotional rather
than the conceptual plane:
They do not add entirely new assumptions which are strongly manifest in
this environment. Instead, they marginally increase the manifestness of a great
many weakly manifest assumptions. In other words, poetic effects create common impressions rather than common knowledge. Utterances with poetic
effects can be used precisely to create this sense of apparently affective rather
than cognitive mutuality. What we are suggesting is that, if you look at these
affective effects through the microscope of relevance, you see a wide array of
minute cognitive effects.
Due to their intrinsic creativity and figurativity idioms are, obviously, prone
to trigger weak implicatures and stylistic effects. In the context of poetic
metaphor, this is also claimed by Carston (2002: 356):
In many good poetic metaphors, there are other apparently non propositional
effects that are achieved: qualitative states of mind, such as sensations and feelings, are evoked and these, rather than anything conceptual, may be precisely
what the poet is striving for through his creative use of words.
As we have seen in Section 4.7, the literal scene can work as a background –
a conceptual and emotional resource or pool – for the communication of
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.23 (1197-1240)
Chapter 6. Idiom variation and variability 
affect or the critical evaluation of an experience apart from its ontologising
conceptual impact. This said, I would like to argue that idiomatic wordplay
goes beyond this conventional and characteristic modelling function of idioms. With wordplay, the formal and semantic structure of an idiom is exploited as a resource for stylistic and poetic effects that also communicate
the speaker’s intention to engage in humour, to be conspicuous and to trigger additional, weakly implicated facets of meaning. In other words, beyond
the idiomatic meaning, idiomatic wordplay has the additional implicature that
the hearer is invited to join in a game of deriving a number of weak implicatures that add communicative spice to the use of the expression. Consider the
following examples:
(25) To coin a phrase, Lewis’s war-time broadcasts on behalf of God put a
pigeon among the cats. (CKN: 477). (derived from: put the cat among
the pigeons)
(26) For a short while. It is the early bird watcher that catches the bird. (F9H:
1380). (derived from: the early bird catches the worm)
(27) Governments argue that they cannot make a development omelette without breaking heads. (CR9: 62). (derived from: you can’t make an omelette
without breaking eggs)
In (25) the conceptual relationship conveyed by the conventional idiom put the
cat among the pigeons is reversed. This creates a striking meaning adaptation.
Being a Christian emblem for peace, the constituent pigeon is semantically
reloaded in the communicative context of war-time broadcasts. Correspondingly, the constituent cats stands for the warlords. Based on these unconventionally highlighted meanings, the literal scene is reconstituted to create the
new context-specific meaning ‘appease an aggressive situation.’ The playful
nature of this usage-event is explicitly marked by the phrase to coin a phrase.
(26) clearly involves ambiguation and thus communicates two strong implicatures. More precisely, the novel literal scene created by the variant can be
understood both literally and idiomatically. Thus, literally, the bird watcher
who starts his activity early in the morning is claimed to have the best chances
of spotting the bird. Idiomatically speaking, being early thus means to be successful in the context of bird watching. Obviously, this conspicuous adaptation
of the literal scene was consciously created by the speaker to link the expression
to the usage-context. Since the conventional literal scene (the bird catches
the worm) still reverberates in the semantic background, the variant explic-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.24 (1240-1298)
 Idiomatic Creativity
itly communicates its own creativity. Thus, as a potential weak implicature, the
bird watcher can himself be equated with a worm-catching bird; this is funny.
Finally, (27) is a pun on the proverb, you can’t make an omelette without
breaking eggs. The substitution of heads for eggs is conspicuous. While the
substitution is motivated by the usage-context (planning political development), the variant alludes to the fact that any political development must
be well-thought; i.e. heads must be ‘broken’ to find an appropriate solution.
The substitution head for egg, with its weakly implicated humorous allusion
to the compound egghead (highly educated but impractical person), creates
additional humorous effects that go well beyond the conventional modelling
function of the proverb. Moreover, the blended conceptualisation of an addled development omelette has the potential of conveying a further critical
evaluation that points to the potentially insane outcome of this planning process. This pun variant nicely reflects the speaker’s intention of engaging in the
weak communication of a great number of affective contextual effects as addressed by Sperber and Wilson. The pun thus invites the hearer to derive an
open-ended set of weak implicatures that transcend the strong communication
of the idiomatic meaning.
Since idiomatic wordplay is a weak form of communication that leaves the
interpretative ball of generating weak contextual-effects in the hearer’s court,
the recognition of different forms of idiomatic creativity is to a considerable
extent dependent on its subjective evaluation by the receiver. In the same way
as idiom motivation can vary from speaker to speaker, the interpretation of alternative forms of idiomatic creativity are thus also dependent on the cognitive
environment that is constructed by a specific hearer in a specific usage-context.
In other words, it is theoretically impossible to define a strict demarcation
line between regular uses of idioms and idiomatic wordplay although this is
implicitly postulated by many theorists.
Nevertheless, to flesh out the cline of idiomatic creativity between the poles
of systematic variation and wordplay, i.e. to explain and evaluate the distribution, function and nature of a given idiom variant, it is possible to propose a
number of parameters that can be used to classify a given variational token.
In what follows, I will discuss alternative concepts by the help of which idiom
variants can be evaluated relative to three parameters:
1. The degree of conventionality
2. The degree of frequency and institutionalisation
3. The degree of intentionality
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.25 (1298-1349)
Chapter 6. Idiom variation and variability 
Based on these parameters, the following classification is not intended to postulate the objective existence of clear-cut idiom-variation classes. However, it
can provide points of orientation that make it possible to make qualitative distinctions on the idiomatic-creativity cline. In other words, in the same way as
compositionality, motivation and analysability provide orientational parameters to structure the continuum of idiom-transparency, the following idiomvariation classes can help us to structure the continuum of idiom variation.
.. Usual and occasional variation
In agreement with Burger’s classification of alternative types of idiom variants
(Burger 1998a: 25–27), I will use the terms usual variant and occasional variant to capture differences in the commonness and institutionalisation of idiom
variants (see also Burger et al. 1982: 68; Korhonen 1992: 50–51). Usual variants
are produced without a context-specific discursive intention; the are recurrent
and institutionalised and can thus be recorded lexicographically. For instance,
the lexical variation between bleed NP dry and bleed NP white is usual and
frequent. As an institutionalised pattern of variation, it can be predicted. The
notion of usual variation thus overlaps with the notion of systematic variation
(as described above) to the extent that it involves codified systematic departures
from an idiom’s base-form.
In contrast, occasional variants are created for the discursive purposes of
one specific usage-event; they are transitory and restricted to the contexts and
cotexts that they appear in. Therefore, they cannot be recorded or become
cognitively entrenched. Occasional variants thus include wordplay but are not
restricted to this variational class (see Burger 1998a: 150 for a classification of
occasional variants). To substantiate this point, we can compare the occasional
variants in (28) and (29):
(28) [...] if they come as close as Jed to upsetting the championship applecart.
(K5A: 702).
(29) Covers a multitude of chins. (Advertising slogan for WILLIAMS SHAVING CREAM).
The premodifier championship in (28) does not change the overall idiomatic
meaning of the expression, but it is inserted to connect the construction to
its context of use. In other words, the modifier functions as a topic-indicator
rather than anything else (see Section 6.4.3 below). Moreover, this variant
cannot be regarded as a creative exploitation or wordplay, because the orig-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.26 (1349-1397)
 Idiomatic Creativity
inal meaning of the idiom is preserved and strongly implicated. Apart from
the conspicuous, hybrid conceptualisation triggered by the novel NP championship applecart (i.e. the championship is like an applecart) the variant does
not trigger weak implicatures that go beyond its conventional modelling function. Thus, although the given occasional variant constitutes a momentary
and transient alteration of the base-form, it must be regarded as systematic.
As a consequence, systematic and regular variation extends into the realm of
occasional variation, which therefore cannot be equated with wordplay as is
too often done in theoretical approaches to idiom variation. This view is also
shared by Burger (1998a: 151), who claims:
Die Erweiterung hat in diesen Fällen nicht den Effekt eines Sprachspiels, sondern sie gibt eine Art Anweisung, wie man die phraseologische Ausdrucksweise in den wörtlichen Gedankengang zu “übersetzen” habe [...].
In contrast, the advertising slogan in (29) modifies the idiom cover a multitude
of sins (‘to compensate for or to mask sins, faults and shortcomings’) to create
a pun-variant through lexical substitution. The lexical alternant chins is quasihomophonous to sins. This makes it most likely that the base-form of this pun
variant is still recognised by the reader. The effected pseudo-literalisation has
the potential of triggering a great number of weak implicatures that directly
link the slogan to the product, namely, a type of shaving cream. The more or
less obvious implicatures of the pseudo-literal meaning, covers a multitude
of chins, are that the shaving cream is used by a great deal of men and that it
is economical because it is not used up quickly. To add further qualities to the
product, the advertiser depends on the idiomatic reading. Namely, the shaving cream is also very efficient in masking the ‘sin’ of a man’s negligence – the
sin of not shaving his beard. The implication of the idiomatic meaning is that
when the shaving cream is used, the shave is so clean that these shortcomings
are no longer visible. The pattern of ambiguation based on the process of occasional lexical substitution thus helps the advertiser to communicate the list of
the positive attributes of the product (Langlotz 2001a: 123). However, the advertiser clearly engages in weak communication here. While he or she does not
mention the qualities of the product explicitly, this leaves it to the consumer
to deduce a great number of weak implicatures from the pun variant to evoke
such qualities. The creative pattern-reforming process thus invites the readers
to join in an active game of meaning generation that binds them to the advert
and, maybe, to the product. Clearly then, the slogan must be seen as a intentional exploitation of the idiomatic base-form designed to create a semantic
surprise-and-binding-effect.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.27 (1397-1439)
Chapter 6. Idiom variation and variability 
On the basis of these insights, it becomes possible to establish a first interpretative criterion to delimit systematic occasional modifications from idiombased wordplays. Idiom variation must be regarded as wordplay if the idiom
is used to trigger weakly implicated semantic effects that go beyond the contextual integration of the idiomatic meaning. This, for instance, is the case
when the idiom is intentionally ambiguated or changed in a way that is inconsistent with its conventional idiomatic meaning. To refine this distinction
further, one can resort to Dobrovol’skij’s classificatory scheme, which is based
the classificatory parameter of intentionality.
.. Intentional vs. non-intentional variants
In his typology of idiom variation, Dobrovol’skij (1997: Section 2.3, see also
Dobrovol’skij 1999a) stresses intentionality as the most important interpretative parameter for idiom variation. On the basis of this criterion, three general
variation classes can be proposed: systematic variation (Variation im Rahmen
des Usus), wordplay (kreative Sprachspiele mit Idiomen) and erroneous variants
(Versprecher), (Dobrovol’skij 1997: 74). With this classification in hand, it is
possible to make a general distinction between systematic and non-systematic
variation.
Obviously, erroneous slips-of-the tongue such as the contaminations in
(30) must be excluded from an analysis of regular variability:
(30) a. the dust clears (the dust settles + the fog clears)
b. plain as day
(plain as a pikestaff + clear as day)
Such formal blends or contaminations of two synonymous or quasi-synonymous idioms can be commonly encountered in spoken discourse (cf. Langlotz
2004a). Usually, they are negligible as momentary deviations. However, the formal blends in (30) are also attested by the Collins Cobuild Dictionary of Idioms
and can thus be interpreted as having gained the status of usual variants.
Following Dobrovol’skij’s classification, a variant can be interpreted as
wordplay if it reflects the speaker’s conscious intention of creating specific
perlocutive effects. Such perlocutive effects correspond to the series of weakly
implicated conclusions that are intended by an advertising slogan such as covers
a multitude of chins. Unlike regular and systematic variability, the creation of
such effects is not limited by an idiom’s formal and semantic structure, but it is
almost unlimited (cf. Burger et al. 1982: 68). Dobrovol‘skij (1997: 74–78) compares two global subtypes of wordplay. On the one hand, idiomatic wordplay is
effected by ambiguation (see section above). On the other hand, Dobrovol’skij
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.28 (1439-1488)
 Idiomatic Creativity
points to exploitations of the image evoked by an idiom’s literal meaning. This
can be illustrated on the basis of the following example taken from Glucksberg
(2001: 73), (see also Section 2.2.5):
(31) Speaker A: Did the old man kick the bucket last night?
Speaker B: Nah, he barely nudged it.
In the reply, speaker B plays with the image evoked by the literal meaning of
kick the bucket. Instead of negating the idiomatic meaning (‘no, he didn’t die’),
the negation is effected by manipulating the literal scene in terms of an inferential analogy: kick the bucket : not nudge the bucket = die : not die. This
analogical modification is not prevented by the specific nature of the idiom’s
formal and semantic structure – kick the bucket is a notoriously opaque idiom – but rather depends on the accessibility of the literal scene in discourse.
Thus, in contrast to the processes of ambiguation, this variational process does
not create a striking semantic effect; the idiomatic meaning is fully preserved.
Obviously, the variant elaborates on the literal scene and thus creates the potential for a series weak implicatures that can be derived from this conceptual
resource. Apart from its obviously disphemistic stylistic-markedness, the modified force-dynamics of the novel scene has the potential to evoke a number of
weak conclusions: the old man was too weak to kick something, or, the old
man is not going to die suddenly, he will just pass away slowly. Note that the
generation of such implicatures from the conceptual background of the literal
scene is in full accord with the general cognitive-modelling function of idioms.
Nevertheless, (30) can be classified as wordplay, because the elaborative modification (he barely nudged it) could not trigger the idiomatic meaning if it were
used in isolation, i.e. without the previous activation of the idiom’s canonical
form and meaning. Put differently, in order to fulfil its intended communicative function or perlocutive effects (the negation of the idiomatic meaning),
the modification feeds on the previous recognition and conventional use of
kick the bucket. In this sense, it is creative, but not independent. I therefore call
this exploitation process parasitic elaboration.
From this discussion, three criteria can be derived to measure idiomatic
wordplay:
1. Stylistic markedness/conspicuousness: the formal structure of the base-form
is adapted in a striking or stunning way (as with covers a multitude of chins).
Often this creates a series of intended but weakly implicated contextual
effects. The hearer is invited the generate weak implicatures that transcend
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.29 (1488-1617)
Chapter 6. Idiom variation and variability 
the idiomatic meaning or add further semantic richness to it (e.g., he barely
nudged it).
2. Ambiguity: both the literal meaning and the idiomatic meaning are activated to refer to the context. Thus, both levels of meaning are strongly
implicated. This goes beyond the modelling function of idioms for which
the literal scene works as a semantic background, but is not evoked for
direct reference.
3. Context-dependency: idiomatic wordplay is intended by the speaker to invite the hearer to a game of sense-making which depends on the generation
of context-specific weak implicatures. Therefore, a pun variant can only be
understood in the usage-context.
Following these criteria, we can advance a preliminary definition of idiomatic
wordplay:
Idiom variation must be regarded as wordplay if a given idiom alternant is
used to trigger a series of weakly-implicated semantic or stylistic effects that go
beyond the systematic contextual integration of the idiomatic meaning in discourse. This is the case when the idiom is intentionally ambiguated or changed
so that it becomes stylistically marked and inconsistent with its conventional
idiomatic form and meaning. Wordplay can also be effected without affecting
the integrity of the idiomatic meaning through parasitic elaboration. This is
the case when a given pun variant cannot activate the idiomatic meaning independently, but depends on the previous recognition of the idiom in accordance
with its conventional usage.
In accordance with this definition, I would like to propose the following idiomvariation classes. Proceeding from the previous discussion, the scheme in Table 6.1 situates the distinctions between systematic vs. non-systematic and
usual vs. occasional variants, respectively, on a cline of idiomatic creativity.
Again, it must be emphasised that these classes constitute landmarks on a
gradual continuum rather than fully clear-cut distinctions. Although the classification refines the distinction between systematic variation and wordplay, it
nevertheless cannot prevent some degree of overlap between different variational classes, which would make an unequivocal description of controversial
cases possible. This is mainly due to the non-objective nature of intentionality. According to Harras (1983: 21) intentionality must not be regarded as an
intrinsic feature of any given action, but as an effect of observation. Put differently, to be able to characterise a given activity as intentional, it must first
be interpreted as being intentional by the perceiver. As a consequence, idiom
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.30 (1617-1649)
 Idiomatic Creativity
Table 6.1 Overview of idiom-variation classes
Occasional variation
a. systematic occasional variants effected by contextual
integration
b. intentional creation of pun variants through exploitation and
wordplay characterised by
(i) ambiguation
(ii) stylistic markedness
(iii) parasitic elaboration/context-dependency
c. non-intentional erroneous variants
(i) idiomatic slips-of-the-tongue
(ii) formal blends
Cline of idiomatic variability
b. systematic variants effected by the preservation of regular
morphosyntactic and syntactic variability (inflections, tense
adaptations, syntactic transformations, etc.)
Non-systematic varibility
Usual variation
a. usual variants
(i) institutionalised formal variants
(ii) institutionalised polysemy
Systematic variability
Frozenness
The lexicogrammatical structure of an idiom cannot be varied.
Pseudo variation
Creation of a pseudo variant: the idiomatic meaning is not activated only the literal
interpretation is possible
variants can only be categorised as differing in intentionality and purpose if
we attribute an echoic speaker-intention to the bare linguistic material. For the
interpretation of some variants a precise intentionality-based classification is
therefore difficult to make. Consider the following lexical substitutions in upset
the applecart that are attested by the BNC but not listed in idiom dictionaries:
(32) a. [...] until something happens to rock the applecart. (AAV: 371).
b. Rocking the energy applecart. (B74: 1556).
These alterations can hardly be regarded as wordplay because they do not create a striking semantic effect and are not based on parasitic elaboration. It is,
however, possible to analyse the variants as non-intentional contaminations
of upset the applecart with the quasi-synonym rock the boat. But although the
variational pattern can be viewed as an erroneous variant in consideration of
its surface structure, this interpretation seems hard to defend with regard to the
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.31 (1649-1697)
Chapter 6. Idiom variation and variability 
meaningfulness of the lexical alteration: the substitution of rock for upset seems
to create a causative meaning adaptation: by rocking an applecart, one may
cause it to be overturned or overthrown. Thus, one can characterise the lexical
substitution as a systematic and regular perspective-adaptation of the idiomatic
meaning. This said, the recurrence of the variation indicates that, although the
variant is not listed in dictionaries, the substitution cannot be described unequivocally as an occasional form of contextual integration, but reflects some
degree of commonness. How then can the variation be interpreted? Are we
faced with: a) a contamination, b) wordplay, c) an institutionalised usual variant that is not yet recorded idiom dictionaries or d) a systematic variant of
the base-form that mirrors the idiom’s systematic variability. This potential
for interpretative indeterminacy illustrates that the boundaries between different idiom-variation classes are open to interpretation and thus become fluent
rather than clear-cut.
. Principles of idiom variation
Systematic occasional idiom-variation and idiomatic wordplay do not proceed
according to uniform patterns. To develop a more sophisticated explanatory
framework to chart the distinction between systematic variation and wordplay,
it is therefore necessary to describe alternative idiom-variation strategies. So
far, idiom variation has been explained as the effect of idiomatic creativity –
the speaker’s ability to manipulate an idiom to encode a target conceptualisation in a communicatively functional way. In most cases, changes in the lexical
surface are therefore communicatively motivated: they are the effect of different intentional or unintentional variation strategies that underlie the creative,
context-specific activation of an idiom activation-set in discourse. Following
Sabban (1998), I call these variation strategies – variation principles. A variation
principle captures a distinct pattern of how an idiom is adapted in response to
the meaning(s) and communicative function(s) it is intended to encode.
Slightly modifying Sabban’s classification, I will discuss five basic variation
principles in terms of my cognitive-linguistic account of idiomatic creativity:
1.
2.
3.
4.
5.
Constructional adaptations
Literal-scene manipulation
Topic indication
Topic-related literal-scene manipulation (including conjunction variation)
Ambiguation (including punning)
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.32 (1697-1752)
 Idiomatic Creativity
This classification is motivated by the basic questions: How is the idiomatic
construction modified in response to the target? What is the precise relationship of the variant to the standard literal scene, the standard idiomatic meaning
and the context-specific target-conceptualisation, including the usage-context?
In cognitive-linguistic terms, these five principles constitute five prototypical
alternatives of how an idiomatic activation-set can be manipulated in the process of coding. Although the principles are salient, there is some degree of
overlap between them.
.. Constructional adaptations
Constructional adaptations include modifications of the base-form that change
the inflectional and syntactic structure of an idiom in a systematic way. Put
differently, the term constructional adaptation covers usual and conventional
grammatical variants that frequently recur in discourse. The notion can thus
be used as a counter-concept to the generative notion of transformational deficiency. For this study, the following constructional adaptations are of interest:
article variation, number variation, passivisation and fronting/topicalisation.
Consider, for instance, the variants of walk a tightrope in (33)–(35):
(33) North-West Thames regional health authority has several districts with
significant overspends, while South-East Thames is facing a potential £5m
shortfall with many of its 15 districts ‘walking a tightrope’. (A1J: 308).
(34) That tightrope was walked every weekend when Charles and Diana took
the train with their nanny from Norfolk to Liverpool Street station in
London where their mother met them. (ECM: 82).
(35) The Khmer Rouge is aware of the tightrope it has to tread, both with the
West and with Vietnam. (A45:159).
(33) involves a fully grammatical use of walk a tightrope. The only adaptation
process featured by this variant consists in adapting the verb to the temporal frame of the target conceptualisation (15 districts are currently acting in very difficult circumstances). (34) and (35) reflect instances of
fronting and article variation. (34) additionally features passivisation. Since
walk a tightrope is an isomorphic idiom, these adaptations can be readily explained on the basis of the phrase-induced figurative senses of the constituents
tightrope (‘very difficult situation’) and walk (‘deal with’). The variants can
be directly understood when the constituents are replaced by these figurative correspondents. Thus, these examples provide clear evidence against a
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.33 (1752-1791)
Chapter 6. Idiom variation and variability 
semantic-unit view of idioms, but support the present account of idioms as
complex scenes. Following our description of idiomatic creativity, the literal
scene (agent walks a tightrope) works as a mediating semantic background
relative to which these constructional adaptations are sanctioned. The variants are fully grammatical because the corresponding adaptation of the literal
scene is consistent with the motivated and isomorphic association of the walk
a tightrope-scene with the deal with/act in a difficult situation-scene.
This motivation is structured by the conceptual metaphors basis/context of
activity/development is a path, quality of basis/context is quality of
path, stability is physical balance and failure is (falling) down.
In sum, constructional adaptation is a systematic idiom variation principle
that covers regular morphosyntactic and syntactic adaptations of an idiom’s
base-form. These adaptations are claimed to be supported and constrained,
respectively, by the internal semantic structure of the idiomatic configuration. This hypothesis is to be tested in the second analytical part of this study
(Chapter 7).
.. Literal-scene manipulation
Literal-scene manipulation triggers a contextually motivated adaptation of the
idiomatic meaning by adapting the literal scene for the purpose of coding the
target conceptualisation efficiently. Thereby, the variation is comprehensible
on the basis of the idiomatic configuration alone, i.e. without its prior evocation in the usage context. The following straightforward example can be used
to illustrate this principle:
(36) The Chancellor had a narrow tightrope to walk and he managed to please
a variety of people. (CEL: 589).
The alteration of walk a tightrope in (36) involves adjectival premodification to
create the variant a narrow tightrope to walk (for simplicity’s sake, I am ignoring
the non-prototypical syntactic evocation of the idiom here). Narrow functions
as an adjective of quality; it specifies an inherent property of the nominal head
(tightrope). The premodifier is fully compatible with the literal scene: the NP,
a narrow tightrope, evokes a fully consistent image that is richer than the standard image: walk a tightrope. The newly computed literal-scene of the variant
thus works to elaborate and intensify the pre-coded scene of the base-form. By
analogy with this literal scene manipulation, the idiomatic meaning is intensified in parallel (‘The Chancellor had to act in a very difficult situation’). Again,
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.34 (1791-1849)
 Idiomatic Creativity
this semantic intensification of the idiom is guided by the set of conceptual
metaphors that motivate the construction (see above).
According to Dobrovol’skij, literal-scene manipulation must be clearly regarded as a form of wordplay. (To substantiate his argument he refers to a study
by Napoli 1988):
Die potentielle Bereitschaft der Kommunikationsteilnehmer, die bildliche
Komponente des Idioms auszubauen sowie die literalen Bedeutungen der
Einzelkonstituenten als selbständige semantische Entitäten zu interpretieren,
bringt nahezu unbegrenzte Möglichkeiten eines kreativen Umgangs mit Idiomen mit sich. Vgl. dazu den ‘extension test’ von Napoli (1988), der eine
prinzipielle Interpretierbarkeit von Äusserungen wie The cat got out of the bag
and wreaked havoc nachweist. ‘Extendability can call for building up a story
around the idiom which is plausible at the metaphorical level’.
(Dobrovol’skij 1997: 75)
Disagreeing with Dobrovol’skij, I claim that it is not possible to classify all
forms of literal-scene manipulation as wordplay. Literal-scene manipulation
is neither context-dependent nor does it trigger ambiguation. Therefore, only
stylistic conspicuousness can work as a measure to qualify it as wordplay. With
regard to this criterion, the adjectival premodification in (36) is relatively inconspicuous. The speaker/writer certainly did not have the intention to trigger
a striking semantic effect or to suggest weak implicatures that transcend the
conventional figurativity of the idiom. The relative gradability of the conspicuousness criterion can be observed when we compare (36) with the literal-scene
manipulations in (37) and (38):
(37) The pub has always had to tread a difficult tightrope in reconciling its social
function, as a ‘house’ for the public, with its commercial function as a
retail shop. (A0B:97).
(38) Only the utter ruthlessness of one ravaged, machine-sustained tyrant and
the overstretched forces of his fierce yet fragile Imperium kept the human
race tottering along its fraying tightrope. (CM4: 1293).
The adjectival premodifications in both (36) and (37) are occasional. However, they are so inconspicuous that they can hardly be termed wordplay. (37)
further features the usual lexical substitution tread for walk. Only for (38)
can one claim a highly creative adaptation of the idiom’s literal scene. Thus,
this variant is the only candidate for which the term wordplay is appropriate.
Note, however, that neither the adjectival premodification fraying tightrope nor
the lexical substitution totter for walk is exceptionally different from the alterations underlying (36) and (37). The conspicuously creative character of (38)
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.35 (1849-1914)
Chapter 6. Idiom variation and variability 
emerges from an accumulation of lexicogrammatical changes rather than from
fundamentally different formal adaptations. Put in cognitive-linguistic terms,
the deviations in (36)–(38) differ from the base-form in terms of elaborative
distance: while (36) and (37) are relatively close to the base-form, the degree of
deviation in (38) is more striking.
From the cognitive-linguistic perspective of creative coding, elaborative
distance is the effect of qualitative and quantitative changes along the coding
parameters of the pre-coded literal-scene. More precisely, elaborative distance
is created by:
a. radically changing the scope of the literal-scene: e.g., let the cat out of the
bag vs. the cat got out of the bag and wreaked havoc (see above quotation),
b. construing the scene by extension from the basic-level (or the typical level
of abstraction at which the base-form is construed): e.g. totter along the
fraying tightrope, call a spade a manually powered entrenching instrument.
c. changing the force-dynamic relationships in a striking way: let the cat out
of the bag vs. the cat got out of the bag and wreaked havoc; put a pigeon
among the cats.
For literal-scene manipulation, the evaluation of a variant as wordplay is thus
dependent on its degree of deviance from the typical construal of the baseform. Following Carter’s distinction between core and non-core vocabulary
(Carter 2004: 115–119), we can regard elaborative distance as the result of using lexical items and constructions that mark a conspicuous departure from
the idiom’s neutral coding-baseline in a given speech situation:
It is [...] clear that the clines which run from core to non-core vocabulary
represent a continuum of expressive possibilities for making utterances more
intimate, as well as for intensifying and evaluating utterances. At such points
in the discourse, speakers are doing more than convey information or transmit
ideas; they are allowing their own attitudes to emerge, they are interacting with
others, they are signalling affective and expressive responses and accordingly
they are making vocabulary choices which often creatively encode their own
feelings.
(Carter 2004: 117)
This statement is in full accordance with the cognitive-linguistic view of idioms
as cognitive micro-models. The manipulation of the literal scene of an idiom is
a salient strategy to fine-tune the model, i.e. to enhance its modelling function
of evaluative and affective concretisation and to trigger additional intimacy
effects. By re-construing the literal scene by means of conspicuous, non-core
vocabulary, its conceptual and emotional entailments can be further exploited
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.36 (1914-1955)
 Idiomatic Creativity
to trigger a series of additional weakly implicated premises and conclusions.
Accordingly, the image of tottering along a fraying tightrope in (38) adds further conceptual detail to the literal scene to widen the pool of its potential
conceptual and emotional implicatures. Moreover, such a marked elaboration reasserts the intimate sense of community between the speaker and the
hearer since it presupposes the mutually-known neutral base-form as a shared
starting point for understanding. In other words, conspicuous literal-scene manipulation playfully highlights and enhances the modelling function of idioms
beyond their conventional pre-coded figurativity and it creates intimacy effects.
This said, it must be re-emphasised that the demarcation-line between systematic adaptations and playful ones is a matter of gradual differences rather than
a clear-cut threshold.
In short, I claim that literal-scene manipulation covers a variety of lexicogrammatical alterations that range from being fully systematic to being
strikingly unconventional (as in The cat got out of the bag and wreaked havoc).
Moreover, I claim that the degree to which literal-scene manipulation is systematic is dependent on the coherence of the metaphorical association that
motivates an idiom. In other words, I suggest that systematic literal-scene manipulation is highly restricted for opaque idioms, while it is more open for
motivated idiomatic constructions. On the basis of this criterion, parasitic
elaboration – which can itself be regarded as a form of literal-scene manipulation – can be clearly defined as wordplay. In the second analytical part of this
book, the relationship between idiom motivation and systematic literal-scene
manipulation will be tested empirically.
.. Topic indication
This principle of idiom variation can be introduced by comparing it to the
preceding one on the basis of a syntactically analogous example:
(39) That sum may seem like a lot of lei (the Romanian currency that purchases
next to nothing abroad) but it still left the Romanians treading a financial
tightrope. (A9R: 7).
In contrast to narrow tightrope in (36), the NP a financial tightrope in (39) does
not create a meaningful literal scene. What could a financial tightrope possibly be? Financial certainly cannot be understood as an inherent, qualifying
adjective. Rather, when considering the literal scene only, its semantic contribution is ungrammatical. This variation principle is thus characterised by the
absence of a consistent literal meaning. However, the use of the premodifier
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.37 (1955-2004)
is no longer problematic if one takes the idiomatic meaning into account: on
the figurative level, the contribution of financial is fully grammatical: act on an
unstable financial basis / go through a difficult financial situation. Thus, financial
does not modify the literal sense of tightrope, but it specifies its phrase-induced
figurative sense ‘unstable basis /difficult situation.’
Following, the terminology used to describe the semantic structure of
metaphors (see Section 3.3.1), literal-scene manipulation is vehicle-related. The
vehicle underlying the idiom’s metaphorical structure is made more specific
by describing the literal scene in more detail (narrow tightrope). In contrast,
figurative-level modification (financial tightrope) specifies how the idiomatic
meaning must be interpreted in the specific usage-context. In this sense, financial is topic-related. I will therefore call this idiomatic type of figurative-level
modification topic indication (cf. also Goatly 1997: Sections 6, 7, 8). The topicindicator financial restricts the denotational range for the interpretation of the
idiomatic-meaning in the specific usage-context.
Clearly, topic indication is a context-specific phenomenon. In this sense,
it differs from literal-scene manipulation. But although the occurrence of the
topic indicator cannot be predicted outside the context of use, this variational
constituent is integrated into the idiom’s formal structure according to general grammatical rules for adnominal modification and lexical substitution.
This variation principle must therefore be described as a fully systematic type
of occasional variation, rather than a form of wordplay. This interpretation
is even advocated by Nicolas (1995: 248), who regards all other forms of occasional idiom-variation as wordplay (also cf. Burger’s statement in Section
6.3.1 above).
Since the topic indicator applies to the idiomatic meaning, it must be embedded within the constituent-structure in a way that is compatible with the
figurative level. Hence, the speaker has to know how the literal meanings of the
constituents can be mapped onto the idiomatic meaning – he or she must know
their phrase-induced figurative senses. In other words, for topic indication to
work, the idiomatic configuration must be potentially isomorphic. Thus, topicindicating lexical substitution and modification can be hypothesised to be only
possible for corresponding idioms. Again, this will be scrutinised in the second
analytical part of this book.
.. Topic-related literal-scene manipulation
An idiom’s topic cannot only be addressed by blending alien topic indicators
into the constituent structure. Rather, the literal scene can also be related to the

JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.38 (2004-2069)
 Idiomatic Creativity
informational structure of the target conceptualisation by manipulating it in a
way that leads to a consistent topic-related literal-scene. Consider the following
example (reproduced from Moon 1998: 51), which stems from a review of a
television comedy:
(40) His old friend and partner in crime Georgio Bertoli (Steve O’Donnell)
talks to the police in return for personal immunity and a Continental
breakfast. Alarmed by this turn of events the Grand Master of the Freemasons (Nosher Powell) employs two inept hit men, Mig (Tim McInnerny)
and Mog (Alexei Sayle), to kill Bertoli before he can spill the pasta.
Here, the literal scene of spill the beans is adapted to the Italian-mafia-gangster
context. Thus, the substitution of pasta for beans relates the literal meaning
to the actual context of the target conceptualisation (Georgio Bertoli reveals
information about the Italian gangster-scene to the police). This topic-related
form of literal-scene manipulation does not alter the idiomatic meaning, which
clearly distinguishes it from variation principle 2. Moreover, unlike topicindication, this variation preserves the conceptual integrity of the literal scene:
spill the pasta denotes a process that is fully consistent in itself. In other words,
the variational strategy involved with this type of idiom variation consists in
finding a linguistic unit that can be associated with both the usage-context and
the conventional literal scene. In the present case, the noun pasta can be directly related to the Italian-mafia context and – via the hyperonym food – it
can also be linked to beans. This makes it possible to recognise spill the pasta
as a variant of spill the beans. In this way, the writer guarantees that the conventional idiomatic meaning can still be activated by the hearer although the
idiom’s standard literal-scene is modified.
Topic-related literal-scene manipulation is also involved in what Ernst
(1981: 52) termed conjunction variation in his analysis of premodifying adjectives in idioms:
(41) (from an article on the making of the movie Jaws)
Bruce, a shark, found it a part he could really sink his three rows of teeth
into. (Ernst 1981: 52, italics by al).
(42) (from a review of a production of the Twelfth Night)
Malvolio deserves almost everthing he gets, but ... there is that little stab of
shame we feel at the end for having had such fun pulling his cross-gartered
leg for so long. (Ernst 1981: 52, italics by al).
Here, the premodifiers (three rows of and cross-gartered) specify the idioms’
nominal heads (teeth and leg) directly, i.e. they modify the underlying literal
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.39 (2069-2115)
Chapter 6. Idiom variation and variability 
scenes in a consistent way. In this sense, they are different from the topicindicating adjective in (39) and similar to the literal-scene manipulating adjective in (36). However, unlike the use of narrow in (36), these premodifiers
cannot be attributed a corresponding figurative interpretation on the level of
the idiomatic meaning. Rather, they can only be interpreted literally. Thus,
cross-gartered in (42) applies literally to Malvolio’s real leg and three rows of
(teeth) in (41) describes the real shark’s most scary feature rather than anything
else. These modifiers therefore work as conjuncts that are integrated into the
idioms’ formal structure, but must be interpreted literally outside the idiomatic
configuration (Ernst 1981: 60). Thus, the conjuncted premodifiers do not alter
the idiomatic meaning at all, but provide additional information about the
idiom’s context-specific referents. Thus, (41) and (42) can be rendered as:
(41) a.
Bruce, a shark, has three rows of teeth and found its role a part he could
sink his teeth into (= he could become deeply involved with).
(42) a.
Malvolio has a cross-gartered leg and we feel ashamed for having pulled
his leg for so long (= for having teased him for so long).
The highlighted phrases in (41a) and (42a) are the conjuncts that extend the
respective literal meanings and intertwine them with their referential contexts.
At the same time, the idioms as such, i.e. read without the premodifiers, must
be interpreted according to their standard idiomatic meaning. In the same way
as pasta in (40), the conjuncts thus also work as vehicle-related topic-indicators
(cf. also Sabban 1998: 210).
Although this variation principle can be seen as a combination of the
systematic principles 2 and 3, topic-related literal-scene manipulation can be
clearly classified as wordplay. This form of variation is highly conspicuous,
fully context-dependent and non-predictable. Moreover, it communicates its
own creativity. For these reasons it is different from inconspicuous literalscene manipulation, which systematically exploits the idiom’s inherent figurative potential irrespective of the usage-context. Moreover, it is different from
topic-indication because the topic-related adaptation of the literal-scene does
not obey systematic rules. In addition, this principle involves ambiguity: the
conjuncted parts of the variant (pasta, three rows of teeth, cross-gartered leg)
have to be interpreted literally, while the underlying idiom-standard must be
understood idiomatically.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.40 (2115-2157)
 Idiomatic Creativity
.. Ambiguation and punning
As we have illustrated on the basis of have a good nose and covers a multitude of chins (see examples (12) and (29) above), ambiguation and punning
involve the creative evocation of multiple referentiality. Therefore, this contextdependent variation principle can be clearly characterised as a form of wordplay. Ambiguation is also triggered by the following structural variant of be
home and dry (having achieved success or a victory).
(43) In the dying minutes Bradford had Lee Sinot send off for arguing, but by
then Swindon were home and well almost dry. (KRM:1957).
This passage concludes the report of a football match which took place under
very bad weather conditions. This meteorological context provides the motivating ground for the pun variant home and almost dry. Home and dry can be
classified as an unmotivated, opaque idiom with a non-isomorphic semantic
structure. Having no figurative-level correspondents, the constituents cannot
be modified systematically. In the present case, however, dry is qualified by
the adverbial downtoner almost. Almost dry must be interpreted literally as
an ironic description of the players’ physical condition after the match. This
literal meaning is activated through semantic priming. The bad-weather context brings the literal meaning of the constituent dry to the foreground. This
ironic resemantisation triggers ambiguity: ‘Swindon were successful, but the
team’s players were drenched because of the terrible weather.’ The variant thus
creates a unconventional and humorous semantic-effect and must therefore
be characterised as wordplay. Ambiguation through punning thus creates a
novel literal meaning (home and almost dry). This meaning must not be interpreted idiomatically, but has its own reference in the usage-context. In the
pun variant, the idiom’s base-form can still be recognised. This makes it possible to co-activate the idiomatic meaning with the novel literal meaning. As
a result, the variant becomes ambiguous between the conventional idiomatic
reading and the novel literal reading. Ambiguation causes the hearer to fluctuate between the different referential meanings produced by the pun variant.
This switching from one meaning to the other causes an abrupt change of perspective. The abruptness of this mental process is puzzling and results in the
evocation of intended but weakly implicated emotional and intellectual effects.
These contextual effects must be actively derived by the hearer and thus reflect
an idiomatic form of playful, weak communication. The reader experiences the
variant as surprising, original or funny (cf. Langlotz 2001a; Hemmi 1994).
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.41 (2157-2206)
Chapter 6. Idiom variation and variability 
After having charted this complex spectrum of idiomatic variability, it is
necessary to change perspective again and to scrutinise the pole of frozenness.
. Cognitive constraints on idiom variation
Above, I have claimed that variability and frozenness constitute two counterforces that determine an idiom’s lexicogrammatical behaviour. Turning back
to the pole of frozenness, we must address one of the most important issues in
the analysis of idiom variation, namely, the question of its limitations: By what
cognitive factors and forces are these alternative classes, types and principles of
idiom variation constrained and thus make them subject to relative frozenness?
Obviously, any novel grammatical construction can only fulfil its communicative purpose effectively as long as it can be understood by the communicative partner. In other words, the novel construction must be communicatively
functional and so conventional that the hearer is still able to decode it with
reference to the context and the linguistic schemas stored in his or her mind.
Since very creative adaptations of idioms depend on the hearer’s ability to derive a variety of weak contextual effects, the speaker’s attempt to maximise the
relevance of the variant amounts to a communicative tightrope-walk that is
subjected to a number of limitations.
In this section, I would like to show that multiple factors work as constraints on idiom variation. Not all of these constraints apply to all idiomvariation strategies in the same way. Thus, for the attempted analysis of systematic idiom-variation, it is most important to answer the question: What global
constraints restrict the systematic variability of a given idiom in discourse?
.. Recognisability
Recognisability is the most general variation constraint. It applies to all
types of idiom variation and defines the boundary between idiom variation
and pseudo-variation (see Table 6.1). According to Burger et al. (1982: 68),
the context-specific variability of an idiomatic expression is virtually unconstrained:
Im Rahmen der Untersuchung transformationeller Defekte von Phraseologismen gingen die Generativisten recht sorglos mit dem Kriterium “das geht /
das geht nicht”, also mit der Zuordnung von Grammatikalitäts- bzw. Akzeptabilitätsgraden zu einzelnen Wortverbindungen um. [...]. Aufgrund der Synopse unseres Belegmaterials aus verschiedensten Textsorten gesprochener und
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.42 (2206-2252)
 Idiomatic Creativity
geschriebener Art und aufgrund der sonst in der Literatur zitierten Belege
muss man wohl sagen: es gibt kaum eine Veränderung eines Phraseologismus,
die in irgendeinem Kontext nicht möglich und durchaus sinnvoll wäre.
Burger et al. are right in pointing to the fact that idiom variation is, fundamentally, a phenomenon of idioms-in-use and thus highly context-dependent.
Moreover, their general observation is certainly correct: any idiom can be
subject to any type of lexicogrammatical alteration if one does not only refer to systematic idiom-variation, but also includes idiomatic wordplay. Of
course, in generative frameworks, idiomatic wordplay has generally been excluded from theoretical description and disclaimed as a matter of performance
without, however, proposing any criteria on the basis of which this exclusion
could be justified. While agreeing with Burger et al.’s statement, one should
nevertheless concede that the variation potential of an idiom is not completely unconstrained but minimally dependent on the partial recognition of
its base-form (cf. Barz 1986: 322; Burger 1998a: 154). The speaker must find a
balance between manipulating the idiom to adapt it to his or her communicative intentions and the hearer’s presupposed ability to recognise the variant
as a departure from the idiomatic standard-configuration. We can thus define
recognisability as a first variation constraint. Recognisability defines the maximum degree to which the variant may deviate from the base-form, while still
performing its idiomatic function.
With reference to the notion of immanence, recognisability can be characterised in cognitive-linguistic terms. In Section 3.1, I have outlined the cognitive ability to recognise a triangle by the help of the top-down projection
of the corresponding mental triangle-schema even if the underlying visual
stimuli are deviant and incomplete. The recognition of the geometrical figure is triggered by our cognitive ability to activate the schema on the basis
of substructures that are immanent to both the visual stimuli and the mental triangle-image. In the same way, the base-form of an idiom constitutes a
complex mental configuration. By analogy with the deviant triangle, the idiom
variant can be recognised as an instance of a given idiom if the cogniser is able
to activate linguistic substructures that are immanent to both the standard idiom activation-set and the variant. The recognition of an idiom in discourse
can thus be explained on the basis of the same set of general cognitive processes as the processing of visual or any other mental input. The fundamental
difference between the recognition of visual and linguistic input lies in the specific quality of the schemas that are activated when the stimuli are processed.
In other words, the key to the question of idiom-recognition lies in the seman-
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.43 (2252-2300)
Chapter 6. Idiom variation and variability 
tic and structural complexity of the idiomatic configuration with its inherent
activation probabilities and salient substructures. But by what specific factors
is idiom recognition guided?
... Activation probabilities
To be stored as a complex configuration with unit status, an idiomatic
activation-set is dependent on the strong mental associations between its constituents. In psycholinguistics, the strong co-activation of informational units
is called transitional probability (see Shannon & Weaver 1949; cf. also Burger
et al. 1982: 179–180). Since constituents such as grasp and nettle are the symbolic components that constitute the complex unit grasp the nettle, they both
carry the potential to trigger the idiom’s idiomatic meaning. Often, the mutual
evocation of parts of the idiomatic transition-chain is sufficient to activate the
entire idiomatic configuration. (Note that this in full accord with the concept
of key as proposed by the configuration hypothesis). This can be illustrated
by means of a trivial experiment. Consider the following example. What is the
most probable continuation of the following sequence of words?
(44) When in Rome ....
If you are able to complete this string by adding do as the Romans do, you have
attested the existence of the corresponding proverb in your cognitive grammar.
Further, you have experienced its representation as a probabilistic unit because
you were able to retrieve the whole configuration on the basis of fragmentary
input only. Having a great deal of proverbial expertise, you can certainly decode
the following familiar string as well (see Langacker 1987: 430):
(45) He who laughs least, laughs best.
When activating this proverb, you have possibly read over the fact that least
deviates from the original form: He who laughs last, laughs best. If so, you have
directly activated the proverb as a mental configuration after reading over the
first of its immanent lexical components. You have ignored the alternant use
of least because you have activated the whole proverb as a mental schema and
projected this schema top-down onto the incoming information before you
scrutinised completely the words underlying the present variant. If, in contrast, you have discovered the deviation while still being able to recognise the
proverb, then the combined activation of the lexical units through bottom-up
processing and the top-down projection of the proverb schema have made it
possible for you to recognise the string as a pun on the proverb. In other words,
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.44 (2300-2354)
 Idiomatic Creativity
you have performed a recognition process highly similar to the recognition of
the deviant triangle.
In sum, the activation probabilities, which pattern an idiomatic configuration as an entrenched network of mental substructures, make it possible to
recognise an idiom even if the lexical stimuli evoked by the variant do not fully
correspond to the base-form. In what follows, it will be shown what formal
and semantic characteristics of idioms may support the direct stipulation of an
idiomatic configuration.
... Salient idiomatic substructures
The previous experiments would not have worked by replacing any constituent. Thus, the fragments ... in Rome ... or he .... last ... are considerably less
likely to trigger the respective proverbs. One must therefore assume that some
structural characteristics have a higher probability to activate the idiomatic
activation-set than others. I would like to call the most typical structural features of an idiom its salient idiomatic substructures. Note that this term is in full
accord with the concept of key as proposed by the configuration hypothesis.
Following our definition of idiomatic standards as idiomatic activation-sets,
salient idiomatic substructures thus define the most distinctive features of these
mental networks. By analogy with the triangle, salient idiomatic substructures
are similar to the configuration of angles that demarcates the triangle schema.
Any triangle can be recognised on the basis of this salient pattern alone without
being dependent on the full perceptual presence of its three sides.
Since uniqueness increases noticeability, cranberry morphs and idiosynractic combinations of lexemes as in go places or trip the light fantastic are very
good candidates to work as salient idiomatic substructures. Again, this can
be tested by means of very simple experiments. Can you retrieve the idioms
corresponding to the following nouns?
(46) applecart, bandwagon, cropper, kibosh, gaff, mountain, tree, hand, blood.
While it is relatively easy to find the idiomatic strings corresponding to the
cranberry morphs (applecart, bandwagon, cropper, kibosh, gaff ), frequent idiom
constituents (mountain, tree, hand, blood) are less distinctive and are therefore
unlikely to isolate one specific idiom from a set of other possible candidates
in the cognitive grammar. The latter constituents, however, become distinctive
once they co-occur together with other lexical units with which they pattern a
specific idiom:
(47) a.
mountain + molehill (→ make a mountain out of molehill)
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.45 (2354-2417)
Chapter 6. Idiom variation and variability 
b. bark + tree (→ bark up the wrong tree),
c. hand + till (→ have your hand in the till)
d. sweat + blood (→ sweat blood)
With a view to the cognitive-linguistic account of the semantic structure of
idioms, this co-occurrence of core lexical constituents functions as a form of
literal-scene setting. It can be assumed that those lexical constituents that carry
the highest probability to activate the literal scene and with it the idiomatic
meaning are the best literal-scene setters. If a lexical constituent alone is insufficient to unfold the idiomatic configuration, its well-entrenched co-occurrence
with a co-constituent creates a high potential to activate it.
In sum, four formal characterisitcs of idioms can be said to be highly salient
and thus work to increase their recognition-probability in variational usages
(see also Sabban 1998: 114):
a. cranberry morphs: cropper, kibosh, etc.
b. highly salient lexical constituents: bandwagon, bitter pill, tightrope, etc.
c. salient constructional features (idiosyncrasies or non-prototypical constructional schemas): pastures new, go places, coming up roses, etc.
d. the co-occurrence of core lexical constituents: spill + beans, kick + bucket,
grasp + nettle, etc.
... Conceptual priming
Conceptual priming works as a further factor to increase the recognition probability of a varied idiomatic construction. Above, we have already discussed the
pattern of variation involved in the following parasitic elaboration:
(48) Speaker A: Did the old man kick the bucket last night.
Speaker B: Nah, he barely nudged it.
The possibility to negate the opaque idiom kick the bucket by altering its lexicogrammatical structure completely, illustrates the importance of the literal
scene for the creation of a variant. Supposing that the two speakers know the
idiom and have already talked about the old man’s state of health, the idiomatic
meaning of kick the bucket can be directly retrieved. However, the corresponding literal scene (kick the bucket) underlies the evocation of the idiomatic
meaning. Otherwise, it would not be possible to negate the idiomatic meaning
by using the symbolic unit nudge. More specifically, on the basis of the literal
conceptualisation, the idiomatic meaning can be altered by coding the literal
scene differently. The variant he barely nudged it reconceptualises it by changing the impulsive-force-schema implied in the concept kick. This works by
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.46 (2417-2463)
 Idiomatic Creativity
replacing kick by nudge. This reconceptualisation is then coded by the formal
structure of the variant barely nudge it, which is completetly different from kick
the bucket on the level of form. In other words, for this specific instance of
variation to work, the recognition of kick the bucket in the variant is mediated
on purely conceptual grounds. The example therefore nicely illustrates that
conceptual priming works as an additional factor to support recognisability.
In short, recognisability is the most fundamental constraint for idiom variation. Due to recognisability idioms may be varied at all. If an idiom variant
is not recognisable as idiomatic, the idiomatic activation-set cannot be activated. As a consequence, the variant becomes non-idiomatic or meaningless.
In the analytical part, the recognisability constraint will be further scrutinised
with a view to the question in how far idiom-transparency supports idiom
recognition.
.. Functionality
While for the comprehension of idiomatic slips-of-the-tongue it may be sufficient to recognise the idiom within the erroneous variant, this variation constraint alone is not enough to motivate wordplays. Reconsider the pun variants
cover a multitude of chins and home and almost dry (see examples (29) and
(43) above). We have interpreted both of these variants as wordplay. Thus, they
were created by the speaker for the purpose of creating perlocutive effects: the
advertising slogan covers a multitude of chins was presumably designed to provoke surprise in order to communicate the positive attributes of the product;
for home and almost dry the author’s (echoic) intention was to create ironic
humour to reduce the soberness of the reported match result. Apart from
recognisabilty, one must therefore assume an additional constraint to guide
the creation of these pun variants: functionality. Functionality limits the intentional adaptation of an idiom to fulfil a specific communicative purpose in
a given communicative situation. This constraint thus leads to the qualitative
distinction between non-intentional, non-functional slips-of-the-tongue and
idiomatic wordplay, which must always be functional by definition.
Since creative idiomatic coding is a complicated communicative act that
presupposes the hearer’s willingless and ability to join in the game of deriving
weak implicatures, the variant’s functionality is always subject to the hearer’s
subjective interpretation of its communicative purpose. In other words, the
functionality criterion is dependent on considerable overlap between the
speaker’s and hearer’s cognitive environments. If similar cognitive effects underlie the speaker’s intention and the hearer’s set of derived implicatures, then
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.47 (2463-2517)
Chapter 6. Idiom variation and variability 
the variants functionality becomes mutually manifest. If, however, the hearer
derives contextual effects that were not intended by the speaker, their impressions of the variant’s functionality can diverge. In other words, functionality is
a considerably subjective criterion.
Of course, functionality also constrains systematic adaptations. However,
functionality is not a sufficient criterion to explain systematic variability.
.. Compatibility
Compatibility is the most important constraint for the creation of systematic
variants. It points to the requirement that any formal modifications of the literal scene must be in accordance with the idiomatic meaning. For example,
kick in kick the bucket is a verb that prototypically describes a momentary action (cf. Quirk et al. 1985: §4.35). In contrast, die refers to a transitional event.
This semantic discrepancy between the two processes blocks certain tense and
aspect adaptations of the verb on the literal level. While it is possible to say
John is dying (describing an anticipatory interpretation of the event), the adaptation of kick to fit this conceptual refinement is not acceptable; the continuous
form is kicking the bucket can only be used to describe the repeated, iterative recurrence of this momentary action. However, repetition does not make sense
relative to the process of dying. Therefore, the idiom cannot appear in the progressive (cf. Glucksberg 1993: 20). In the same way, the use of the simple present
is restricted due to incompatibility. One major function of the simple present
is to frame habitual actions and events (habitual present). Thus, John kicks the
bucket in (49a) implies that this action is performed frequently. Again this does
not make sense relative to the idiomatic meaning. Consequently, the use of the
simple present is restricted to occur in combination with plural subjects only
(49b). In this syntactic context the simple present of kick can describe a general
state of affairs, which makes sense relative to the idiomatic meaning:37
(49) a. *John kicks the bucket
b. Too many soldiers kick the bucket in unnecessary wars.
Apart from tense and aspect adaptations the compatibility constraint also
prevents this idiom from modifying or pluralising the NP. The bucket does
not have a conceptual correspondent on the level of the idiomatic meaning.
Therefore, any adaptations that modify the denotational range of the bucket
are incompatible with the idiomatic meaning die. In the analytical part, this
correlation of systematic variation with the compatibility constraint will be
subjected to close scrutiny.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.48 (2517-2572)
 Idiomatic Creativity
.. Non-ambiguity, non-conspicuity and evocational autonomy
To set off systematic variation from idiomatic wordplay, three further constraints must be postulated: non-ambiguity, non-conspicuousness and evocational autonomy.
To exclude patterns of ambiguation and conjunction variation from systematic idiom variation, the (pseudo-)literal scene of the idiom variant must
not establish direct reference to the target conceptualisation. On the basis of
this constraint, wordplays such as covers a multitude of chins and home and
almost dry can be described as unsystematic because they are multi-referential.
When discussing the conventionality of literal-scene manipulation in Section 6.4.2, we noticed that fairly conspicuous adaptations of the literal scene
through lexical substitutions and extensive NP-modifications may create stylistic effects that make one perceive these variants as strikingly deviant from the
base-form. Although conspicuousness is a gradual and imprecise criterion, it
can be added here to distinguish ‘neutral’ literal-scene manipulations from
highly conspicuous ones.
Finally, we have claimed that parasitic literal-scene manipulations must be
described as wordplay, because they depend on the previous priming of the idiomatic activation-set. Thus, although the extension did he kick the bucket – no,
he barely nudged it is non-ambiguous and fully compatible with the idiomatic
meaning, it must be excluded from the description of systematic idiom variants
because it is not evocationally autonomous.
.. Grammaticality
Finally, I will use the term grammaticality to highlight the fact that any variation of the formal structure of any idiom must be in accordance with the
grammar of the language in consideration. This constraint was already addressed in early studies on transformational restrictions (see, e.g., Burger 1973;
Newmeyer 1974; Fleischer 1982). In cognitive-linguistic terms, grammaticality
demands that the formal structure of an idiom variant must accord with the
general constructional schemas provided by the language. Grammaticality thus
amounts to the constructional and conceptual consistency of the literal scene.
Since the literal scene is evoked to organise the target conceptualisation, its
symbolic structure must remain inherently consistent. For instance, the grammaticality constraint demands that the literal scene cannot be construed in the
passive if the verbal constituent that profiles it cannot be passivised.
(50) have a screw loose → *a screw was had loose
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.49 (2572-2704)
Chapter 6. Idiom variation and variability 
The grammaticality constraint even applies if the passive were compatible with
the idiomatic meaning:
(51) get the show on the road → *the show was got on the road
‘start a project’
→ ‘the project was started’
The grammaticality of the literal scene also constrains the following transformations listed by Schenk (1995, see Section 2.2.3):
(52) a. *John instructs the beans to be spilled.
b. *The beans try to be spilled.
Counter to Schenk, I claim that the unacceptability of these variants does not
have anything to do with the overall semantic structure of the idiom. Rather,
they must be rejected because the respective literal scenes are ungrammatical.
It is important to re-emphasise that topic indication (variation-principle 3)
does not lead to a fully grammatical literal meaning. Rather, the specific topic
indicators interfere with the construction of a conceptually consistent literal
scene. Note, however, that topic indication obeys the compatibility constraint.
The use of the topic indicators is fully grammatical with a view to how they
modify the phrase-induced figurative senses of the literal constituents. Thus,
while topic indication does not create a consistent literal scene, it is a fully
grammatical and systematic process nevertheless.
By integrating these variational constraints, our overview of different variational classes can be refined (see Table 6.2 below).
In the following chapter, the systematic lexicogrammatical variability of
different idioms will be analysed in order to establish a cognitive-linguistic
idiom-variation grammar. Systematic idiom variants, i.e. variants that do not
trigger strikingly unconventional semantic effects, reflect the strategies of constructional adaptation, (inconspicuous) literal-scene manipulation and topic
indication. By means of these strategies, idioms can be manipulated to give
expression to context-specific target-meanings in a creative but regular way.
As we have seen in this chapter, the creation of a systematic idiom-variant is
determined by the complex interplay of both motivating and constraining factors. Following the general hypothesis advocated in this study, the motivating
basis for systematic idiom-adaptations must consist in the conceptual correspondences that pattern the semantic structure of a given idiom and render
it transparent. Only for motivated idioms is it possible to considerably adapt
the lexicogrammatical structure in a way that is functional, grammatical and
fully compatible with the idiomatic meaning. In the following analysis this
hypothesis is tested.
JB[v.20020404] Prn:20/02/2006; 15:11
F: HCP1706.tex / p.50 (2704-2704)
 Idiomatic Creativity
Table 6.2 Overview of variation classes including variation principles and variational
constraints
Frozenness
The lexicogrammatical structure of an idiom cannot be varied.
Usual variation
a. systematic occasional variants effected by contextual
integration
topic indication
literal-scene manipulation (inconspicuous)
constrained by grammaticality, compatibility,
non-conspicuousness, non ambiguity, evocational autonomy,
compatibility, functionality and recognisability.
b. intentional creation of pun variants through exploitation and
wordplay including
literal-scene manipulation (conspicuous)
topic-related literal-scene manipulation
ambiguation/punning
constrained by grammaticality, functionality and
recognisability.
c. non-intentional erroneous variants
constrained by recognisability
Pseudo variation
Creation of a pseudo variant: the idiomatic meaning is not activated only the literal
interpretation is possible; the idiom is not recognisable in the context of use.
Cline of idiomatic variability
Occasional variation
Non-systematic variability
b. systematic variants effected by the preservation of regular
morphosyntactic and syntactic variability
constructional adaptations constrained by grammaticality,
compatibility, non-conspicuousness, non-ambiguity,
evocational autonomy, compatibility, functionality and
recognisability.
Systematic variability
a. usual variants (lexicalised alterations)
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.1 (47-128)
chapter 
The lexicogrammatical variation of idioms
denoting success, progress and failure
More studies of lexical and syntactic variation are needed, to
ascertain which FEIs repeatedly show up in real text as frozen
and which as fluid; and to classify and correlate the different kinds of variation. It may then be possible to separate
out formal classes, to determine which FEIs could indeed
be logged as ‘big words’, uninflectable, unvariable, and uninterruptable, and at the same time develop ‘rules’ governing
other more flexible kinds of item.
(Moon 1998: 309)
Having proposed a cognitive-linguistic account of idiom representation and
variation, we are now in a position to apply the theoretical apparatus to a
well-defined set of variation data. In this second analytical part of the book,
the idiom activation-set model is thus subjected to further empirical scrutiny.
The model is applied to actual idiom-variation data to shed cognitive-linguistic
light on the linguistic and psycholinguistic controversy carved out in Chapter 2.
The central empirical question to be scrutinised reads:
To what extent can the systematic variation of spf-idioms in discourse – or
constraints on their variability – be explained on the basis of their conceptual motivation by well-entrenched patterns of semantic extension, particularly
conceptual metaphors?
Since different idioms cannot be accounted for in terms of a uniform mental
activation-set and since idiom-variation cannot be described according to one
single variation strategy only, one cannot expect a uniform variation potential
for different idioms in different context-specific usage-events. Therefore, the
aim of this qualitative analysis of actual idiomatic usage-data is to reveal the
creative interplay of the motivated or unmotivated semantic structure with the
context-specific variation strategies.
For systematic idiom-variation, constructional adaptations, topic indication and weak forms of literal-scene manipulation are of particular interest.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.2 (128-171)
 Idiomatic Creativity
With reference to these variation principles, this chapter is an attempt to
develop a cognitive-linguistic idiom-variation grammar which explains the
lexicogrammatical behaviour of spf-idioms as a reflex of their systematic conceptual structuring. In other words, this chapter pulls the previous theoretical
arguments and empirical insights together to test them against empirical data.
This analysis of idiom-variation tokens combines the insights gained from
the idiom-representation model and its analytical application to spf-idioms
in Chapter 5. Moreover, it correlates these insights with the idiom-variation
model that was outlined in the previous chapter. In this way, it revisits the
idiom-variation controversy by means of the proposed cognitive-linguistic account of idiom representation and variation.
. Database and analytical procedure
The usage-based foundation of Cognitive Grammar suggests that genuine
usage-data rather than constructed examples should form the basis for linguistic analysis and theory construction. In accordance with this spirit, the recent
development of linguistics has seen the rise of corpus-linguistics, which bases
linguistic analyses on large computer-assisted corpora of both spoken and written language (cf. Kemmer & Barlow 2000: xv). By using computer-corpora
for the analysis of idiom variation, important insights have been gained by a
number of descriptive and classificatory studies (Moon 1998: Chapter 4; Coffey 2001). Following this inductivist spirit, I chose the 100-million-word BNC
World Edition as a database for the selection of actual tokens of idiom variation (for a good overview of the features of this corpus, see Aston & Burnard
1998). However, the present analysis differs from descriptive approaches; unlike these quantitative classifications, it tries to systematise and explain the
corpus-data qualitatively in relation to a coherent theoretical framework – the
cognitive-linguistic model of idiom-representation and idiom-variation.
Although corpora such as the BNC undoubtedly provide valuable insights
into the actual use of idiomatic constructions, they have their limitations, too.
Quantitative corpus analyses show that idioms occur relatively infrequently in
both spoken and written communication (Moon 1998; Minugh 1999: 57–71;
see also Minugh 2001). In accordance with the general frequency ranges revealed by these studies, idioms can be expected to appear in an approximate
range between 10 and 50 tokens per type in the 100-million-word BNC. Being
a subphenomenon of idiom use, idiom variants occur with an even lower frequency. Given the low overall frequencies and the striking discrepancies in the
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.3 (171-218)
Chapter 7. The lexicogrammatical variation of spf-idioms 
occurrence-probabilities between different idiomatic constructions, the BNC
cannot be used as a tool to derive stable quantitative predictions about idiom variability. For this reason, the present approach only concentrates on
a qualitative cognitive-linguistic comparison of the lexicogrammatical adaptations reflected by the different idiom-transparency types. Accordingly, the
corpus is not used as a basis for detailed statistical analysis, but as an extensive
resource of attested usage-data. In other words, the cognitive-linguistic analysis is expected to reveal very strong qualitative trends rather than statistically
measurable patterns.
To deduct the cognitive-linguistic idiom-variation grammar lexicogrammatical variants of spf-idioms will be discussed relative to their motivating
conceptual bases. However, this grammar must not be understood as a ruledriven apparatus. Rather, it is designed to make explanatory if-then statements:
if a given variation occurs, then salient conceptual structures and correspondences must be active; alternatively, if a given idiom can be motivated relative to striking conceptual regularities, then a given variation-potential can
be expected.
To develop the cognitive-linguistic idiom-variation grammar, the following dimensions of lexical and grammatical alteration will be considered:
a. Variation within idiom noun-phrases: article variation, number variation
(pluralisation/singularisation), adnominal modification, including premodification by adjectives, participles and nouns as well as postmodification by prepositional phrases and relative clauses.
b. Clause-level transformations: passivisation
c. General lexical flexibility through lexical substitution
I will discuss these phenomena because I expect them to reflect the influence
of the internal semantic structure of idioms on their malleability. I chose this
whole spectrum of alterations because I hope them to reveal a holistic picture
of the grammatical behaviour of idioms. In other words, I want to prevent
my analysis from producing claims about the linguistic status of idioms that
are discussed against one single adaptation process only. Thus, while the alterations in (a) and (b) are mainly expected to be determined by isomorphism,
(c) is hypothesised to illustrate lexical flexibility as induced by motivation. It
has to be explained in how far the underlying conceptual regularities reflected
by spf-idioms make these variations systematically possible or to what extent
the absence of internal semantic structuring inhibits variation. If the data substantiate my hypotheses, this speaks against the semantic-unity and long-word
conception of idioms. Moreover, if the lexicogrammatical alterations can be
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.4 (218-276)
 Idiomatic Creativity
motivated by the underlying conceptual associations, then this speaks for the
strong influence of general conceptual structuring on idiom representation and
processing.
To extract corresponding usage-tokens for representatives of the overall
set of 600 spf-idioms, the corpus was searched with the client programme
SARA-32 version 0.98, the customary search tool supplied with the BNC World
Edition. I conducted approximately 800 individual searches for verbal and
prepositional idioms with base-forms featuring the following constructional
schemas: V + NP (e.g., grasp the nettle), V + NP + PP (e.g., take the bull by
the horns), V + NP to-V (e.g., have a hard act to follow), P + NP (e.g., off
the hook, on the rails); between 0 and approximately 100 idiom-usage tokens
were retrieved for each of these. I concentrated on these constructional frames
to make my object of linguistic analysis fully compatible with the linguistic and psycholinguistic research tradition in idiom variation, which mainly
concentrates on verbal idioms. The starting points for my searches were the
citation-forms of all spf-idioms that I selected from the CCDI (see Chapter
5). To find tokens of a given idiom such as rock the boat, I used the Query
Builder, a general search mode customised by means of a visual interface. In
general, I specified the searches by asking the program to look for combined
instances of the lexical key-constituents, i.e. rock and boat, within a span of
up to 10 words. To include all morphological inflections of these head-words,
I used the Lancaster lemma-scheme. For each search all tokens were downloaded unless they exceeded the arbitrarily chosen limit of 150 tokens. In the
latter case, only a random selection of 100 tokens was downloaded. This search
strategy proved very effective for the extraction of all types of morphosyntactic and syntactic variation. However, since this search mode is dependent
on the direct stipulation of the conventional lexical key-elements, it is inappropriate for the retrieval of potential lexical substitutions such as rock the
ship/canoe/trawler. Lexical substitutions can only be approached by intuitive
adaptations of the base-form. Indeed, the search for occasional lexical substitutions is “a matter of serendipity” (Moon 1998: 51). To tackle this problem, I
tried to interpolate lexical substitutions by searching alternative collocates of
the key-constituents. While this method proved relatively easy for idioms with
infrequent constituents (e.g., upset the applecart) or more than two salient keyconstituents (e.g., take the bull by the horns), it was practically impossible to
apply to idioms with very frequent constituents (e.g., rock the boat); for these
idioms it proved far too time-consuming to read through all the individual occurrences of rock + N or V + boat. Consequently, lexical substitutions could
not be extracted systematically for them.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.5 (276-327)
Chapter 7. The lexicogrammatical variation of spf-idioms 
In the electronic appendix (www.idiomatic-creativity.ch), the extracted
corpus-data is presented and analysed for a representative set of 120 spfidioms. The data are categorised for the type of lexicogrammatical alteration
as well as the idiom-variation principles that they reflect. Moreover the data
are correlated with the conceptual backgrounds that motivate these idioms.
Thus, this appendix correlates the analysis of the conceptual backgrounds that
shape the semantic structure of spf-idioms (as sketched in Chapter 5) with the
context-specific lexicogrammatical alterations that they reflect. The appendix
thus comprises the detailed analyses on which the cognitive idiom-variation
grammar is built up.
In what follows, the major trends that can be deduced from these lexicogrammatical analyses are presented. The occurrence and specific quality
of idiomatic article-variation, pluralisation, passivisation, adnominal modification and lexical substitution as they appear with the different idiomtransparency types are discussed individually. The discussion is guided by the
questions:
a. What specific functions and qualities do these alterations adopt with the
different idiom-transparency classes?
b. To what extent do the variants still follow the general grammatical conventions of English as presented in Quirk et al.’s Comprehensive Grammar of
the English Language (1985)?
c. In how far and why are the alterations enabled or blocked? What is the
influence of underlying conceptual regularities on this behaviour?
By discussing the lexicogrammatical behaviour of idioms relative to the canonical grammatical patterns of English, it becomes possible to judge in how
far the presence of idiom-transparency preserves the systematic variability of
idiomatic constructions in discourse.
. Articles and article-variation in spf-idioms
Articles are closed-class grammatical words which realise the function of indicating the specific kind of reference that a given NP-head establishes to its
referent in the linguistic or situational context of discourse. In general, English
articles mark either definite reference (the cake, the cakes) or indefinite reference
(a cake, Ø cakes). In the case of definite reference, the referent is identified as
a unique and specific entity in the discursive context, whereas indefinite reference points to some virtual entity that has not yet been clearly identified (Quirk
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.6 (327-384)
 Idiomatic Creativity
et al. 1985: §5.10). Accordingly, Langacker claims that articles have the cognitive function of indicating the type of “mental contact” that the conceptualiser
establishes to the contextual instance of the conceptual entity-type that is designated by the NP-head. Cognitively, definite reference thus implies that the
speaker has established mental contact with the referent of the NP, i.e. he or she
has identified and isolated it as a unique and clearly delimited entity within the
target conceptualisation. The speaker further assumes that this referent can also
be identified by the hearer (Langacker 1991a: 98). With indefinite reference, the
addressed entity-type must still be grounded in the discursive context.
There are different ways in which the uniqueness of the referent can be determined or ‘recovered’ by the hearer when relying on the presupposed shared
knowledge between him or her and the speaker (for a detailed overview see
Quirk et al. 1985: § 5.28–35):
1. Situational reference: the uniqueness of the referent of the NP is pragmatically presupposed in the immediate situation in which the sentence is
uttered, e.g.: Have you signed the contract? (Said in a business situation).
2. Reference to encyclopedic uniqueness: on the basis of general world knowledge the unique identity of the referent can be established, e.g.: The US
President has taken strange decisions. (Encyclopedic knowledge tells us that
there is only one current US president).
3. Anaphoric and cataphoric reference: anaphoric and cataphoric reference
is established when the uniqueness of the referent is inferable from the
preceding or following discursive/linguistic context, e.g.: Jane was given a
diamond ring and the stone was very precious (anaphora), The bicycle over
there belongs to Lance Armstrong (cataphora).
4. Reference to institutions (Fellbaum 1993: 280): If the head-noun refers to a
uniquely identifiable social or cultural institution, the can be used, e.g.: The
press has been very aggressive recently.
In contrast, the indefinite articles (a and Ø) address a conceptual entity for
which no unique referent has been identified, i.e. no concrete mental contact
has been established, yet. The indefinite articles point to some unknown instance of a class of items that is still open to be materialised by a concrete
referent (see Quirk et. al. 1985: §5.36–37). The following subtypes can be distinguished:
1. Introduction of a new referent: the indefinite and the zero article are used
whenever the referent has not been mentioned before and is unfamiliar to
the speaker and hearer, e.g.: He has bought a book.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.7 (384-446)
Chapter 7. The lexicogrammatical variation of spf-idioms 
2. Non-referring use with a predicative nominal: in copular (predicative)
noun-phrases, the indefinite article has a descriptive rather than a referring role, e.g.: Chomsky is a great linguist (the article ascribes the quality
great linguist to the copular subject Chomsky), (cf. Quirk et al. 1985: 273).
3. Existential predication: existential sentences generally have an indefinite
noun phrase as a notional subject, e.g.: There’s a dog in the hut, He has
a book to read.
4. Use of the zero article with institutions, meals, means of transport, etc.: A
great number of everyday ‘institutions’ such as means of transport, meals,
times of day and night or illnesses do not have an article in English. This
use of the zero article is a matter of fixed collocations which have become
part of the English grammar, e.g.: go to town/church/prison, be in bed.
Interestingly, articles have only raised little interest in the analysis of idiom variation. Apart from Fellbaum (1993), there is no other study focussing exclusively
on the function and behaviour of these words in idioms. This striking lack of
interest for idiomatic article-use and article-variation conflicts with their high
potential for offering insight into the semantic structure and cognitive modelling function of idioms. Since articles mark the specific kind of reference that
a given NP-head establishes to its referent in the discursive context, the use and
variation of articles in idioms should clearly indicate the semantic status of
the idiom-internal NPs that they introduce. If all idioms were semantic units,
then their internal NPs would have to be non-denoting, i.e. they could not
be autonomously related to a given discursive referent; therefore, they could
not mark the referential definiteness or indefiniteness of the idiomatic NP. In
contrast, if the NPs can be attributed some degree of semantic autonomy and
motivation, then the use of the articles should correspond to the type of figurative reference established by the NP. In cognitive-linguistic terms, the use of
articles should therefore indicate to what extent the literal concepts, profiled by
the nominals, can be independently mapped onto a definite or indefinite target
to concretise it.
Article variation is a form of constructional adaptation (see Section 6.4.1).
Following our cognitive-linguistic account of this phenomenon, any changes
of the lexicogrammatical structure of an idiom must be mediated by the idiom’s literal scene as well as its relationship to the idiomatic meaning. As
we have seen, an idiom’s literal scene is pre-coded in terms of a conventionalised lexicogrammatical structure – its base-form. The article that appears
in a given idiomatic base-form, for instance the in grasp the nettle, thus belongs to the conventionalised symbolic profile of the literal scene. Since articles
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.8 (446-499)
 Idiomatic Creativity
specify whether the noun phrase’s discursive referent is definite or indefinite,
the pre-coded article should theoretically point to the conventionally presupposed type of reference of the idiomatic noun phrase. Thus, the in grasp the
nettle suggests that nettle typically has a uniquely identifiable referent when the
idiom is evoked in discourse. This, however, is where the intricacies of article use in idioms lie. Since the internal semantic structure of different idioms
is multifaceted, the motivating conditions for article use in different idioms
must be multifaceted, too. The following cognitive-linguistic analysis therefore involves two analytical foci. First, the functional quality of the articles
in the base-form is examined. Thus, it has to be shown if a) the article is
either definite or indefinite, and b) why the given quality appears in the baseform. Second, the variability of the articles can be scrutinised relative to the
idiom-transparency types.
.. The functions and behaviour of the definite article in idioms
A great number of idioms have the definite article in their base-forms. Following the grammatical function of the definite article in non-idiomatic constructions, these idioms can be said to presuppose the referent of their NPs
to be uniquely identifiable and specific. The functional status of this anticipation of definiteness is conditioned by the internal semantic organisation
of the complex scene conveyed by the idiomatic construction. The following
straightforward trends can be observed.
... Unique situational reference in relation to the presupposed
literal domain
This idiomatic use of the definite article corresponds to its standard use to
establish situational reference. Many idioms contain the definite article because their NP-concepts are embedded in a motivating conceptual sourcedomain relative to which the NP becomes uniquely identifiable. Since this
idiom-internal form of pragmatic presupposition is dependent on the apprehension of the literal source-domain, this idiomatic function of the definite
article is dependent on the motivation of the idioms themselves. Consider the
following examples:
(1) out/off the starting blocks
*The athlete got off some starting blocks
(2) dip a toe in the water
*The boy dipped a toe in some water
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.9 (499-531)
Chapter 7. The lexicogrammatical variation of spf-idioms 
These idioms presuppose the conceptual scenarios of their literal scenes.
Within these scenarios their NPs are unique. Thus, (1) presupposes the racedomain. In this context, the NP starting blocks has a unique referent and must
therefore be marked as being definite. Accordingly, the scenario of swimming (in a pool) motivates the use of the definite article in (2). The use of
the indefinite article, dip a toe in some water, would not make sense relative
to the swimming-frame, which motivates the idiomatic meaning according
to the conceptual metaphor purposeful activity/development is swimming. Due to the compatibility and recognisability constraints, the article must
therefore not be varied in these motivated constructions: by substituting the
indefinite article for the definite article, the literal meaning either becomes ungrammatical or it is altered in a way that does no longer correspond to the
conventional literal scene that is associated with the idiomatic meaning. Article variation in these constructions thus destroys the motivating conceptual
background relative to which the idiomatic meaning can be foregrounded. The
definite article thus works as a general marker for the inherent figurativity of
these idiomatic constructions (cf. Fellbaum 1993: 287).
A relatively small set of spf-idioms contain nouns whose uniqueness is
determined by our world-knowledge.
(3) reach for the sky/the moon
*reach for a sky/a moon
Sky and moon in (3) are unique entities and therefore demand the definite article. The substitution of the indefinite article for the definite one is inhibited
by the grammaticality constraint; the definite article must be used in order to
guarantee the well-formedness of the literal scene. This literal-level constraint
works independent of the idiom’s motivation and isomorphism. A subtype of
this constraint occurs with idioms that contain definite emblems.
Many emblems have the status of strong, uniquely identifiable cultural
institutions. Such an emblematic value can be attributed to road in (4).
(4) get the show on the road
*get the show on a road
The association of the road-concept with the abstract concepts of progress
and development is so deeply entrenched in our culture that it has gained
the quality of an emblem: road = development. (The cinematic genre of the
road-movie is a point in case). This emblematic reading of road can only be
activated if it is marked as uniquely identifiable through the use of the definite
article. The alteration of the article in this idiom destroys the integrity of the
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.10 (531-607)
 Idiomatic Creativity
literal scene. Indeed, the BNC does not reveal uses of this construction with a
varied determiner.
The widespread presence of the definite article in idioms is the effect of
their more general tendency to presuppose the definiteness of their NPs relative to the motivating literal source-domain. This speaks for the validity of the
theoretical assumption that idioms are activated as complex scenes in which
the literal scene works as a conceptual scaffolding, a standard of comparison
relative to which the target conceptualisation is understood. If the literal scene
is modified in a way that prevents it from performing this modelling function, the idiomatic meaning cannot be evoked, i.e. the target conceptualisation
cannot be accessed figuratively.
In the following idioms, the motivating source domain is also pragmatically presupposed. (5) activates the machine or vehicle-frame, whereas (6)
implies a canoeing-scenario:
(5) set the wheels in motion
He set some wheels in motion
(6) up the creek without a paddle
He was up a/some creek without a paddle
Contrary to (1)–(4), the use of the indefinite article does not destroy the cognitive micro-models here, although it distorts the postulated immediacy of the
implied action and situation, respectively. This, however, does not affect the
underlying motivating frame: the indefinite literal meanings still make sense
relative to the idiomatic meaning; they are compatible with it. Consider the
following uses attested by the BNC:
(7) Thank you, but it was Rob’s idea. I merely set a few wheels in motion.’
(HA7: 2544).
(8) But they’re still alive and spitting, and here then is the third album from a
band who have found themselves paddle-less up more creeks than they’d
care to count and have emerged [...] more glorious than anyone could
have hoped. (ED7: 2354).
The variation of the determiner in these constructions can be accounted for by
the latent isomorphic structure of these idioms. In (5) and (7), respectively, the
action of putting or setting a vehicle or machine in motion, makes the act of
starting some abstract configuration (a plan, idea, etc.) concrete. This ontologisation is guided and patterned by the conceptual metaphors: progress is forward movement in space and the developing configuration is a vehicle
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.11 (607-640)
Chapter 7. The lexicogrammatical variation of spf-idioms 
(or progress is the efficient progressing of a machine and the developing configuration is a machine), (see Sections 5.2.5 and 5.2.6). Following
these patterns, the constituent wheels can be attributed the phrase-induced figurative sense: (part of) developing configuration. Since the conceptual
node, wheels, constitutes only part of the machine or vehicle that is actually started, this correspondence involves the part for whole-metonymy:
wheels for vehicle/machine. Thus, the conceptual correspondence wheels
= developing configuration is motivated metaphtonymically. This, makes
an isomorphic analysis of the idiom’s meaning possible. Accordingly, the understatement in (7) is effected by triggering a subtle semantic distinction by
substituting the indefinite quantifier a few for the definite article. The definite
usage (the wheels) implies totality: by setting the wheels, i.e. all the wheels, of
a machine or vehicle in motion, the subject becomes fully responsible for its
processing. In contrast, a few implies indefinite partiality. Rather than starting
the whole machine on his or her own, the agent only turns on an undetermined portion of the total apparatus. On the level of the contextually-adapted
idiomatic meaning, this variation of the literal scene therefore reduces the influence of the subject. This triggers the modest understatement that can be
paraphrased as: I am only partially responsible for the successful activity.
A similar interpretation can be provided for the use of the quantifier more
in (8). The pluralisation and quantification of the constituent creek in this comparative construction can be explained on the basis of the constituent’s distinct
figurative contribution. The idiom’s semantic structure is motivated by the
progress as movement-model as well as a set of archetype metaphors: situations are locations, water is insecurity, problems are physical forces,
quality of path is quality of basis of development and instability is
negative development. These metaphorical mappings make it possible to
attribute the figurative sense difficult situation to creek in the context of the
idiom. This figurative sense can be quantified. Consequently (8) can be read
as: they have found themselves in more difficult situations than...
In a nutshell, these examples nicely illustrate the modelling function of
idioms. Through constructional adaptation, the article is slightly varied to manipulate the pre-configured conceptual content of the literal scenes creatively.
Since the conventional association of the literal scenes with the idiomatic
meaning can be motivated and structured on the basis of a set of conceptual
mappings, the manipulations do not destroy the literal background, but trigger
corresponding adaptations of the idiomatic meaning (understatement, quantification). These adaptations preserve the integrity of the cognitive micromodels because they can be motivated on the basis of underlying conceptual
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.12 (640-694)
 Idiomatic Creativity
metaphors and metonymies. In contrast, article variation was not possible for
(1) and (2); for these spf-idioms the presupposed conceptual background demands the presence of the definite article; article variation is incompatible with
their motivating bases.
If one compares, (1)–(8) to unmotivated idioms such as (9)–(11), one notices that the absence of a motivating domain influences our perception of the
definiteness of their NPs:
(9) buy the farm
(10) bite the dust
(11) shoot the breeze
Although the definite article in these opaque constructions suggests the NPs’
referential definiteness, it is unclear relative to which discursive referent, domain of knowledge or situational context this definiteness could be motivated.
Thus, our inability to evoke a structuring conceptual basis for these opaque
expressions also affects our ability to motivate the function of the articles. The
non-transparent status of the definite article in these constructions, freezes its
use within their opaque semantic structure (cf. Fellbaum 1993: 286). In other
words, the article is deprived of its conceptual scaffolding function in these
expressions; this corresponds to the opaque use of red in red herring (see Section 4.3). As a result, the variability of the article is blocked for these opaque
constructions. Indeed, I have not found any variation of the definite article in
tokens of these idioms. We can thus formulate a first trend for the variability of
idioms with the definite article in their base-forms:
The definite article points to the presupposed literal source-domain relative to
which the NP becomes uniquely identifiable. It thus works as a general indicator for the inherent figurativity of the construction. Article variation is blocked
for opaque idioms because the pattern of figuration cannot be remotivated
for these expression. With transparent idioms the definite article must not be
varied if this destroys the conceptual background on the basis of which the
association of the literal scene with the idiomatic meaning can be motivated.
Otherwise, the function of cognitive modelling could not be guaranteed. Thus,
motivation alone cannot be taken as the basis to trigger article variation in idioms. Motivation can only affect variation if it makes it possible to isolate the
nominal constituent semantically to attribute it some independent figurative
meaning within the context of the idiomatic expression. In a nutshell, the alteration of the article must be compatible with the motivating conceptual bases
as well as the relative semantic autonomy of the following NP.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.13 (694-748)
Chapter 7. The lexicogrammatical variation of spf-idioms 
These insights lead us to the analysis of determiner variation in isomorphic
idioms.
... Idiomatic anaphora and cataphora
Following the cognitive function of idioms to work as micro-models, NPs of
isomorphic idioms with definite-article base-forms are evoked to evaluate and
concretise a uniquely identifiable abstract referent. More specifically, they ontologise an abstract target-referent to which mental contact has already been
established in the immediate discursive context. This idiomatic use of the definite article corresponds to its anaphoric and cataphoric use in literal language.
Due to the inherent figurativity of idioms, anaphora and cataphora established
with idiom-NPs is based on their phrase-induced figurative senses rather than
their literal meaning. I will therefore call this indirect, idiomatic form of discursive co-reference idiomatic anaphora and cataphora. Consider the following
uses of the idiom jump on the bandwagon:
(12) Ever since Mr Gorbachev attended our G7 meeting last year and Mr Lamont went to inspect the Russian chaos, Britain [has effectively adopted
the Soviets and we will be pressing hardest to help]bandwagon – even George
Bush is now jumping on [this bandwagon]. (AL2: 215). (emphases and
bracketing by author)
(13) CHANNEL 4’s Saturday night series TV Heaven which [dishes up gems
from the archives]bandwagon has been so successful that the BBC have
jumped on [the bandwagon]. (C88: 2045). (emphases and bracketing
by author)
The constituent bandwagon has the figurative meaning successful (social)
activity / movement. This sense is motivated by the progress-ICM including the conceptual metaphor, a supporting means for progress is a
vehicle, the metonymy vehicle for pursued activity together with the
image of the bandwagon as a vehicle that transports a group of people that
are united in a common activity (see Section 5.2.5). To perform its idiomatic
modelling function, the constituent bandwagon must refer anaphorically to an
antecedent, i.e. a discursive referent that is uniquely identifiable in the usagecontext. Put differently, the figurative sense of bandwagon must be related to
some preceding target-conceptualisation that it can categorise. In (12), the
complex VP has effectively adopted the Soviets ... to help constitutes the target
to which the figurative sense of bandwagon refers anaphorically; the VP is the
antecedent, while bandwagon is its figurative anaphor. In accordance with this
pattern of anaphoric reference, the demonstrative determiner this is substituted
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.14 (748-805)
 Idiomatic Creativity
for the to index the direct accessibility of the idea that is to be concretised.
The same interpretation can be provided for (13), in which the VP dishes up
gems from the archives constitutes the specific successful activity that works as
an antecedent for bandwagon and motivates its use with the definite article.
Idiomatic cataphora with the same expression occurs in (14):
(14) When great ex-players join the media [bandwagon] of [knocking successful
teams]bandwagon one can’t help feeling there may be a tinge of jealousy in
their comments. (CHV:110) (emphases and bracketing by author).
Here, the of-complement knocking successful teams specifies the abstract activity referred to and concretised by bandwagon. The complement can be interpreted in terms of apposition (see Section 7.5 below). This makes it possible
to read bandwagon and knocking successful teams co-referentially: bandwagon =
(vehicle for) shared activity = knocking successful teams. The cataphoric
use of the definite article is further motivated by the premodifier media which
renders the activity more specific and uniquely identifiable in the discursive
context. (14) can thus be paraphrased as: When great ex-players join in the media’s (activity of ) knocking of successful teams. Note that unlike this paraphrase,
the idiomatic use triggers a number of weak implicatures that can be directly
derived from its conceptualising force. Thus, jumping on the bandwagon entails the effortless gain of advantages at the expense of others, the participation
in an activity that one does not control or steer oneself, the mindless following
of others. In other words, the conceptual impact of the idiomatic micro-model
makes it possible to explain, evaluate and criticise the target conceptualisation
in far greater detail.
It is important to emphasise that the definite article is only used if bandwagon has an identifiable definite referent in the immediate discursive context.
If this is not the case, the indefinite article, a, is used as in (15):
(15) One of the reasons being put forward is that they are jumping on a bandwagon which unfortunately is worldwide. (HCU:73).
(16) The public silence was broken early in 1987 by Marxism Today, who can
seldom resist a bandwagon, even if they often fall under the wheels in the
process; and [the particular bandwagon] they joined on this occasion was
the one about what the Tories had labelled [‘Loony Left’ councils]bandwagon .
(C9S:648).
The systematic grammatical use of the article with bandwagon becomes particularly obvious in (16). In this passage, the activity referred to by the use
of bandwagon remains non-specific at first and is therefore used with the in-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.15 (805-862)
Chapter 7. The lexicogrammatical variation of spf-idioms 
definite article. At the second occurrence, the definite article is evoked in cataphoric use (the bandwagon → ‘Loony Left’ councils). The referential uniqueness of the nominal’s referent is further highlighted by the premodifier (the
particular bandwagon) and the NP’s topicalisation through fronting (the particular bandwagon they joined).
Of course, jump on the bandwagon is a highly transparent idiomatic construction (an isomorphic idiom with constituental motivation). Therefore, it
is not surprising that the lexicalised figurative sense of bandwagon is actualised
in accordance with the general rules for article use. In other words, it could be
argued that the systematic variation of the determiner is not a result of the idiom’s isomorphic semantic structure, but that it rather reflects the lexicalisation
of bandwagon. Extensive corpus-data reflecting the usage of other isomorphic
constructions can be put forward to counter this argument.
For instance, the idiom upset the applecart also belongs to the group of
isomorphic idioms with the definite article in the base-form (its underlying
conceptual motivation has been discussed extensively in Section 4.7.1 and is
further sketched in the electronic appendix). The BNC-data indicate that its determiner is consistently used according to the trend outlined above. In (17) the
idiom is evoked in a general statement about some troublesome people spoiling satisfactory constellations. The generality of the statement is reflected by
the indefinite plural use of applecart, which is accompanied by the zero article.
(17) These people are fortunately few and far between because they are extremely odd and have a way of upsetting applecarts. (G0X: 1786).
(18) All this has upset [the applecart] of [the relation of fertility to prosperity]applecart . (EDK: 1433).
(19) A foreigner who is stunningly attractive will readily upset [the apple-cart]
of [our preconceptions]applecart , as will a friend whom we discover has been
slandering us behind our back. (CM8: 582).
(20) Both of these sides will have to bring out something special if they are
to come as close as Jed to upsetting [the championship]applecart [applecart].
(K5A: 704).
In (18) and (19) the precise semantic value of the constituent is made specific by postmodifying of-complements. In (20) the premodifier championship
fulfils the same function. These modifiers are appositional; they describe the
constellations that are metaphorically modelled and evaluated by applecart. In
other words, these examples become fully comprehensible if the modifiers are
substituted for applecart. Accordingly, the definite article appears in cataphoric
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.16 (862-918)
 Idiomatic Creativity
use. These data nicely illustrate that the use of the definite article with applecart
must be justified. If the discursive context does not imply or specify the necessary information to isolate its figurative target previous to the evocation of
the idiom, this referent must be made accessible by other means. In (18)–(20)
referential uniqueness is guaranteed by restricting the NP’s referential scope
through premodification and postmodification.
On balance, the behaviour of the definite article in isomorphic idioms conforms to the general rules for non-idiomatic standard constructions. One can
thus propose the following trend:
Isomorphic idioms with definite-article base-forms are prototypically used in a
discursive context where the NP’s context-specific figurative referent is known
to the speaker and hearer. If the discursive context does not provide information about the referent to which the NP could refer either anaphorically
or cataphorically, the figurative NP-head must be rendered unique by premodification or postmodification. Otherwise, the indefinite article (or another
indefinite determiner) is used. Note that this trend clearly speaks against the
uniform treatment of idioms as semantic units. Moreover, it points to the fact
that the choice of the article in idioms is not guided by autonomous syntactic
rules, but it is determined by the conceptual arrangements that idioms convey.
.. The functions and behaviour of the indefinite and zero article
in idioms
In standard constructions, the indefinite articles (a and Ø) are used when the
NP points to a new referent that has not become uniquely identifiable yet. As
we have noticed with the definite article, the presence of motivation is not a
sufficient condition to create the potential to vary the article in a given idiomatic construction. Rather, the NP must be construed as having some degree
of semantic independence within the overall motivated structure. The same
conditions apply to motivated idioms with indefinite-NP base-forms:
(21) throw a spanner in the works
(21) profiles a machine-scenario. More specifically, it concretises the deliberate
prevention of an activity in terms of stopping the processing of the machine.
While its meaning may be paraphrased as a semantic unit (‘stop, thwart, prevent sth.’), the idiom’s literal scene can be construed isomorphically relative
to the idiomatic meaning: create (throw) a problem (a spanner) for some
developing activity (in the works). This isomorphic construal is motivated
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.17 (918-962)
Chapter 7. The lexicogrammatical variation of spf-idioms 
by the machine-model and the conceptual metaphors: development is the
efficient processing of a machine and problems are obstructions. Following these metaphors, spanner concretises a given interference, while the
works stand for the developing activity in terms of the metaphtonymic
mappings: part for whole → works for machine and the progressing
activity is a machine. One can therefore call this idiom a motivated idiom
with a latent isomorphic structure or a latently isomorphic idiom. The idiom’s
latent isomorphism accounts for the variability of the article in actual use:
(22) As the season draws to a thrilling climax, with Swansea and Llanelli both
bearing down on a possible league and cup double, Pontypridd and Newport get their chance to throw a spanner in the works at the Bank Holiday back-to-back Schweppes Cup semi-finals at the Cardiff Arms Park.
(CKA: 894).
(23) Borrowings will fall immediately as the company releases advanced corporation tax reserves, and two-thirds should be paid by 1992. But [the spanner] in these works could be [falling advertising revenue]spanner . (A55: 108)
(emphases and bracketing by author).
In (22) – presumably a commentary on a rugby championship – the idiom’s
semantic structure is activated as a unit and can be paraphrased as Pontypridd
and Newport get their chance to stop/prevent this at the Bank Holiday. Thus, the
discursive context does not provide any specific clue about how these teams
want to stop Swansea and Newport. Therefore, the semantic contribution of
spanner remains unanalysed in the given instance of use. In contrast, (23)
involves an isomorphic interpretation of the idiom, which attributes the constituent spanner the autonomous reading interference/problem. Following
this isomorphic reading, spanner makes cataphoric reference to falling advertising revenue, which works as the discursive referent that is concretised and
evaluated by spanner. As this example indicates, the use of the article in idioms
with the indefinite (or zero) article in their base-forms seems to correspond to
the previously outlined trends for the definite article. This is substantiated in
the following sections.
... Isomorphic idioms with new, non-definite figurative referents
In the previous subsection, we have observed that the use of the definite article
in isomorphic idioms must be justified by specifying the NP’s discursive referent if it has not been pre-established in the discursive context. For isomorphic
idioms with an indefinite-NP base-form this is not necessary; the NPs in these
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.18 (962-1017)
 Idiomatic Creativity
idioms typically concretise a new abstract referent that has not been addressed
or specified in the previous discursive context:
(24) walk a tightrope
(25) skate on thin ice
The NPs in these constructions have a phrase-induced figurative value that
is evoked to introduce a new, abstract concept. In (24) tightrope has the
figurative sense unstable basis (for development/progress). This meaning is motivated by the conceptual metaphor system: progress is forward
movement in space, basis/context of development is a path, quality
of basis/context is quality of path, positive/negative development is
stability/instability → tightrope = insecure basis for successful development. A fully compatible motivation underlies the figurative value of
thin ice in (25). Due to the semantic autonomy of the nouns relative to this
system of conceptual metaphors, the evocation of the article is in full accordance with its use in standard constructions: when the figurative referent in
the usage-context remains indefinite, the indefinite (or zero) article is used;
otherwise, the definite article appears (see (26)–(29) below). Note that the appositive postmodification in (29) guarantees the referential uniqueness of the
discursive referent of ice. The of-complement speculation and ... specifies the
abstract referent that is idiomatically captured by this constituent:
(26) This weekend Mr Gorbachev has walked the finest diplomatic tightrope.
(A3T: 259).
(28) To many Churchill was not so much a buccaneer as a straightforward pirate, a political outcast who skated on thin ice deliberately to keep himself
in the public eye, [...]. (HWA:407).
(29) The book is perhaps especially valuable for the way in which the author is
prepared to extend onto [the thin ice] of [speculation and comparison with
other animals]thin ice , a trait sadly lacking in so many agricultural scientists.
(B7M: 1571).
The systematicity with which the article is used in these examples works as a
straightforward proof that the respective NPs have an autonomous semantic
value. Also note how the conceptual source-domain scenario is intertwined
with the target conceptualisation to model it in figurative terms. The variation
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.19 (1017-1081)
Chapter 7. The lexicogrammatical variation of spf-idioms 
data, therefore, are not compatible with an overly narrow, orthodox view of
idioms. In contrast, they illustrate the conceptualising power of idioms and the
corresponding adaptation of the lexicogrammatical scaffolding. In a nutshell,
the data substantiate the above mentioned trends for article use in idioms:
Motivated and isomorphic idioms with an indefinite-NP base-form are prototypically used in a discursive context, where the NP’s metaphorical referent
is not yet known to the speaker and hearer. The presence of the indefinite NP
speaks for the high degree of semantic autonomy of this constituent. Due to
its semantic autonomy, the NP can be actualised either definitely or indefinitely depending on whether its discursive referent is uniquely identifiable or
not. Consequently, the article is highly variable in these idioms and follows the
general rules for the use of determiners in English.
In my corpus I have found a great number of verbal idioms that follow this
trend: swallow a bitter pill, like a rat leaving a sinking ship, hitch one’s wagon to
a star, bang your head against a brick wall, be on/walk a knife-edge, a drowning
man will clutch at a straw, put one’s head in a noose, (be) in a rut, strain at a
gnat and swallow a camel, stir up a hornet’s nest, between a rock and a hard
place and many others. In contrast, semantic opacity constrains the potential
for attributing idiom-constituents an autonomous figurative meaning. As with
definite-NP idioms, article variation therefore cannot be expected for opaque
idioms with an indefinite NP in their base-form. For instance, I have not found
any article variation for take a bath, come a cropper or go places in the BNC. This
substantiates the position that the systematic variation of articles in idioms
follows their underlying semantic motivation as effected by conceptual models.
Still, as the following subtypes of idioms with the indefinite and zero article
show, isomorphism does not necessarily guarantee article variability.
... The use of the indefinite article in existential idiomatic constructions
A number of idioms can be described by analogy with existential sentences.
Thus, (30)–(32) suggest that the existence of the indefinite referents of the their
NPs is a given. Apart from this existential reading, these idiomatic have-a-NPto-V constructions have the modal sense of obligation, i.e. have a thing to do.
(30) have a hard row to hoe
(31) have a mountain to climb
(32) have a bitter pill to swallow
Relative to different metaphorical models all of these idioms can be attributed
an isomorphic semantic structure: the NPs denote some problem or difficulty
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.20 (1081-1120)
 Idiomatic Creativity
that has to be solved or accepted. Thus, hard row and mountain in (30) and (31)
concretise obstacles and obstructions following the conceptual metaphor
progress is forward movement in space. And bitter pill in (32) metaphorically denotes an unpleasant fact relative to the consumption is dealing with
an issue-model (see Section 5.2.4). But although their underlying conceptual
motivation makes it possible to isolate the semantic contribution of the NPs,
the indefinite determiner is hardly varied in discourse. Here, the variability
of the indefinite article is not limited by semantic opacity, but it is prevented
by the overall constructional constraint of the existential construction, which
conventionally takes an indefinite NP as its notional subject. In other words,
in these constructions article variation is blocked by the grammaticality constraint. The same is true when the constituents are used as predicative nominals
in copular constructions:
(33) This is a mountain to climb
(34) This is a bitter pill to swallow
... The zero article in asyntactic idioms
A considerable number of idiosyncratically constructed (asyntactic) idioms
contain singular count nouns in combination with the zero article and thus violate the grammatical distribution of the zero article in standard constructions.
(35) jump/abandon ship
(36) back to square one
(37) tread water
Although constructionally-idiosyncratic, the semantic structure of these idioms is motivated. (35) refers to the metaphor-model development is a
voyage; it expresses the idea of deserting a failing organisation by the image
of leaving a ship. purposeful activity is a game constitutes the conceptual
mapping underlying (36) with square one being related to board games. Finally,
the association of treading water with stagnation in (37) is fully sanctioned
by the metaphors progress is forward movement in space and water is instability/insecurity. The reason for the absence of the definite or indefinite
article in these expressions is their terminological character: abandon ship is
a specific term in seafaring, square one is a board-game term and tread water evokes the specific health cure-domain. When an article is used with
these idioms, the terminological reading, which relates the constructions to
these motivating source-domains, is blocked. As a consequence, the idiomatic
meaning cannot be activated. In other words, the frozen use of the zero article
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.21 (1120-1172)
Chapter 7. The lexicogrammatical variation of spf-idioms 
in (35)–(37) is primarily restricted by the recognisability constraint. The constraint prevents the article in these idioms to be varied although some of them
have a conceptually-motivated isomorphic semantic structure.
In conclusion, the variation-data on systematic article use in spf-idioms
support the general hypothesis of this study: the variation of the articles is mediated or blocked by the presence or absence of underlying conceptual bases
that motivate the association between the literal meaning and the idiomatic
meaning. The conceptual patterns make it possible to motivate idioms and to
attribute phrase-induced figurative meanings to the constituents of isomorphic idioms. However, conceptual motivation alone does not guarantee article
variability. Rather, the integrity of an idiom as a cognitive micro-model must
be preserved. Thus, isomorphism can be overridden by the grammaticality,
compatibility and recognisability constraints, which block the flexibility of the
determiners. These insights provide strong evidence for the idiom-activationset model and the complex interaction and integration of context, literal scene
and idiomatic meaning in idiom processing.
. Number and number variation in idioms
The grammatical category of number points to the quantity of the entity referred to by the nominal head. The English number system involves a binary contrast between the singular and the plural. The singular denotes ‘one,’
whereas the plural indicates ‘more than one.’ Morphologically, the regular plural is marked by adding the plural suffix, -s, to the noun stem. The singular
is unmarked (for a cognitive-linguistic account see Langacker 1991a: Section
2.3.1). With regard to their sensitivity to the number contrast, three classes of
nouns can be distinguished (for a comprehensive overview, see Quirk et al.
1985: § 5.3):
1. Variable nouns: variable nouns or count nouns can adopt both the singular
and the plural, e.g.: The dog is sleeping vs. The dogs are sleeping.
2. Singular invariable nouns: this class only allows the singular. It is most typically represented by noncount nouns, which usually denote concrete but
undifferentiated masses, substances or continua (milk, wood, water, etc.): I
love milk vs. *I love milks.
3. Plural invariable nouns: this class involves summation plurals (scissors,
ashes, jeans, etc.), pluralia tantum (arms ‘weapons,’ brains ‘intellect,’ clothes,
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.22 (1172-1232)
 Idiomatic Creativity
etc.), unmarked plural nouns (people, police, etc.), personal adjective heads
(the poor, etc.) and some proper nouns (the Seychelles).
For the number of an idiomatic NP-head to vary systematically, two necessary
conditions must be met. First, the NP-head must have some degree of semantic autonomy; this makes it possible to adapt the number of this constituent
to the number of its referent in the target conceptualisation. Second, the literal NP-head and its figurative correspondent must both profile countable
conceptual entities.
The two necessary conditions for number variation can only be fulfilled
by VP-idioms and PP-idioms that have an isomorphic semantic structure.
The following idioms (which become motivated and analysable through the
metaphors basis of development is a path and problems are obstacles)
meet this requirement and feature fully grammatical, systematic number variants for their NPs in the BNC. The idiomatic NPs in (38)–(40) vary their
number relative to the plurality of their referents in the specific usage-contexts:
(38) bounce back when things go wrong and find new pathways, blaze new
trails. (B2F: 633).
(39) ‘I’ve been looking into something like this for about two years and come
up against brick walls constantly,’ she said. (K54: 6493).
(40) What remains a mystery is how Virgin would have fared had its directors
decided to weather the storms in the stock market. (A1S: 315).
Thus, idiomatic isomorphism can be taken as a valid measure to explain number variation. In contrast, number variation is blocked for opaque idioms such
as kick the bucket, bite the dust or bite the bullet, because no functional relationship between their NP-heads and the target conceptualisation can be
established.
For a considerable number of isomorphic idioms number-variation is
grammatically constrained on the literal-level because their NP-heads are uncountable mass nouns or plural invariable nouns:
(41) skate on thin ice
(42) gain/lose ground
(43) dip a toe in the water
(44) in the doldrums
In the BNC, I have found only one token featuring number variation for one
of these expressions:
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.23 (1232-1283)
Chapter 7. The lexicogrammatical variation of spf-idioms 
(45) Nelson is flagging appropriate titles in its catalogue as ‘self-access’, there
are one or two new publishing names dipping toes in the swirling waters of
EFL, and during the close season there have been some author-transfers,
including David Vale from CUP to Heinemann and Alan Maley to Penguin. (F9J: 712).
This token is a stylistically marked, conspicuous literal-scene manipulation of
(43). The extended NP the swirling waters of EFL mimics poetic language to
create a pathetic elaboration of the corresponding literal scene. This elaboration leads to the intensification of the image provided by the idiom. Note
that the novel, intensified literal scene conforms to the underlying motivating conceptual metaphor water is uncertainty and progress is forward
movement/swimming. However, this playful elaboration triggers stronger implicatures than the base-form: EFL is a very uncomfortable field to publish in;
EFL will drag you (and your career) down. Moreover, it links the idiom to the
stylistic context of writing and publishing.
For the use of the definite article in idiom base-forms, we have seen that
a number of these expressions presuppose motivating scenarios in relation to
which only the definite article is functional. The same is true for the number of
the NP-heads in many motivated idioms:
(46) on the rails
(46) is motivated by the railway-scenario, which works as one specific sourcedomain within the model: purposeful activity is a journey. Within the
railway context, rails has the quality of a summation plural, i.e. of two equal
parts which are joined together (see Quirk et al. 1985: §5.76). Although rail is a
count noun and could therefore appear in the singular, the singular is typically
associated with other senses: a rail is a ‘metal bar to work as a barrier or support,’ whereas rails has the specialised meaning ‘railtrack.’ Thus, the motivating
railway-scenario can only be evoked on the basis of the plural. Hence, the use
of the singular would destroy the conceptual background (the source-domain)
for the figurative association of the literal scene and the idiomatic meaning and
thus trigger a non-idiomatic interpretation. In short, the use of the singular is
incompatible with the idiomatic meaning and therefore blocked. This claim is
supported by corpus-data:
(47) She picks up this blue skirt and holds it against her legs and then puts it
back on the rail. (A74: 580).
(48) Keegan got Newcastle back on the rails at Birmingham, following a run of
three defeats. (CEP: 2537).
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.24 (1283-1337)
 Idiomatic Creativity
The singular of rail in (47) evokes the subsense ‘bar to hang clothes on,’ while
only (48) reflects the metaphorical evocation of this source domain with the
idiom. Thus, singularisation is prevented by the compatibility, recognisability
and functionality constraints.
The preservation of number in accordance with the idiom’s motivation
is also prominent with somatic idioms. In the following constructions the
concrete image associated with the literal scenes prevents the number of the
body-part nouns from being varied:
(49) stand on your own two feet
(50) start/get off on the wrong foot
The image in (49) is motivated by the conceptual metaphor stability (of success) is physical balance. The image of physical stability would be destroyed
for this idiom, if the body-part constituent feet were singularised (stand on
your own foot). In contrast, (50) conveys the scene of a person taking an uncoordinated step to conceptualise the unsuccessful start of a given activity. The
underlying metonymic association body-part for movement of body-part
(foot for step) would not make sense if the body part were used in the plural.
In sum, motivation prevents number variation in idioms if the motivating conceptual background is destroyed by this alteration. Number variation
in these idioms is non-functional, incompatible with the idiomatic meaning
and prevents the recognition of the idiomatic construction. Number variation is also blocked for idiosyncratically-constructed idioms be on course, back
at square one, jump/abandon ship (as discussed with the zero article above).
For these constructions, the use of the plural or singular would destroy their
specific idiomatic reading that is conveyed due to their terminological nature.
Finally, a small number of isomorphic idioms have plural nouns in their
base-forms that are metaphorically mapped onto singular or mass-noun figurative correspondents:
(51) make great strides
(51) is motivated by the metaphor progress is forward movement in space.
The idiom conceptualises the amount of progress made by an agent in terms
of the distance covered by a mover. Thus, the length of a mover’s strides (great
strides) figuratively expresses the amount of progress. The use of the indefinite plural is well-motivated relative to the idiomatic meaning ‘make much
progress.’ The abstract concept of progress has the quality of an abstract mass
or indefinite quantity of perfection. To suit the non-discrete character of this
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.25 (1337-1388)
Chapter 7. The lexicogrammatical variation of spf-idioms 
abstract concept, the concretising literal-scene must also refer to a non-discrete
mass. This is achieved by the indefinite plural great strides, which captures
an indefinite amount – a mass – of stride-instances. To make the amount of
progress countable, the literal scene can be manipulated as in (52):
(52) A great stride forward was made after 1931 by J. N. L. Myres who began the
systematic investigation of early Anglo-Saxon pottery (Myres 1969: 1–5),
[...]. (CFK:163).
The singular (a great stride) points to a countable ‘piece’ of progress (the systematic investigation of Anglo-Saxon pottery). Since the passage addresses a
particular developmental step rather than progress in general, the singular,
a great stride, must be chosen instead of the mass-construal provided by the
plural (great strides). In short, this manipulation of the literal scene is fully
functional and compatible with the idiomatic meaning. Again, it shows how
the motivated literal scene of an idiom interacts with the usage-context and
triggers corresponding lexicogrammatical alterations.
On balance, number variation reinforces the trends found for the use of
articles. Number variation occurs systematically with isomorphic idioms that
contain a number-variable noun that can be attributed an autonomous figurative reading. Due to the grammaticality constraint, number variation is
however prevented for isomorphic idioms that contain mass-nouns, plural
invariable nouns or referentially-unique constituents. For a great number of
motivated idioms number variation is non-functional and incompatible with
the idiomatic meaning because it would prevent the evocation of the sourcedomain on the basis of which their motivation can be established. Finally,
number variation does not occur systematically for opaque idioms due to the
absence of a transparent semantic basis to which this grammatical adaptation
could be applied.
. Passivisation – idiom variation at the clause level
The passive is one side of the coin of the grammatical category of voice. To
be open to passivisation an active clause must minimally fulfil one requirement: its main verb must be transitive. The grammatical distinction between
active and passive makes it possible to construe the content of a transitive
action from two alternative pespectives without changing the content of the
reported scene: John kicked a ball vs. The ball was kicked by John. The voice
category thus establishes two grammatical alternatives to code the dynamic in-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.26 (1388-1449)
 Idiomatic Creativity
teraction between an agent (John) and a patient (ball) in accordance with the
process that is profiled by the verb (kick), (for a comprehensive overview see
Quirk et al. 1985: §3.64–72). The problem of why certain idioms can take the
passive, while others cannot has been discussed controversially in the phraseological and syntactic literature. The following cognitive-linguistic account of
idiom passivisation and its inhibition proceeds from Langacker’s account of
the passive construal.
According to Langacker (1991a: 229): “The two voices [...] represent alternative philosophies with respect to the linguistic coding of events.” The structure of the passive can be explained straightforwardly as a marked construction
that departs from canonical and unmarked transitive clauses in the active voice
(see Section 3.4.4). The passive typically changes the levels of prominence in a
force-dynamic relation between an agent and patient. While an active clause
highlights the most energetic participant (the agent) as its syntactic figure
(→ John kicked the ball), a passive clause elevates the patient, i.e. the tail in the
energetic interaction, to this status (→ The ball was kicked). In other words,
the primary function of the passive is to defocus the agent, while foregrounding the patient and the process that it undergoes (see also Stein 1979; Croft
1998: 90).
As article variation and pluralisation, idiom passivisation is a subtle and
fairly inconspicuous form of constructional adaptation. The potential of verbal idioms for passivisation was intensively discussed in the context of the
analysis of transformational deficiencies (see Section 2.1.3). Valuable explanatory models were proposed by Chafe (1968) or Newmeyer (1974) for instance.
Further, the phraseologists Burger (1973), Fleischer (1982) and Dobrovol’skij
(1997) point to a number of supporting and constraining factors. Their sophisticated frameworks take both the quality of the literal meaning and the
idiomatic meaning into account. With reference to the grammaticality and the
compatibility constraints, their lists of supporting and constraining factors can
be systematised as follows (Table 7.1). This list of supporting and constraining factors indicates that the minimum requirement for the passivisation of
a verbal idiom is the presence of a transitive relationship in both the literal
and the idiomatic meaning: both literally and figuratively the idiomatic configuration must code an inherently dynamic interaction between an agent that
functions as an energy source and a patient that functions as the affected element in this energetic interaction. In accordance with this condition, verbal
idioms belonging to groups (a) and (c) in the table are uncontroversial. While
verbal idioms of group (c) cannot take the passive due to the grammaticality
constraint, passivisation occurs most naturally with verbal idioms that have a
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.27 (1449-1543)
Chapter 7. The lexicogrammatical variation of spf-idioms
Constraining factors
Supporting factors
Table 7.1 Influencing factors for idiom passivisation
Grammaticality of literal scene
Compatibility with idiomatic meaning
(a)
Open syntactic slot for direct object in
idiom base-form:
(d)
Idiomatic meaning corresponds to a
transitive process:
drive NP to distraction
→ he was driven to distraction
set the wheels in motion = start sth.
→ the wheels were set in motion
(b)
Base-form verb is open to passivisation
(literal scene describes a concrete transitive
action involving agent + patient):
(e)
Idiom has an isomorphic semantic
structure (constituents can be attributed
an autonomous meaning):
set the wheels in motion
→ the wheels were set in motion
burst/prick + the bubble =
tackle + the problem
→ the bubble was pricked
(c)
Base-form verb is not open to
passivisation (intransitive, reflexive,
stative, etc.):
(f)
The idiomatic meaning corresponds to
an intransitive process or a state, and/or
the idiom is not isomorphic:
shoot yourself in the foot
→ *Himself was shot in the foot
kick the bucket = die
→ *The bucket was kicked
transitive base-form with a schematic patient-slot: blind someone with science, send someone to the showers, etc. (= group (a) in the table). Structurally,
these idioms can be described as complex transitive verbs with an unspecified direct object (someone): send to the showers = expel someone, blind with
science = deceive someone. Thus, verbal idioms that belong to this group are
characterised by their fully grammatical transitive valence; they reflect what
Burger (1973: 81) calls “echt transitive Struktur” (true transitivity). In other
words, they are naturally isomorphic with regard to the ascription of agent and
patient roles. Therefore, these idioms can be straightforwardly passivised independent of whether they can be motivated or not: he was blinded with science,
he was sent to the showers.
The most interesting class for an analysis of idiom passivisation thus consists of transitive verbal idioms with a lexically specified patient-slot: grasp the
nettle, spill the beans, kick the bucket, bite the dust and so forth. For these idioms
semantic analysability seems to be a necessary requirement for passivisation.
If, for instance, we analyse the institutionalised base-form of prick the bubble

JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.28 (1543-1591)
 Idiomatic Creativity
in terms of its immanent grammatical and semantic relations, the following
isomorphic structure appears:
{sbj}
{sbj}
{agent}
stop
V
process
an activity
obj
patient
{agent}
{sbj}
{sbj}
process
V
prick
patient
obj
the bubble
meaning paraphrase
grammatical relations
semantic roles
idiomatic scene
literal scene
semantic roles
base-form
The institutionalised base-form of transitive verbal idioms thus typically codes
a literal scene with a thematic relationship that consists of a process (prick)
and a patient (bubble) that undergoes this process. The agent within the profiled action chain is not lexically-filled in the base-form [sbj pick the bubble].
That is, it has the status of a highly schematic subject-slot that can be specified variably depending on its specific referent in the discursive context. (In
the figure the schematic status of the agent/subject slot is indicated by curved
brackets). The distribution of the process and patient slots on the level
of the idiomatic meaning is analogous to the literal meaning. That is, the
figurative correspondents, stop and activity, reflect the same force-dynamics.
This can be readily explained with reference to the motivating conceptual
metaphors: integrity of a constellation is physical stability, failure
is physical disintegration and problems are forces. These ontological
metaphors map the concrete force-dynamic structure that is immanent in the
source-domain (pricking a bubble) onto the target domain (stop an activity). As a result, the empirically abstract process stop an activity/idea,
etc. can be understood in terms of the same force-dynamics (see Lakoff 1993b;
Tucker 2001).
In contrast, the semantic structure of the opaque idiom bite the dust does
not have a coherent transitive structure: although its literal meaning is transitive, this meaning cannot be mapped onto a corresponding force-dynamic interaction on the level of the idiomatic meaning. Rather, the idiomatic meaning
‘die’ describes an intransitive change of state:
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.29 (1591-1653)
Chapter 7. The lexicogrammatical variation of spf-idioms 
[sbj]
[sbj]
[experiencer]
die
V
process
meaning paraphrase
semantic roles
idiomatic scene
↑
[agent]
[sbj]
[sbj]
process
V
bite
patient
obj
the dust
literal scene
semantic roles
base-form
Given their analysable structure, isomorphic idioms such as prick the bubble
can be readily passivised because the patient can be attributed an autonomous
semantic value. This makes it possible to foreground this constituent as a passive subject and to defocus the agent, while highlighting the resultative process
that the patient undergoes:
the activity
passive sbj
patient
was stopped
passive-V
process
patient
passive sbj
the bubble
process
passive-V
was pricked
meaning paraphrase
semantic roles
idiomatic scene
literal scene
semantic roles
passivised form
The ability of an idiom to take the passive is evidence for its semantic
analysability (cf. Burger 1973: 84; for a critical discussion of this problem see
Dobrovol’skij 2004). This, of course, speaks against the unqualified view of
idioms as semantic units. For completely non-isomorphic idioms the passive
transformation, clearly, is not possible. Thus, for bite the dust the passive is
incompatible with the non-transitive nature of its idiomatic meaning and the
corresponding absence of a potential passive subject:
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.30 (1653-1701)
 Idiomatic Creativity
*was died
passive-V
process
meaning paraphrase
semantic roles
idiomatic scene
↑
patient
passive sbj
*the dust
process
passive-V
was bitten
literal scene
semantic roles
passivised-form
The comparison of prick the bubble with bite the dust thus suggests that idiom
passivisation is blocked in the absence of a motivated isomorphic structure. By
and large, this hypothesis is confirmed by the relatively few instances of idiom
passivisation found for spf-idioms in the BNC. Consider the following instance
of prick the bubble:
(53) In the House of Commons on the afternoon of Thursday, 3 December,
he got a sizeable cheer when he spoke against any ‘irrevocable step’. Baldwin rightly thought that this particular bubble would be pricked over the
weekend. (EFN: 2025).
In this variant the idiomatic NP is highlighted and emphasised (this particular
bubble), while the implicit agent is defocused. Together with the topicalisation
of the figurative NP, the passive construction is thus clearly motivated by the
conceptual analysability of the construction. Note that this is also reflected by
the determiner-variation in accordance with the process of idiomatic anaphora
(the → this). The same systematicity is reflected by grasp the nettle:
(54) The experiences of recent years have challenged this presumption, but the
nettle has still not really been grasped. (G31: 305).
In (54) the passive is used in full accordance with the isomorphic semantic
structure of the construction. It can be readily paraphrased as:
(54) a.
The experiences of recent years have challenged this presumption, but
the problem has still not really been tackled.
The function of the passive to foreground the patient and defocus the agent
becomes most obvious in (55). This example is analoguous the use of prick the
bubble in (53):
(55) Moreover, deep-rooted public opposition to urban road construction
emerged in the 1970s, so that the only feasible way forward would have
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.31 (1701-1756)
Chapter 7. The lexicogrammatical variation of spf-idioms 
been [a policy of car restraint]. But [this nettle]policy was never grasped; [...]
(C8F: 189).
Again, the idiomatic NP nettle becomes the passive subject, while the agent
is not mentioned. The foregrounding of the patient is made explicit by the
demonstrative determiner: this nettle. The demonstrative indicates that the
context-specific metaphorical target of nettle has already been mentioned in
the preceding linguistic context. More specifically, the NP a policy of car restraint works as the abstract referent that is concretised by it. The foregrounded
idiomatic NP makes figurative anaphoric reference to this antecedent. Clearly
then, the idiomatic NP in this instance of passivisation can be semantically
isolated in accordance with its phrase-induced figurative meaning (problem)
as motivated by the underlying conceptual metaphors. It makes full sense to
use nettle as a relatively independent semantic unit and to foreground it as a
topic by making it the subject of the passive clause. I strongly doubt that it is
possible to handle these data by an orthodox or generativist account of idiom
representation and variation.
Passives can also be found for metaphtonymically motivated idioms with a
latent isomorphic structure.
(56) ‘Look, Mark,’ Muldoon said forlornly ‘What’s the point? The wheels have
already been set in motion. (AC2: 1481).
In (56) the metaphtonymy developing configuration is a vehicle, wheels
for vehicle, moving wheels for starting development provides the motivating conceptual basis for the passive. As such, the idiom put the wheels in
motion is not isomorphic, but rather has the status of a complex verb with
the meaning ‘start something.’ Wheels can, however, be directly related to
this idiomatic patient-slot via the wheels for developing configurationmetonymy. Accordingly, this constituent can be foregrounded as a passive
subject in (56) to mean The process has already been started.
In sum, the few instances of passivisation found for spf-idioms in the BNC
provide further support for previous insights on the influence of transparency
(particularly isomorphism) on the systematic constructional adaptability of
these expressions. In other words, the idiomatic potential for constructional
adaptations can be readily explained as an effect of the conceptual organisation
of idioms and the creative evocation of their conceptualising force in discourse.
Idiom passivisation is not based on autonomous syntactic processing. This
said, the restricted data for idiom passivisation found for spf-idioms in the
BNC (for more instances see electronic appendix), are far from providing a sta-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.32 (1756-1801)
 Idiomatic Creativity
ble basis from which any clear-cut rules on idiom passivisation can be derived.
Thus, for a more profound cognitive-linguistic analysis of idiom-passivisation
a more extensive database is still needed.
. Adnominal modification
The complexity of noun phrases can be increased (virtually indefinitely) by
two forms of adnominal modification: premodification and postmodification.
Premodifiers and postmodifiers add information to the NP-head, which either restricts its referential scope or adds descriptive detail to a specific referent
(for an extensive overview see Quirk et. al. 1985: Chapter 17). In cognitivelinguistic terms adnominal modification reflects the more elaborated construal
of a conceptual entity along the dimension of schematisation and elaboration.
Elaboration consists in the cognitive ability to construct a mental representation with refined informational precision; this involves the integration of
conceptual units to constitute a more fine-grained conceptualisation. On the
linguistic level, conceptual elaboration triggers the creation of corresponding composite structures such as head-modifier constructions (see Langacker
1987: 235–236).
Adnominal modification in idioms affects their syntactic fixedness. Frozen
idioms should not allow any conceptual elaboration of their pre-coded literal scenes by inserting premodifiers or postmodifiers. In positive terms, it can
therefore be hypothesised that adnominal modification is only possible if the
speaker is able to establish a systematic, directly meaningful and functional
relationship between the elaborated literal scene and the target conceptualisation. Moreover, this manipulation of the literal scene must be fully compatible
with the idiomatic meaning.
Apart from the variable semantic structure of different idioms, the description of idiomatic adnominal-modification is complicated by the alternative
idiom variation principles outlined in Section 6.4. Reconsider the following
adjectival modifications:
(57) The Chancellor had a narrow tightrope to walk and he managed to please
a variety of people. (CEL: 589).
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.33 (1801-1854)
Chapter 7. The lexicogrammatical variation of spf-idioms 
(59) This weekend Mr Gorbachev has walked the finest diplomatic tightrope.
(A3T: 259).
While (57) is a clear example of literal-scene manipulation, (58) reflects topic
indication. Since these modifications have already been discussed in Section
6.4.2 above, it is not necessary to go into further detail here. (59), however,
shows that literal-scene manipulation and topic indication may appear in
combination. Diplomatic functions as a topic indicator to narrow down the
context-specific interpretation of the figurative sense of tightrope. This results
in the context-specific reading: a difficult diplomatic situation. The superlative finest, however, qualifies the literal meaning of tightrope to intensify the
usage-specific idiomatic reading. The resulting NP finest tightrope has the elaborated figurative meaning: the most delicate situation. Thus, (59) can be paraphrased as: This weekend Mr Gorbatchev has acted in the most delicate diplomatic
situation.
These examples illustrate that adnominal modification within idioms is
a semantically heterogeneous phenomenon. Thus, when trying to answer the
question to what extent different idioms can be modified adnominally, it is
necessary to pay close attention to the precise semantic contribution of the
modifier. In what follows, I will discuss those types of premodification and
postmodification that proved to be most prominent in my corpus of spfidiom-variation.
.. Idiom-internal and idiom-external relativisation
Postmodification by finite and non-finite clauses is a relatively scarce variational phenomenon for the idioms in my corpus. Tokens can be found for
postmodification by adnominal relative clauses (Quirk et al. 1985: §17.9), -edparticiple clauses (Quirk et al. 1985: §17.29) and infinitive clauses (Quirk et al.
1985: §17.30). These postmodifiers only occur with clearly isomorphic idioms
such as jump on the bandwagon, blaze a trail, walk a tightrope, grasp the nettle, swallow the bitter pill, have a mountain to climb, skate on thin ice, burst the
bubble and so forth.
With some isomorphic verbal idioms such as (60) and (61), the transitive [v-np] structure can be transformed to become a [np + postmodifying
clause]-construction that works as the complex embedded NP of a higher
order clause (cf. Moon 1998: 5.6.4):
(60) The Khmer Rouge is aware of the tightrope it has to tread, both with the
West and with Vietnam. (A45: 159).
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.34 (1854-1912)
 Idiomatic Creativity
(61) This was immediately prior to the Nicholson ‘star’ era; the wave of which
he rode the crest was beginning to build [...]. (AP0: 656).
In these examples, the idiomatic NPs (tightrope, wave) become the heads of the
[np + postmodifying clause] constructions, while the postmodifying relative clauses (it has to tread, of which he rode the crest) contain the idioms’ verbs
as their VP-heads. Since the verbal core of the postmodifying relative clause
constitutes the V-node of the idiom’s base-form, I call this variation process
idiom-internal relativisation.
Idiom-internal relativisation is a constructional adaptation that changes
the relative prominence of the profiled conceptual substructures. More specifically, it places the idiomatic NPs in focal position: the NP is fronted and
topicalised (see Nunberg et. al. 1994: 501). These relative-clause-construals of
verbal idioms are similar to idiomatic pseudo-clefts such as (62):
(62) And that’s fairly thin ice on which to skate one’s credibility! (C9J:29).
Idiom-internal relativisation is only compatible with and functional for
analysable idiomatic expressions. This condition is fulfilled for (60)–(62). The
relative clause constructions underlying these variations are evoked to place
emphasis on the phrase-induced figurative senses of the constituents tightrope,
ice and wave. Thus, these variants can be directly paraphrased by substituting
the figurative senses ‘go through’ + ‘difficult situation’ and ‘experience’ + ‘a
positive trend’ for the constituent nouns and verbs. For non-analysable idioms
(both motivated and opaque ones) the construal of the literal scene according to this schema is not compatible with the idiomatic meaning. Due to this
constraint, they cannot be manipulated accordingly:
(63) *the dust that they had bitten was ...
(64) *the toe which she dipped into the water was ...
(65) *the bucket that he kicked was ...
With a view to Schenk’s argument that syntactic transformations of idioms
reflect the application of autonomous syntactic rules on lexicalised phrases (see
Section 2.2.3), these data provide very strong counter-evidence. The variants
reveal the semantic analysability of the phrases and further indicate that the
syntactic adaptations are the effect of their conceptual motivation.
A couple of variational tokens of isomorphic idioms reflect postmodification by restrictive and non-restrictive relative clauses that are external to their
constituents structure.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.35 (1912-1972)
Chapter 7. The lexicogrammatical variation of spf-idioms 
(66) Thus, while the health, agricultural and voluntary work committees remained social rather than political in their programmes, they blazed a trail
that others widened into a national highway. (APD: 1318).
(67) One of the reasons being put forward is that they are jumping on a bandwagon which unfortunately is worldwide. (HCU:73).
In (66) and (67), the relative clauses modify the idiom-internal NPs (a trail and
a bandwagon) restrictively to add more specific information about their figurative referents. Since these relative clauses add information that is external to
the idiom’s constituent structure, I will call these postmodifiers idiom-external
relativisations. The content of the idiom-external relative clause can both extend the literal scene, as in (66), or it can work as a topic-related relative clause
as in (67).
The relative-clause in (66) is determined by the conceptual metaphors
progress is forward movement in space and the context/basis of development is a path. In accordance with these metaphors, the elaborated NP a
trail that others widened into a national highway can be mapped onto the level
of the idiomatic meaning to create the actual reading: they created a basis for
future development that others expanded so that, nationwide, other committees
could proceed in the same way. This variant thus nicely illustrates the principle of literal-scene manipulation. The postmodifer enriches the literal scene to
widen the pool of conceptual entailments for the concretisation and evaluation
of the communicative target. To understand the variant, the hearer must reinterpret the postmodifying relativiser in terms of a figurative analogy to derive
its context-specific interpretation: trail : nationwide highway = small basis
for (political) development : strong basis for nationwide political
activity. The evaluative and critical weak-implicatures of this analogy are obvious: blazing a trail is an individual and effortful pioneering activity, while
widening this trail into a national highway means to earn the public reward
for the original efforts of others. Since the idiom is motivated by conceptual
metaphors, and since the variant exploits the source-domain knowledge of
these projections, the creative production process that underlies the creation of
(66) must be shaped by the metaphorical models. Psycholinguistically speaking, these metaphorical concepts must be strongly activated for the productive
manipulation of the idiomatic configuration to become meaningful.
Unlike (66), (67) does not recruit source-domain knowledge, but the topicrelated relativisation (which unfortunately is worldwide) applies directly to the
phrase-induced figurative sense of bandwagon (successful movement). To comprehend this variant, the metaphorical source-domain knowledge does not
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.36 (1972-2028)
 Idiomatic Creativity
play a supporting role. On the contrary, the interpretation of the postmodifier in combination with the literal meaning of jump on the bandwagon creates
a highly inconsistent literal scene. Therefore, the conceptual metaphors cannot
be assumed to play a decisive role in the creation and comprehension of this
specific idiom-variant. To produce (67) the speaker must be able to process
bandwagon according its the phrase-induced figurative sense. Although this
sense can be motivated by conceptual metaphors and metonymy (progress
is forward movement, support for progress is a vehicle, vehicle for
supported activity), the corresponding source-domain knowledge is not of
direct relevance for the production and comprehension of the variant.
Looking back at the cognitive-linguistic idiom-production model in Section 6.2, we can thus propose the following refinement. Psycholinguistically,
the difference between the literal-scene manipulation of (66) and the topicindication of (67) is significant. While both variants reflect systematic adaptions of the corresponding idiom activation-sets, these adaptations cannot be
processed along the same linguistic and conceptual pathways. While (66) suggests an active involvement of the underlying conceptual associations, the conceptual impact of the literal-scene must be downplayed for (67). One can thus
postulate both an active as well as a passive influence of conceptual metaphors
on idiom processing. In the case of literal-scene extension, the source-domain
knowledge must be retrieved and mapped onto the idiomatic meaning to derive the context-specific idiomatic reading. Thus, when the literal scene of an
idiom is extended in accordance with its underlying conceptual motivation,
conceptual metaphors must be assumed to play a decisive and active role in
the production and comprehension process. In contrast, conceptual metaphors
cannot and must not be attributed an active role in a topic-related modification
such as (67). Rather, the activation of the respective source-domain knowledge would interfere with the derivation of the actual idiomatic reading of the
varied construction. Nevertheless, conceptual metaphors must be attributed a
passive role for the processing of this variant. Namely, they support the process
of attributing a phrase-induced figurative sense to the constituent bandwagon;
the conceptual metaphors provide the conceptual scaffolding relative to which
this sense can be constructed. However, once this sense has been established
in the mental lexicon of the speaker/hearer, it is not necessary for him or her
to recur to the motivating conceptual background in every instance of idiom
processing. Rather, the phrase-induced figurative sense may be activated like a
word-meaning that can be modified irrespective of its metaphorical basis.
This qualified view of the role of conceptual metaphors in idiom-processing
is in accordance with our previous observations on the intricacies of idiom rep-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.37 (2028-2066)
Chapter 7. The lexicogrammatical variation of spf-idioms 
resentation and processing. Being heterogeneous, idioms cannot be regarded
as behaving in a homogeneous way. Rather, it must be assumed that alternative pathways through the mental networks are (unconsciously) selected
when specific idiomatic expressions are activated for linguistic processing in
specific contexts. In other words, the variation-data support the psycholinguistic claim that, in the process of idiom production and comprehension, the
activation-potentials of both the conceptual and the linguistic substructures of
an idiomatic activation-set vary considerably and change dynamically depending on how the semantic levels of a given idiom interact with the context of use.
These alternative activation-patterns can be expected to correlate with different
idiom-variation strategies. At the moment, it is impossible to judge or predict
precisely how activation-potentials spread through different idiom-activation
sets in different usage-contexts. However, for future psycholinguistic research,
variation-data like (66) and (67) implicate that closer attention should be paid
to the impact of alternative idiom-variation principles on idiom processing.
After this short psycholinguistic deviation, let us now turn back to our analysis
of idiomatic postmodifcation by idiom-external relative clauses.
In some isolated instances, the idiomatic NP of an isomorphic verbal idiom
is repeated in isolation from the collocational context of the phrase. Moreover,
this isolated NP is postmodified by an idiom-external relative clause to render
its reference more specific:
(68) The ‘one-last-pushers’ of the Labour movement need to be reminded
of the electoral mountain that remains to be climbed (a mountain which,
thanks to the Boundary Commission, will be some 20 or so Tory seats higher
by the time of the next election). (CAG: 110).
(69) The final salvo was that of jumping on the bandwagon, a wagon we were
informed which was already jam-packed with Black and Jewish women:
women whose political history among white gentile English women, in
some ways, paralleled our own. (CF4: 698).
In combination with their postmodifying relative clauses, the isolated repetition of the idiomatic NPs (a mountain, a wagon) places these constituents
in focus. This is a further clear indicator of the autonomous semantic contribution (the figurative senses) of these constituents. Again, the specifying
relative-clause can be literal-scene-related (69) or both literal-scene and topicrelated (68).
Note that in (69) the conceptualisation of bandwagon as a vehicle is exploited in accordance with the conceptual metaphor support for development is a vehicle. Thus, unlike in (67) above, the source-domain knowl-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.38 (2066-2111)
 Idiomatic Creativity
edge related to the literal scene, jump on the bandwagon, is activated. This
speaks for the adequacy of the view outlined above: the motivating conceptual metaphors can adopt both an active role, as in (69), and a passive role,
as in (67), when one and the same idiom is processed according to different
idiom-variation principles.
Modifying the NP a mountain of have a mountain to climb, (68) is a particularly interesting idiom-variant because it reflects the process of conceptual
blending (cf. Mena Martínez 2002). More specifically, the postmodifying relative clause integrates elements belonging to the discursive context of election with elements from the literal source-domain, mountain. Following
the metaphor problems are obstructions, the variant conveys a blended
conceptualisation of the target-domain task: winning the majority in an
election against the tories is conceived as an electoral mountain that
must be climbed by the Labour party. The postmodifying relative clause further elaborates this blend by conveying the image of a mountain whose height
is measured in terms of Tory seats. The variant thus nicely reflects the complex interaction between the three activated levels of meaning in its processing:
the literal scene (climb a mountain), the phrase-induced figurative sense
(overcome a problem) and the actual target conceptualisation (win against
the tories). The variant thus provides further support for a simultaneous
or interactive processing model of idiom comprehension and production (see
Sections 2.2.5 and 6.2.2). The data clearly speak against a lexical representation
and lexical processing view of idiom storage and use. In contrast, they underline their cognitive functionality as complex scenes. In other words, such blends
most explicitly reflect the speaker’s cognitive creativity in the process of adapting the idiomatic standard configuration to the communicative necessities of
the usage-event.
.. Prepositional postmodification in combination with the
development is a journey-metaphor
Many idioms that characterise progress as physical movement in space contain nouns like road, track, lane, trail, etc. These nouns profile path-concepts
that are evoked to concretise the abstract context, direction or basis of a given
development (see Section 5.2.5). In accordance with this metaphorical conceptualisation, the nouns can be postmodified by prepositional [to/toward +
NP]-phrases to indicate the aim of the development:
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.39 (2111-2161)
Chapter 7. The lexicogrammatical variation of spf-idioms 
(70) There were quite a few battles before Pearce got British Aerospace on the
road to privatisation, [...]. (A6L: 1777).
(71) As it once showed the way toward democratic success, today it blazes the
trail toward democratic failure’. (J57: 614).
These systematic variations can also be explained in terms of conceptual blending. The conceptual metaphors progress is a journey and aim of development is the destination establish conceptual correspondences between
the literal path-concepts and the developmental aims addressed by the writers (privatisation, democratic failure). These topic-indicating aims are
mapped into the literal-scene to fill the destination-nodes in the source frame
(journey). On the linguistic surface, the metaphorical target-destinations occupy the NP-slot in the prepositional construction [to/toward + NP]. The
postmodifications in (70) and (71) thus evoke hybrid images that integrate elements of both the literal-scenes and the context-specific topics. These hybrid
images provide further evidence for the complex discursive interplay of the semantic levels of an idiomatic activation-set with its contextual target. All of
these semantic representations must be activated in the combined implementation of topic-indication and literal-scene manipulation. But, the necessary
precondition for these variants to become meaningful is the motivated association between the conceptual source and target domains as patterned by the
conceptual metaphors.
.. Idiomatic postmodification by of-complements
Of-constructions are the most frequent postmodifying PPs to be found in idiom variants by far. Interestingly, the of-complements in the variation-data
only contain target-related NPs and thus perform the primary function of topic
indication. Semantically, these postmodifiers work as appositive or partitive
constructions primarily (for a definition see Quirk et al. 1985: §5.116, §17.47).
With a great number of variants, the postmodifying of-complement is
apposed to the phrase-induced figurative-sense of the idiomatic NP that
it modifies:
(72) The book is perhaps especially valuable for the way in which the author is prepared to extend onto the thin ice of speculation and comparison
with other animals, a trait sadly lacking in so many agricultural scientists.
(B7M: 1571).
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.40 (2161-2228)
 Idiomatic Creativity
(73) All this has upset the applecart of the relation of fertility to prosperity. (EDK:
1433).
The topic-indicating, appositive of-constructions are co-referential with the
postmodified NPs (thin ice, applecart); the semantic relationship between the
idiomatic NPs and the of-complements is one of identity. These appositional
patterns can thus be paraphrased in terms of a copular clause as:
(72) a.
speculation and comparison with other animals is the thin ice; thin
ice = an insecure basis for argumentation
(73) a.
the relation of fertility to prosperity is the applecart; applecart = labile
constellation
Clearly then, this appositive form of postmodification by topic-indicating
of-complements is dependent on the analysable nature of the idiomatic
constructions to which they apply. The postmodifiers are used to make the
context-specific referents of the postmodified NPs explicit. In other words, the
of-constructions are inserted to ‘translate’ the autonomous figurative meaning
of these constituents into the usage-context. Accordingly, the variants can be
directly understood, if the topic indicators are taken to replace the idiomatic
constituents that they postmodify:
(72) b. The book is perhaps especially valuable for the way in which the author is prepared to extend onto speculation and comparison with other
animals, a trait sadly lacking in so many agricultural scientists.
(73) b. All this has upset the relation of fertility to prosperity.
This substitution test proves the idioms’ isomorphic character (see also Sabban 1998: 157–159). Appositive postmodification by of-constructions can only
work for idioms that reflect this transparency-type. Indeed, the BNC does not
contain this form of idiom-variation with non-isomorphic idioms both of the
motivated and opaque types. The metaphorical conceptualisation of a weak argument in terms of an unstable path (thin ice) in (72) is readily guided by the
conceptual metaphors stable development is physical balance, stability
is security and basis/context of development is a path. Feeding on the
phyically-motivated experiential entailsments of these models, the idiomatic
NP performs the function of ontologising and evaluating the target conceptualisation in very concrete terms – the brittleness of speculative thinking becomes
conceptually and emotionally ‘graspable.’
On the basis of these insights, it now becomes possible to further substantiate our claims on article-variation in idioms. Above, I have suggested
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.41 (2228-2278)
Chapter 7. The lexicogrammatical variation of spf-idioms 
that the use of the definite article in isomorphic idioms with definite-article
base-forms must often be justified by rendering the figurative NP-head unique
later in the text. This is done by premodification or postmodification in order
to trigger a cataphoric use of the definite article. Appositive of-complements
are one particular means to trigger idiomatic cataphora. Since they unveil the
actual referent of the idiomatic NP explicitly, the specified topics directly motivate the definiteness of the idiomatic NP. This trend is very systematic. It
thus substantiates the view of an idiom-variation grammar as a form of figurative competence that depends on the cogniser’s knowledge of the conceptual
analysability of idiomatic expressions. An autonomous grammar cannot model
such intelligent human cognitive processing.
Figurative competence based on conceptual structuring also underlies the
following types of of-complementation. In (74) the use of the of-construction
can be labelled partitive:
(74) If you’ve been thinking of making domestic changes or improvements, but
lacked the money or momentum to set the wheels of change in motion use
today to make a start. (HJ3: 5838).
The presence of the partitive postmodifier in this variant can be readily
explained in terms of metaphtonymy and blending. As we have seen, the
metaphor the developing configuration is a vehicle motivates set the
wheels in motion. Moreover, the idiom is shaped by the part for wholemetonymy: the wheels are the parts responsible for the car’s movement; starting the wheels’ movement causes the car to run. The partitive of-construction
in (74) makes use of this internal metonymic motivation. The postmodifier
change introduces the whole of which the wheels constitute the part. In other
words, the topic-indicating concept change is blended into the literal scene
to occupy the vehicle-slot in the source-domain. The variant thus evokes a
hybrid, blended image in which the abstract topic-concept change is conceptualised as a vehicle. Following the metonymic mapping, start wheels
for start vehicle, triggering change corresponds to setting the wheels of
change into motion. This variant thus nicely reflects the complex interplay of
metaphor, metonymy and blending with the context-specific manipulation of
an idiomatic-activation set. The creative variant elaborates on the idiom’s instrinsic creativity to evoke an appropriate contextual adaptation of the complex
symbolic configuration.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.42 (2278-2328)
 Idiomatic Creativity
.. Premodification by adjectives and nouns
Adjectival and nominal premodification is an intricate phenomenon influenced by a variety of fine-grained functional and semantic distinctions (see
Halliday 1985: 164; Warren 1984; Quirk et al. 1985: §7.40–7.44; Ferris 1992).
The same is true for the occurrence of adjectives in idiom variants (Ernst 1981;
Nicolas 1995; Allerton 2002). Due to the limitations of this book, it is impossible to do full justice to this complexity. Here, I would mainly like to refer
back to Nicolas’ (1995) claim that premodification in idioms can only be syntactic, i.e. that the premodifiers have the semantic status of external adjectives
or nouns with an adverbial interpretation (see Section 2.2.3). In contrast, I
would like to show that semantic modification is fully systematic and grammatical process that can be explained as a reflex of the motivated and analysable
semantic structure of idioms.
... Literal-scene extension through adjectival
and nominal premodification
In the following variants the premodifiers have the status of adjectives of quality that function as descriptors of the head and extend the idiom’s constituent
structure as a form of literal-scene manipulation:
(75) Brittle Power is very much a continuation of the Lovins canon: the concerted challenge to most of the percepts apparently underlying current
official thinking about energy – this time overturning the most ponderous
applecart of all. (B74: 1576).
(76) They know as well as anyone else that Labour will have an even higher
mountain to climb come the next general election as a result of Boundary
Commission changes (K5D: 10019).
(77) But where to go? We debated hopelessly, going round in ever decreasing
circles: the Alps too crowded, the Himalaya too frightening (ECG: 1088).
(78) In quick succession, Sandy Lyle, Faldo twice and then, last year, Ian Woosnam came riding in on the crest of a British tidal wave. (AJJ: 816).
In (75)–(77), the adjectives of quality work as internal, semantic modifiers.
Contrary to Nicolas’ claim, none of these literal-scene manipulations triggers wordplay because they do not create ambiguity or conspicuous comic or
stylistic effects. Rather, the manipulated literal-scenes fulfil the function of providing more fine-grained conceptual micro-models against which the idioms’
context-specific meanings can be assessed. Interestingly, all of these literal-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.43 (2328-2368)
Chapter 7. The lexicogrammatical variation of spf-idioms 
scene extending adjectives must be interpreted as intensifiers on the level of the
idiomatic meaning. Thus, most ponderous applecart in (75) intensifies the motivating image of instability, whereas even higher mountain in (76) strengthens
the idea of a great problem that must be overcome or borne; and in (77), ever
decreasing stresses the image of senseless, stagnating movement to elaborate
the idea that the subjects’ argument is ‘moving on the spot.’ The same is true
for the noun premodification in (78). The inserted element tidal (wave) must
be interpreted figuratively to work as an intensifier (very) that applies directly
to the autonomous figurative meaning of this noun (wave = favourable situation/period). It can thus be paraphrased as: ...Ian Woosnam came profiting from
the very strong British period.
Literal-scene manipulation with the effect of intensifying the idiomatic
meaning can also be found with some idioms that already contain a premodifiying adjective of quality in their base-forms. To intensify the idiomatic
meaning of these constructions, the adjective is used in the comparative or
superlative form or modified by an intensifier or degree adverb:
(79) Add to this some very crude graphics (they were poor two-and-a-half
years ago) and sparse sound FX, and you’re skating on very thin ice.
(EB6: 1181).
(80) Do you want,’ Anna said, embarking impulsively on the thinnest ice, ‘do
you want to reconsider everything? (CMJ: 1281).
In (79) and (80), the motivated concretisation of an unstable, insecure context
of activity is intensified by strengthening the image underlying the idiom’s NP
thin ice. Both the degree adverb very and the superlative form thinnest apply to
the phrase-induced figurative reading of the NP and thus exploit the motivated
and isomorphic nature of the expressions. The general trend for literal-scene
extending nouns, quality adjectives and superlatives to work as intensifiers on
the level of the idiomatic meaning can only be explained as an effect of the motivated transfer of source-domain entailments onto the target-domain. Both
for the production and the comprehension of such variants the underlying
patterns of semantic extension must therefore be assumed to play an active
role. Thus, being able to mobilise the conceptual metaphors and metonymies
as guidelines for these lexicogrammatical adaptation processes, it becomes possible for a speaker to elaborate the content of the idioms’ literal scenes such
that the intensified idiomatic-meaning can be construed. Since, on the linguistic surface, intensification can be triggered by a variety of slightly different
NP-premodification and adjective-adaptation processes, the cognitive idiom-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.44 (2368-2425)
 Idiomatic Creativity
variation grammar is determined by conceptual modulations rather than pure
syntactic knowledge. Thus, the systematic data under (75)–(80) speak against
Nicolas’ syntactic view of adjectival premodification in idioms.
An interesting subtype of literal-scene manipulation that also accords with
this interpretation can be found with metonymically-motivated kinetic and
somatic idioms. Thus in (81) and (82), the inserted adjectives qualify the kinetic concept stride. However, this specified information does not merely
have the function of providing some refined description of how the persons
move. Rather, the characteristic movement has to be interpreted as a symptom
that reveals a more general characteristic or ability of the subjects.
(81) And though, once she hit her formidable histrionic stride, it was possible to feel that a little of Bette Davis went a long way, her excesses were
surely preferable to the homogenised blandness of those mass-produced
Hollywood sirens of whom a lot went a very little way indeed. (A3V: 38).
(82) Best and Fairest: Fijian hat-trick hero Mesake Rasari getting into his immense stride against Japan in their Pool A clash (CKA: 377).
Thus, she hit her formidable histrionic stride implies: she is an excellent actress,
while Mesake Rasari’s immense stride points to his outstanding physical qualities as a Rugby player. Again, these variants nicely illustrate how the underlying
metonymic motivation of kinetic and somatic idioms can be exploited by the
syntactically simple means of adjectival premodification to communicate relatively complex information in a concise format. In other words, this form of
literal-scene extension exploits the part for whole or possessor for possessed metonymies, which are implied in kinetic and somatic idioms (see
Sections 4.5.2 and 5.3).
... Adjectival and nominal premodification as topic indication
The most frequent type of premodifier within idiom variants are topicindicating adjectives and nouns such as political, financial, economic, intellectual, legal, tax or energy. These premodifiers function as classifiers. For many
of them both external and internal interpretations are equally possible (cf. also
Sabban 1998: 156). If they function as external, adverbial adjectives they classify
the idiomatic meaning as a whole; if, however, they are interpreted as internal
modifiers, they only define a subclass of the phrase-induced figurative sense
of the idiomatic head-noun. Counter to Nicolas’s argument that premodifying
adjectives are predominantly used as external modifiers, I have encountered
topic-indicating premodifiers most frequently with analysable idioms. Thus,
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.45 (2425-2481)
Chapter 7. The lexicogrammatical variation of spf-idioms 
the use of topic indicators as externally-modifying classifiers seems to be the
exception rather than the rule. Indeed, with many isomorphic idioms the internal interpretation seems to be the only grammatical one. Most strikingly, an
internal interpretation is highly favourable for those topic-indicating adjectives
that modify the NP in combination with another premodifier or postmodifier
which further ‘isolates’ the NP syntagmatically:
(83) Within a year Labour was recapturing some of the political ground which
it had lost at the local elections in November 1931. (ACH: 808).
(84) No doubt disposals of that kind will continue over the years, but in an
organised way. They have weathered every kind of financial storm over the
centuries and have skills which the public sector is only just beginning to
learn. (ARX: 305).
(85) Without plunging into the moral and intellectual ‘deep end’ of the arguments about retribution, I suspect there is a relationship of cause and effect
between the state of opinion and the incidence of crime [...], (A69: 683).
Thus, (83) must be paraphrased in terms of internal, semantic premodification as the political advantage/power it had lost, (84) reads every kind of financial
crisis and (85) translates: without getting into the moral and intellectual difficulties/problems of the arguments about retribution. Moreover, an internal interpretation seems mandatory for isomorphic idioms with the definite article in
their base-form. As has been argued above, these idioms are typically used in
discursive contexts in which the referent of their NPs is known to the speaker
and the hearer. If this is not the case, the figurative NP-head must rendered
specific by premodification or postmodifiation to trigger a cataphoric use of
the definite determiner. Apart from of-complementation, the topic-indicating
premodifying adjective is a further syntactic means to fulfil this function. This
can be illustrated by means of (86):
(86) I suspect that Microsoft has been reluctant to grasp the virus nettle and it
will be interesting to see how it copes with the continuous upgrading that
such a product requires. (FT8: 1045).
Nettle has the phrase-induced figurative meaning ‘problem, challenge’. The use
of the definite article implies that this meaning is actualised to concretise a
uniquely identifiable referent in the discursive context. This, however, is not
possible without the classifying topic-indicator virus that functions to restrict
its referential scope. This favours an internal interpretation of the premodifier
over an external one. Therefore (86a) and (86b) are not synonymous:
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.46 (2481-2543)
 Idiomatic Creativity
(86) a.
I suspect that Microsoft has been reluctant to deal with the virus problem (the problem related to viruses) and ...
b. I suspect that, with respect to viruses, Microsoft has been reluctant to
deal with the problem and ...
While (86a) refers to the virus problem, as the specific problem in question,
the external interpretation in (86b) merely alludes to the existence of a specific problem – which, however, is not mentioned in the context – and points
to Microsoft’s reluctance to deal with this problem as far as viruses are concerned. The non-synonymity of these two interpretations provides strong evidence against Nicolas’ uniform treatment of adjectival premodifiers in terms
of external modification.
The semantic identification-function of premodifiers becomes most obvious with appositive noun-premodification. Appositive nouns are inserted to
render the context-specific referent of the premodified constituent explicit.
In this sense their function is equivalent to appositive of-complements (as
discussed above):
(87) The intolerance he had shown for free-traders before 1914 was now turned
on rebel Unionists who rocked the coalition boat. (EW1: 1465).
In the same way as appositive topic-indicating of-complements, the premodifying noun coalition can replace the modified idiomatic constituents without
altering the idiomatic meaning:
(87) a.
The intolerance he had shown for free-traders before 1914 was now
turned on rebel Unionists who rocked (= challenged) the coalition.
The existence of appositive noun-premodification further substantiates the
trend for adjectival premodification in idiom-variants: if the varied idiom has
an isomorphic semantic structure, the modifier must be directly interpreted as
a classifier that specifies the context-specific referent of the idiomatic headnoun. For appositive premodifying nouns this classifying function is maximised: the premodifier itself defines the context-specific referent of the idiom’s
nominal constituent explicitly. One can thus posit the following trend for the
idiom-variation grammar:
Topic indication by means of a premodifier is a regular process that works
in agreement with an idiom’s semantic structure. In general, the head-nouns
of idioms of all transparency types may be premodified by a topic-indicating
(domain-delimiting) adjective or noun. If, however, the modified noun does
not have a figurative correspondent on the level of the idiomatic-meaning, the
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.47 (2543-2588)
Chapter 7. The lexicogrammatical variation of spf-idioms 
premodifiers can only be interpreted adverbially. Thus, external modification
is triggered by the presence or absence of isomorphism. For isomorphic idioms, which possess figuratively-denoting NPs, the internal interpretation is
preferable. This is particularly true for isomorphic idioms that are used in
combination with another adnominal modifier and/or with the definite article.
With these constructions, the topic-indicator must be interpreted internally to
restrict the reference of the idiomatic NP. The premodifier thus motivates the
cataphoric use of the definite article. In contrast, an external interpretation of
the modifier changes the meaning of such idiom variants. I claim this to be a
refinement of Nicolas’ position on idiomatic premodification.
In a nutshell, the overall spectrum of data on adnominal idiom-modification
clearly support my hypothesis: idioms are complex activation-sets that can be
variably unfolded in actual discourse. To the extent that it is possible to ascribe a motivated and/or analysable structure to an idiomatic configuration, it
can be systematically manipulated according to the principles of literal-scene
manipulation and topic indication. In the absence of motivating conceptual
associations, the systematic variation potential of an idiom is highly restricted.
While this speaks against a lexical view of idioms, it reflects the cognitive conditions that determine the systematic variability of idiomatic constructions:
adnominal idiom variation is a reflex of idiomatic analysability and motivation as effected by the cogniser’s ability to recruit metaphorical and metonymic
models; it cannot be explained on the basis of autonomous syntactic rules.
. Lexical substitution
Lexical flexibility is one of the central problems in the analysis of idiom variation. To what extent can the constituents of an idiom can be altered without
damaging its symbolic integrity, i.e. without interfering with the recognisability
constraint? An explanatory model of lexical variation within idioms should be
able to answer this question; it should explain how the lexical constituents can
be systematically replaced to be compatible with the conventional association
between the literal and the idiomatic meaning.
When discussing the psycholinguistic configuration hypothesis and when
introducing the recognisability constraint, we have argued that many idioms
possess lexical key elements on the basis of which the idiomatic configuration can be activated (see Sections 2.1.2 and 6.5.1). The assumption of such
salient idiomatic substructures is in agreement with the distinction between
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.48 (2588-2643)
 Idiomatic Creativity
core and peripheral constituents as proposed by phraseological analyses of idiom variation and modification (see Koller 1977; Burger et. al. 1982: 68). In
this subchapter, it is to be shown to what extent the psycholinguistic notion of
‘idiomatic key-constituent’ can be supported by variation data for spf-idioms.
Moreover, it is to be seen if these lexical keys are promoted by the underlying
conceptual correspondences that motivate the association between the literal
scene and the target scene. Lexical variants for spf-idioms reflect the following
salient substitution patterns.
.. Topic-indicating lexical substitution
A very interesting form of lexical substitution occurs in combination with the
variation principle of topic indication. With topic-indicating lexical substitution, the context-specific target of a constituent is mapped into the idiom’s
formal structure; this topic-indicator substitutes the idiom constituent that
concretises it. Topic-indicating lexical substitution reflects conceptual blending and results in a hybrid, blended image. Consider the following variant of
dipping one’s toe into the water:
(88) This is the entrance to Turbo Ted’s Nursery. British Rail is dipping its toes
into child care. (K1S: 2813).
In (88) water is replaced by its context-specific target child care. This substitution can only be explained in terms of isomorphism. Following the conceptual metaphors difficult purposeful activity is swimming, insecurity
is water and intensity of involvement in an activity is depth of position within a container (water/swimming pool = container in which
swimming takes place), the constituent water can be attributed the phraseinduced figurative meaning novel, difficult activity/issue which the
actor feels insecure of. In accordance with the swimming-script and the
corresponding motivating metaphorical backgrounds, the profiled action dipping one’s toe in the water models the event of tentatively starting a more
abstract target-activity. In (88), the nominal child care symbolises the targetactivity with which the agent, British Rail, attempts to deal. In the variant,
this topic is mapped into the idiom’s constituent structure to substitute its
literal correspondent water. In this way, the topic that is concretised by the
literal scene of the idiomatic configuration is made fully explicit in the formal structure of the variant. Topic-indicating lexical substitution thus works
as an irrefutable proof for the semantic analysability of isomorphic idiomatic
constructions.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.49 (2643-2686)
Chapter 7. The lexicogrammatical variation of spf-idioms 
The process of topic-indicating lexical substitution substantiates the insights that we have gained from the analysis of topic-indicating premodification and postmodification. Many adnominal topic indicators (postmodifying
of-complements and premodifying nouns in particular) reflect apposition: appositive adnominal topic-indicators are co-referential with the idiomatic headnouns that they modify. Therefore, appositive premodification and postmodification constitute ‘preliminary’ states of topic-indicating lexical substitution.
While the former types preserve the lexical structure of the idiomatic baseform but extend it by the adnominal topic-indicators, the latter type modifies
the lexical structure more radically and replaces the idiomatic constituent that
it stands for. To further substantiate this argument, two adnominal variants of
dip one’s toe in the water can be compared to (88):
(89) Once again, I remind the Minister of the problems into which the Government ran when they first tried to dip their toes in the water of ferry
privatisation with the Gourock-Dunoon service. (G3H: 955).
(90) Just when we thought it was safe to dip a toe into the poll tax water, we are
faced with a poll tax mark II. (HHW: 5161).
In (89) and (90), the constituent water is modified by the appositive ofcomplements ferry privatisation and the compound poll tax, respectively.
The idiomatic integrity of these variants is preserved even if they are rendered in terms of topic-indicating lexical-substitution by analogy with the
variant in (88):
(89) a.
Once again, I remind the Minister of the problems into which the
Government ran when they first tried to dip their toes into ferry
privatisation with the Gourock-Dunoon service.
(90) a.
Just when we thought it was safe to dip a toe into the poll tax, we are
faced with a poll tax mark II.
Corresponding, systematic patterns of topic-indicating lexical substitution,
premodification and postmodification occur with a great number of isomorphic idioms. The existence of topic-indicating lexical substitution and its close
correspondence to topic-indicating adnominal modification speaks for its conceptually motivated regularity in the idiom-variation grammar.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.50 (2686-2739)
 Idiomatic Creativity
.. Synonyms, antonyms and perspectival variants
A salient form of lexical substitution consists in creating a synonym or
antonym of the idiom’s literal meaning to trigger a corresponding synonym or
antonym of the idiomatic meaning. Technically, this works by replacing a constituent verb, adjective or preposition by a synonym or antonym. For instance,
antonymous substitution is also possible for some motivated idioms that point
to an inherent quality. Consider the following variants of the idioms be on the
right track (91a–b), the fast track (92a–b) and the fast lane (93a–b):
(91) a.
I knew I was on the right track when I felt that thrill of pleasure at
placing object, not painting it (A08: 2528).
b. Any delay in arriving at a solution will be due not to our being on the
wrong track but simply to the enormous complexity of the problem.
(A0T: 1238).
(92) a.
The Gatwick express, on the fast track to the private sector. (K6A:
564).
b. Because of the Government’s apparent lack of enthusiasm for all
things European, and their determination to go along a slow track,
will not Scotland lose out again without any chance of the central
bank being cited in Glasgow or Edinburgh? (HHV: 8634).
(93) a.
Election 1992: Gillian and Virginia take the fast lane to the top. (AKH:
450).
b. Meanwhile, Scotland is expected to remain in the slow lane of
economic recovery, according to two surveys published yesterday.
(K5D: 10104).
The creation of these antonyms is motivated by the underlying conceptual
metaphor: quality of basis of development is quality of path. In accordance with this metaphor, the conceptual cores track/lane can be modified with attributes (right vs. wrong and fast vs. slow) that indicate the
(in)appropriateness and quality of the developmental bases. On the lexical surface, this leads to the substitution of the corresponding adjectives by their
antonyms. Although the creation of these antonymous alternants is conceptually motivated, the actual occurrence and lexical realisation of such variants
often is not fully predictable. While some spf-idioms show antonymous tokens frequently, others do not. Thus, although the present cognitive-linguistic
framework can explain the conceptual motivation for the creation of an idiomatic antonym, it cannot delimit these differences in actual occurrence. The
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.51 (2739-2795)
Chapter 7. The lexicogrammatical variation of spf-idioms 
same is true for the presence or absence of synonymous variants. Consider the
following uses of be on/walk a knife’s edge and skate on thin ice:
(94) a.
CRISIS club Barnet’s future was balanced on a knife edge last night
after Football League officials rejected their request for £91,000 withheld in lieu of fine payments and commission costs. (K3P: 696).
b. In the last waning glow of daylight, she saw Nicolo’s face tauten with
fury as he teetered on the knife-edge of control. (JY7: 5755).
c. Their new feature film ‘Dream On’ is a delightful blend of magic
fantasy and dream, which treads a knife edge between humour and
tragedy. (KS8: 1102).
(95) a.
The book is perhaps especially valuable for the way in which the author is prepared to extend onto the thin ice of speculation and comparison with other animals, a trait sadly lacking in so many agricultural
scientists. (B7M: 1571).
b. She felt as though she were treading on thin ice and any false
move might plunge her once more into the dark depths of despair.
(JYE: 4440).
c. ‘Do you want,’ Anna said, embarking impulsively on the thinnest ice,
‘do you want to reconsider everything? (CMJ: 1281).
These variants are all motivated by the same conceptual backgrounds: progressing is moving forward, basis/context of development is a path,
quality of basis/context is quality of path and successful development is physical balance. The variants exploit these backgrounds in a systematic way. While leaving the path-concepts unaltered, the verbs are substituted to qualify and intensify the underlying image of moving on an unstable
basis. Tread, teeter, be balanced on, extend on and embark on all describe alternative forms of the mover’s interaction with the respective path-concepts.
Although the choice of these specific lexical items cannot be predicted, they
elaborate the literal scene systematically. Some of these substitutions lead to an
intensification of the idiomatic meaning or increase the stylistic formality with
which the literal scene is coded. Nevertheless, they can be regarded as being
grossly synonymous to the meanings of the corresponding base-forms.
Apart from such synonyms and antonyms, a considerable number of verbal idioms vary their lexico-syntactic structure systematically to construe the
literal scene – a state or action – from alternative perspectives (see Section
6.1.1). Perspectival variants such as get the show on the road vs. keep the show
on the road involve alternative construals of possession, causation, aspect and
reciprocal relationships. In cognitive-linguistic terms, perspectival variants re-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.52 (2795-2854)
 Idiomatic Creativity
flect alternatives of how the relational predication between a trajector (usually
the subject or object, e.g. show) and a given landmark (usually a locative PP,
e.g. on the road) is conceived. Again, the variation-data for spf-idioms suggest
that conceptual metaphors play an important role in motivating these variants.
Following the archetypal metaphorical model being in a state/situation is
being in a location, a number of spf-idioms characterise favourable and
unfavourable (developmental) states/situations in terms of alternative
source-domain scenarios that express an agent’s position relative to a specific
location (see Section 5.2.7). Following the logic of this metaphorical conceptualisation, causing or initiating a state, as well as entering, maintaining and
leaving it are conceived by modulating the agent’s position relative to the spatial
location in the source domain. On the linguistic surface, these manipulations
of the literal scene result in verb substitutions. Consider the following variants
of walk into the lion’s den:
(96) ‘We’re in the lion’s den. (HTY: 3840).
(97) I tremble at the thought of Mr Kinnock entering the lion’s den of Europe,
going ‘naked into the conference chamber’, as his old hero Aneurin Bevan
put it. (AHN: 1937).
(98) They wouldn’t be expecting her to go further into the lion’s den. (CH2:
7950).
(99) At 15, Davey was suddenly thrown into the lion’s den on the Saturday afternoon programme with such home-grown greats as Mick McManus, Jackie
Pallo, Big Daddy and Les Kellett. (CH5: 3132).
In the context of the idiom, the nominal lion’s den has the autonomous semantic value difficult/dangerous situation. This sense it is motivated by
the metaphors: situation as container/location, quality of situation is
quality of container/location; these metaphors interact with the struggle for life-model and the metaphor a problem is a dangerous animal,
with lion being an emblem for ferocity. Since it concretises the difficult situation that a given agent is confronted with, lion’s den constitutes the figurative
core of the idiom. Relative to this core, the agent’s state can be perspectivised.
Thus, (96) involves a static construal of the subjects’ (we) being in a difficult
situation. In the literal scene this corresponds to their being in the container
(the den). In contrast, (97) construes the literal scene more dynamically by
highlighting Mr. Kinnock’s active decision of putting himself into a difficult
situation. On the linguistic surface, this alternative perspective is reflected by
the verb variation be vs. enter. In (98) the dynamic conception of entering a
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.53 (2854-2898)
Chapter 7. The lexicogrammatical variation of spf-idioms 
difficult situation is intensified by further manipulating the literal scene relative to the conceptual metaphor intensity of involvement in a situation
is depth of position within a container. Thus, on the level of the idiomatic meaning, the subject’s going further into the lion’s den corresponds to
her getting into more trouble. In (99), finally, the literal scene is construed to
reflect a causative relationship: the subject (Davey) is brought into difficulties
by some unknown agent. Again, this perspectivisation is triggered by manipulating the literal scene through verb substitution: the force-dynamic transitive
verb throw is used instead of go or enter, which reflect self-propelled movement. Consequently, the passive subject (Davey) can be construed as being the
victim of malignancy rather than being the initiator of his own difficulties.
In a nutshell, (96)–(99) reflect systematic verb variation round the nominal
lion’s den, which functions as the conceptual core of the literal scene. The systematic substitution of the verbal periphery is motivated and supported by
the underlying metaphorical conceptualisation of changing state is changing position relative to the container/location. Thus, the conceptual
metaphors make a strong and active contribution the creation of these variants. Further examples to substantiate this trend are listed in the electronic
appendix (see, e.g., back to square one, from square one, gain/lose ground, get off
the starting blocks, etc.).
Qualitative departures from the originally motivating literal scene are reflected in the following variants of in the doldrums as well. As walk into the
lion’s den, this idiom is motivated by the state as location-metaphor. With
this particular idiom, the archetype-metaphor was originally concretised on
the basis of the voyage-frame. More specifically, doldrums denotes a windless
area on the ocean where sailing ships cannot move forward. Relative to this
frame, doldrums can be attributed the autonomous phrase-induced figurative
sense ‘state of stagnation or developmental difficulties.’ Again, the literal scene
in the doldrums can be manipulated to create different perspectives on this unfavourable state. On the lexical level, this is reflected by verb and preposition
variation:
(100) That’s why sports cars went through such doldrums for so many years too,
[...]. (A6W: 1187).
(101) Traditional Irish combativeness was nowhere to be seen as they allowed
Robert Jones and his back row to steer Wales out of the doldrums.
(CB2: 698).
(102) There was a feeling that Mill House might just be coming out of the doldrums: [...]. (AD7: 94).
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.54 (2898-2945)
 Idiomatic Creativity
In (100)–(102) the actual scenario of physical movement into, through, and
out of the unfavourable location is highlighted to concretise the actual state
of progress or stagnation of the developing configuration. However, the original source-scenario of voyage is not directly elaborated by these variants.
Rather, they evoke the doldrum-concept as a general physical area or container that has to be passed, avoided or left to regain developmental momentum. So instead of manipulating the pre-coded voyage-scenario, these variants
are strongly motivated by the more abstract intensity of involvement in
a situation is depth of position within the container-metaphor. This
archetype-metaphor is particularly prominent in the following variants, whose
literal-scene cannot be motivated on the basis of the voyage-domain.
(103) SIEMENS NIXDORF PLUNGES INTO THE DOLDRUMS. (CNN: 171).
(104) Firstly it belongs to John Major himself, for taking to the hustings, and
restoring, a cutting edge to the party dialogue – and thus taking the Central Office by the scruff of the neck and lifting it out of the doldrums into
which it had sunk. (B03: 1136).
The verb substitutions plunge into and lift out of are clearly motivated by
the container-model. Thus, this conceptual metaphor determines the production of these variants. The active presence of this conceptual background
instead of the voyage-frame indicates that, for many speakers, in the doldrums
is no longer motivated by the latter model – its derivational basis. This does
not speak against the cognitive-linguistic model of idiom representation advocated in this study. Although the original source-domain, voyage, does not
seem to be associated with in the doldrums any longer, the conceptualisation
of doldrums as a container clearly reflects the speakers’ comprehension of
this idiom relative to a system of conceptual metaphors (as outlined in Section
5.2), rather than processing it as an unmotivated long word. With a view to
the Glucksbergian arguments against the influence of conceptual metaphors
on idiom processing (Section 2.3.3), these variants suggest that conceptual
metaphors are activated. However, speaker’s have the relative freedom to remotivate idioms against different metaphorical ICMs. In other words, conceptual
remotivation-processes cannot always be fully predicted. But what is important
from a cognitive-linguistic perspective is that complex metaphorical-models
are recruited and accessed for the purpose of idiom-representation and (most
forms of) variation.
Systematic perspectival variability is also reflected by idioms that are motivated by the metaphtonymic conceptualisation being in a state of develop-
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.55 (2945-3000)
Chapter 7. The lexicogrammatical variation of spf-idioms 
ment/activity is being on a path. Get the show on the road, be on the rails, be
on course and be on the right track instantiate this metaphtonymy relative to the
more specific car journey, railway journey and voyage frames. The shared
metaphtonymic background connects these idioms mentally – they are governed by the same conceptual associations. This cognitive cohesion is reflected
in their congruent lexical variability. Thus, following the metaphtonymy, all of
these idioms show parallel perspectival variants:
(105) a.
The Peoples’ Charter, [...], was approved by the Conference, the show,
as they say, was on the road. (GXG: 2192).
(106) a.
Mr Kinnock, reaffirming his confidence that Labour was on course for
an overall majority, dismissed reports that Mr Ashdown was seeking
four Cabinet places in a coalition with Labour. (AJ6: 64).
(107) a.
The warm reception given to Project 90, designed by John Heffernan
and Ken Greenley, provided positive proof to Ward that he was on the
right track in his long-term strategy of building a separate and much
stronger image for the Bentley marque. (A6X: 334).
The literal scenes in (105a)–(107a) describe the static constellation of the
mover being on the path as reflected by the use of be. On the level of the
idiomatic meaning this static construal of the literal scene indicates that the
respective subjects (the Peoples’ Charter, Labour, Ward) are in a state of successful development. To highlight the desired continuation of the configuration’s
development, the literal scene can be varied by substituting the verb keep (on
the road/course/the rails/the right track) for be as in (105b)–(108b):
(105) b. Car dealers were on a collision course over the Chancellor’s drive to
keep the motoring industry on the road to recovery. (CF8: 455).
(106) b. Their tasks will be to win over the hearts of the doctors and teachers,
while keeping the reform programme firmly on course. (AKH: 790).
(107) b. It also keeps me on the right track as well. (JJP: 306).
(108) b. We are all going through hard times in the construction industry and
I just can’t deal with endless interruptions and keep a firm on the rails
as well.’ (K55: 5965).
In contrast, the forced termination of development can be conceptualised
metaphorically in terms of taking or forcing the mover off the path. In (105c)–
(107c), this causative construal of wilfully hindered movement and development is coded by using the force-dynamic verbs knock off, force off, and
shunt off :
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.56 (3000-3062)
 Idiomatic Creativity
(105) c.
PREMIER John Major was ambushed by the Germans last night in
a move which could force Britain off the road to European unity.
(CBF: 9880).
(106) c.
STEFAN EDBERG’s Wimbledon preparations were shunted off the
rails last night as he crashed to a humiliating defeat [...]. (K3X: 1595).
(107) c.
The postponement of the most painful tax increases – [...] – is in order
not to knock off course the fragile economic recovery. (K5M: 3024).
In accordance with the logic of the motivating metaphtonymy, recovering successful development corresponds to getting back on the path. This process
can be supported by an external force (an agent) which helps the developing configuration reassume a successful basis of development as implied in
(105d)–(108d):
(105) d. [...]. Rainbow gets the show back on the road, and tries to give herself a
little holiday from auntly interventions by searching the radio bands
for a local traffic report. (HGN: 1695).
(106) d. MICHAEL THOMAS will be handed the chance to set his Liverpool
career back on the rails this weekend. (CBG: 4198).
(107) d. The 20-year-old from St Helens scored five goals in four games to put
City back on course [...]. (CBG: 5873).
(108) d. Several key factors will influence just how quickly Ireland can get its
rugby back on the right track: fitness, tough competition and coaching.
(CHW: 1129).
The great amount of overlap between these perspectival variants shows that
the underlying conceptual metaphtonymy being in development is being
on path motivates all of these idioms in the same systematic way. The conceptualisation of be on a path constitutes the conceptual core of the literal
scenes. On the level of the idiomatic meaning, this core must be paraphrased
as be in a state development/progress. The conceptual core can be subject
to alternative perspectival construals. On the lexical level, these construals lead
to systematic verb variation. Hence, the distinction between core constituents
and peripheral constituents can be readily explained relative to the motivating
conceptual model. The variants are driven by the underlying conceptual motor. The core is a scene-setter and the conceptual periphery varies according to
which aspect of the scene is focused on.
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.57 (3062-3141)
Chapter 7. The lexicogrammatical variation of spf-idioms 
The strong conceptual coherence that connects these idioms mentally
is further underlined by the following variants, which show that the pathconcepts may be mutually exchanged occasionally:
(109) [...] those of us who are obliged to use Network SouthEast at fares ‘much
higher than in Europe to help keep the whole show on the rails’ might be
pardoned for taking a less sanguine view. (EDT: 481).
(110) [...] only a careful analysis of natural conversational data can get us on the
right road again. (HGH: 219).
(111) She needed someone [...] to agree that she was on the right course,
[...].(AD1: 3235).
Thus, in (109) rails instead of road is used as a path-concept. In (110) and
(111), the constituent track in the expression, on the right track, is replaced by
the alternative path-words road and course, respectively. The strong cognitive
cohesion on the conceptual level creates this potential for lexical substitution.
This cognitive-linguistic argument can be further substantiated with idioms for which several salient constituents may act as idiomatic keys interchangeably. Consider the following variants of swallow a bitter pill and the
conceptually corresponding expression sugar the pill (make and unpleasant
situation more acceptable):
(112) a.
Someone he loved more than he loved her, which was a hard pill to
swallow. (H85: 1685).
b. And a firm’s environmental history has emerged as another potential
poison pill no-one wants to swallow. (HC2: 700).
c. THE relegation clouds descended on Darlington last night as Ray
Hankin was obliged to swallow his first bitter taste of managerial
defeat. (K52: 3975).
d. ‘Claudia swallowed bitter disappointment; Roman didn’t even care
enough about her to confide in her, [...] (H8J: 2347).
e. Mr Kohl’s glossy coating of the bitter pill of economic union undoubtedly won him votes, but the coating has already proved to be thin.
(ABK: 375).
f. Mr Lamont will sugar the bitter pill with another one per cent drop in
interest rates [...]. (CEN: 4127).
In (112a–f) none of the lexical constituents remains constant. The canonical
nominal bitter pill in swallow a bitter pill can be subjected to both variation
of the head and the premodifying adjective to adopt the occasional forms: hard
pill, poison pill, bitter taste, bitter disappointment. In addition, bitter pill can also
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.58 (3141-3165)
 Idiomatic Creativity
be evoked as a constituent of the idiom sugar the pill, which can be used variably itself. This indicates that the two idioms and their corresponding variants
are strongly motivated by the fixed conceptual scenario being urged to consume an unpleasant drug/substance, which is evoked in accordance with
the conceptual metaphors dealing with an issue is consuming it, quality
of issue is quality of consumed substance. The striking amount of lexical
variability reflected by (112a–f) can only be explained if one conceives the idioms swallow a bitter pill and sugar the pill as complex mental activation-sets
that consist of the entrenched association of a literal scene with an idiomatic
target-scene; the association between the two scenes is motivated and structured by the conceptual metaphors. The lexical implementation of this association, i.e. the canonical symbolisation of the literal scenes, consists of the salient
key-constituents: swallow, sugar, bitter and pill. Since they are backed by the
metaphorical model, all of these constituents can be attributed an autonomous
phrase-induced meaning: swallow = accept, sugar = improve, bitter = unpleasant and pill = issue. In accordance with these meanings and the motivating
conceptual background, the lexical structure of the two idiomatic configurations can be altered in discourse to the extent that they remain recognisable.
In (112a–f), this is achieved by the combined use of two salient constituents
while the third one is varied. Thus (112a–b) contain the constituents pill and
swallow and vary the adjective bitter, (112c–d) include bitter and swallow but
replace pill by another noun from the domain unpleasant substance/issue,
whereas (112e–f) feature bitter and pill but alter the verb. In other words, the
variants show that the figurative core of the respective idioms may be variably distributed over different key-constituents. The idiomatic configuration
can be activated if two of these constituents are co-activated in combination
with a third lexical item that is also compatible with the underlying conceptual
motivation.
These idiom variants are empirical grist for the cognitive-linguistic idiomaticity mill. Lexical fixedness is strongly relativised for these idioms. What
they share is the fixed conceptual core (on the path = in state of development; swallow unpleasant substance = accept an unpleasant fact) in
relation to which their lexical implementation is organised and becomes variable. The evocation of the core-concepts via the lexical key-constituents (e.g.,
rails, road, course, doldrums, lion’s den, swallow, pill, bitter) is sufficient to guarantee that the idiomatic configuration is recognised for the idiomatic meaning
to be activated. With a view to Moon’s (1998: 121) statement that idiomatic
fixedness “is not always lexical fixedness,” we can therefore propose that fixedness resides in the mental entrenchment of the lingua-conceptual substructures
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.59 (3165-3214)
Chapter 7. The lexicogrammatical variation of spf-idioms 
of idiomatic activation-sets rather than a fully codified lexical sequence. Given
this revised trend for lexical substitutions within isomorphic and motivated
idioms the following variants of be on a knife’s edge, upset the applecart and
skate on thin ice can be readily accommodated. Although the figurative cores
of these idioms (applecart, knife’s edge and thin ice) are varied, the idiomatic
reading of the resulting variants is preserved because other salient constituents
or conceptually corresponding words can compensate for this deviation: they
are sufficient to code and evoke the underlying literal scene:
(113) Life has always been on a razor-edge of survival [...]. (H7K: 399).
(114) That’s nice. Well, towards it, upset the cart now ain’t she?. (KBE: 8964).
(115) Cumulatively, however, the aura of association with the US was of immense political benefit to those who called themselves a provisional government but who were still skating on the thin surface of political respectability and had by no means attained a state of acknowledged legitimacy. (EFA: 310).
In conclusion, the trend observed for lexical substitutions in spf-idioms
can be reformulated as follows: if idioms have a relatively autonomous
conceptual-core, this core and the corresponding constituents are sufficient to
activate the idiomatic configuration as a whole. The core may be encoded by
a fixed lexical substructure (doldrums, lion’s den, tightrope) or the combined
evocation of salient co-occurring constituents (bitter pill, swallow bitter, swallow pill, skate on thin, on X-edge). These items function as interchangeable keys
or formal trigger-points on the basis of which the idiomatic meaning can be
retrieved or activated. Relative to the pre-coded, stable conceptual core, the
literal scene can be manipulated and encoded creatively (through lexical substitutions) as long as this is compatible with the idiom’s underlying conceptual
motivation. In a nutshell, idiomatic fixedness is defined by its complex and
dynamic interplay of conceptual and lexical substructures in relation to an
‘amoebic’ idiomatic activation-set. In accordance with the general hypothesis of this study, this trend is most fully pronounced with isomorphic idioms
that have constituents with lexicalised figurative senses; it also occurs with isomorphic and motivated idioms, but it is not present with opaque idioms. In
other words, strong conceptual motivation seems to support systematic lexical flexibility, while opacity blocks such variation. For opaque idioms all basic
lexical constituents must be present to activate the idiomatic meaning (parasitic elaboration being excluded). Since these idioms cannot depend on a
well-structured conceptual scaffolding that works as a transparent model for
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.60 (3214-3266)
 Idiomatic Creativity
the idiomatic meaning, their lexical structure is the main mental factor to provide for the internal cohesion of the idiomatic activation-set. Put differently,
with opaque idioms the standard lexical implementation carries most of the
mental ‘responsibility’ for triggering the idiomatic meaning. In this sense, idioms provide fascinating clues that point to the close and dynamic interaction
of conceptual and linguistic structuring. They are a window that allows us to
peek into the intricacies of intelligent human cognitive processing.
. Controversial data
After having provided overwhelming evidence for the explanatory power of
the cognitive-linguistic model for idiom representation and variation, we now
have to point to its limits. Although the trends outlined above can explain the
majority of variation-data extracted from the BNC, it is not the case that the
model can explain all variants. Consider the following systematic variants of
the highly opaque idiom come a cropper:
(116) a.
That overweight bounder Billy Bunter has come the most frightful
cropper at the BBC. (CH2: 2403).
b. Unless you learn how to control your work and recreation you’ll come
an almighty cropper and it will be your own fault.’ (FEU: 51).
c. Tories more famous than he had come nasty croppers in the past at
conference time, returning tight to their hotels. (HNK: 841).
d. DURHAM Squash Club’s challenge for the Durham and Cleveland First Division championships has come a catastrophic cropper.
(K52: 2980).
e. I don’t know he’s gonna come a right bloody cropper. (KCG: 1602).
Following the central argument of my study, the use of premodifiers should be
explicable on the basis of underlying patterns of semantic extension that motivate the internal semantic structure of the idiom. For the present tokens of
come a cropper, this argument does not hold. In my view, there is no obvious
conceptual basis in relation to which these patterns of premodification could
be explained. The only plausible explanation that I can present follows Langacker’s usage-based conception of grammar: the variants in (116a–f) seem to
follow a regular variational pattern. The adjectival premodifiers consistently
work as intensifiers of the noun cropper. One can therefore speculate that
these variants are created on the basis of an entrenched constructional schema:
[come a intensifier cropper]. Following Langacker’s usage-based view, such a
JB[v.20020404] Prn:20/02/2006; 15:20
F: HCP1707.tex / p.61 (3266-3284)
Chapter 7. The lexicogrammatical variation of spf-idioms 
schema could be derived from recurrent usage-events that involve the use of
come a cropper according to the intensifier-pattern. This would suggest that
some restricted types of idiom variation with opaque idioms could be subject to the entrenchment of constructional schemas that guide the constrained
range within which the alterations can be produced. Note that Moon develops
a similar argument in the context of idiom passivisation. Discussing the frequencies found for the passives of some idioms in the Bank of English, Moon
(1998: 109) states:
There may indeed be semantic motivations here, but phraseological patterning also plays a part. Cases like spill the beans show a strong fossilization
in an active structure, irrespective of potential passives and deep semantics.
This is another area which needs to be teased out more fully in the light of
much larger corpora and robust models of case structures, verbal processes,
and meanings.
Such a view of idiom variation would again be phraseological in nature and underline the dynamic interplay of routine and creativity in language use (cf. also
Kay & Fillmore 1999). This, however, is where my cognitive-linguistic account
of idiom representation and variation has to stop.
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.1 (47-115)
chapter 
Conclusions and outlook
This book dealt with the phenomenon of idiomatic creativity from a cognitivelinguistic perspective. It explored the hypothesis that many idioms can be attributed an intrinsically creative semantic structure. In addition, it was claimed
that the degree to which this structure can be (re)motivated by a cogniser in
relation to a set of entrenched conceptual metaphors, metonymies and emblems correlates with their potential for systematic lexicogrammatical variation
(Chapter 1).
This hypothesis addressed a central controversy in the field of theoretical
and psycholinguistic approaches to phraseology: the question whether idioms
must be regarded as word-like lexical units or whether they can be described
as semantically analysable, compositional constructions (Chapter 2). The study
approached this hypothesis from a comprehensive cognitive-linguistic perspective (Chapter 3). It proposed a systematic cognitive-linguistic model of idiom
representation and variation, which constitutes an integrated and theoreticallymotivated attempt to re-interpret these phenomena. In doing so, the model
paid tribute to both the heterogeneous nature of idioms as well as the diverse
communicative qualities of different types of idiom-variation.
Following the basic principles of Cognitive Grammar, idioms were described as complex symbolic units. The model anatomised the formal and semantic structure of these units cognitively by characterising them as idiomatic
activation-sets (Chapter 4). Idiomatic activation-sets were conceived as mental
networks that consist of a structured set of symbolic and conceptual substructures. The semantic pole of these units constitutes a complex conceptual scene:
the abstract scene comprising the idiomatic meaning is conceived against the
conceptual background of a literal scene. This complex semantic pole is organised along various dimensions of internal semantic structuring, most importantly, motivation and isomorphism/analysability. Following the cognitivelinguistic account, both motivation and isomorphism emerge from a speaker’s
ability to relate the association between the literal and the idiomatic meaning to
complex patterns of conceptual metaphor, metonymy, blending and emblems.
These patterns function as extensive conceptual bases against which the seman-
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.2 (115-161)
 Idiomatic Creativity
tic structure of idioms can be assessed by a cogniser. In accordance with the
highly variable nature of figuration, the internal semantic structure of different
idioms is highly heterogeneous itself. By means of the idiomatic-activation-set
model, different but salient idiomatic types and subtypes of figuration can be
established in relation to these conceptual backgrounds.
With a view to the controversy about the influence of conceptual metaphor
on idiom motivation (Section 2.3), the model was applied to analyse the mental
representation and variation of idioms belonging to the word-field of success,
progress and failure (spf). First, it was shown that many of these idioms
can be attributed an internal semantic structure relative to a highly coherent
system of metaphorical models (Chapter 5). Thus, the orthodox view of idioms as semantically-unmotivated and unanalysable units could be refuted.
If idioms are not only conceived as isolated lexical units but as pre-coded
cognitive-micro-models that can be related to a complex architecture of alternative metaphorical, metonymic and emblematic ICMs, their conceptual motivation and cognitive functionality can be revealed. Accordingly, the semantic
regularity of spf-idioms was charted relative to well-entrenched conceptualmetaphor systems.
To model the phenomenon of idiom variation in cognitive-linguistic terms
(Chapter 6), idiomatic activation-sets were desribed to constitute the mentallyentrenched idiom-configuration-standards that can be potentially activated
by a speaker in discourse. Idiom variation emerges as the process of adapting an idiomatic construction in order to encode a context-specific target
conceptualisation in a communicatively functional and relevant way. Depending on their internal semantic structure, idioms function as alternative types
of cognitive micro-model. Idiom variation is the process by which an idiomatic construction is adapted to the usage-context to fulfil its cognitive
modelling function. However, this adaptation process is not a homogeneous
phenomenon. To account for this heterogeneity, five alternative principles of
idiom variation were described: constructional adaptation, literal-scene manipulation, topic indication, topic-related literal-scene manipulation and ambiguation/punning. These principles define salient areas on a cline of idiomatic
creativity that ranges from fully systematic alterations to strikingly conspicuous forms of wordplay. Correspondingly, the principles constitute alternative
strategies to adapt an idiomatic construction to the discursive context. The first
three of these are claimed to be systematic.
Systematic idiom-variation was claimed to be motivated as well as constrained by the internal organisation of idiomatic activation-sets. Since this
organisation is itself dependent on the presence of underlying patterns of se-
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.3 (161-218)
Chapter 8. Conclusions and outlook 
mantic extension, systematic idiom-variation was hypothesised to be dependent on the speaker’s ability to (re)motivate a given idiom in relation to these
underlying conceptual bases.
Putting cognitive-linguistic light on the controversy about idiom variation
and variability, the actual discursive behaviour of a subset of these spf-idioms
was tested on the basis of corpus-data extracted from the BNC (Chapter 7).
The data pointed to straightforward qualitative trends for the correlation of
idiom-transparency-type and lexicogrammatical variability. Relative to these
trends, a basic cognitive-linguistic idiom-variation grammar was outlined. As
hypothesised, it could be shown that the lexicogrammatical behaviour of these
spf-idioms accords with how the metaphorical backgrounds organise their semantic structure internally. In a nutshell, the study substantiated the following
basic assumptions about idiom representation and variability:
1. A great number of idioms are not semantic units, but have a motivated and
analysable semantic structure.
2. Idiom-transparency has a conceptual basis, it emerges relative to wellestablished ICMs and entrenched patterns and mechanisms of semantic
extension.
3. Idioms do not function as words, but constitute linguistically pre-coded
cognitive micro-models.
4. Systematic idiom-variation is a reflex of idiom-transparency, i.e. conceptual structuring. Systematic lexicogrammatical alterations are constrained
and motivated by the specific quality of a given idiomatic activation-set.
5. General syntactic rules of article use, number variation, adnominal modification and lexical substitution are preserved to the extent that an idiom
can be semantically penetrated in relation to the conceptual backgrounds.
6. Attempts to explain the syntactic flexibility of idioms relative to autonomous syntactic transformations or rules can be rejected with a view
to the existence of alternative, conceptually motivated types and subtypes
of idiom-transparency.
7. The systematic lexicogrammatical behaviour of idioms can be explained in
terms of the speaker’s ability to manipulate an idiomatic activation-set to
make it fulfil its cognitive modelling function effectively.
8. Idiomatic creativity reveals intelligent human behaviour that is based
on general cognitive principles and processes. Idiomatic creativity is linguistic competence as holistic cognitive competence. It emerges as the
speaker’s figurative competence to manipulate an idiom as a cognitive
micro-model – a mental network that can be evoked to organise and com-
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.4 (218-278)
 Idiomatic Creativity
municate the abstract conceptual relationships in a target conceptualisation figuratively.
On balance, the study has revealed trends in the representation and variation of
idioms that falsify their classical treatment as semantically non-compositional
lexical units.
Having highlighted the insights gained from a cognitive-linguistic approach to idiom representation and variation, it is now necessary to turn
to the limitations of this study and to look out for corresponding implications for future research. Both idiom representation and idiom variation are
highly complex phenomena that are strongly influenced by the heterogeneous
linguistic status of idioms and the mental correlates that underlie this heterogeneity. Although the description of idiomatic configurations as complex
mental activation-sets makes it possible to structure and model the complex
grammatical space occupied by these constructions, it must nevertheless be
modestly conceded that it is not possible to draw irrefutable conclusions about
the variability of idioms on the basis of this heuristics only. The following limitations of the study and corresponding desiderata for future linguistic and
psycholinguistic research can therefore be highlighted:
a. Desiderata for (cognitive-)linguistic research
1. This study analysed the systematic motivation and variation of idioms that
belong to the word-field of success, progress and failure. Idioms of this
word-field are motivated by a coherent system of conceptual metaphors.
Moreover, this system exerts a strong influence on the variable lexicogrammatical behaviour of these constructions. To substantiate these insights
further, it would be desirable for future research to extend the cognitivelinguistic approach to include idioms from other words-fields, e.g. criticism (throw NP to the wolves/lions), power (pull strings), secrecy (keep
one’s lights under a bushel), mind (have a bee in one’s bonnet) or communication (chew the fat) in order to find out if the trends revealed in this
study can be directly transferred to these other domains. On the basis of
the present evidence, one should predict that they do so.
2. The present study used corpus-data from the BNC to investigate the systematic variability of idiomatic constructions. Although this corpus offers
a massive database of spoken and written language, idioms in general
and idiom-variants in particular tend to occur relatively infrequently in
this 100-million-word collection of British English performance. Thus, although the data could be used to discover some salient qualitative trends
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.5 (278-324)
Chapter 8. Conclusions and outlook 
that support the proposed model, they did not suffice to further substantiate the trends on a well-motivated quantitative basis. In other words, future
cognitive-linguistic and corpus-based research on idiom variation should
attempt to incorporate even more extensive databases as, for instance, WebCorp to substantiate or qualify the insights of the present study.
3. Moreover, it could be revealing for further analyses to explore the distribution of alternative forms of idiom-variation in spoken vs. written language
as well as alternative text genres. For instance, topic indication seems to
be a variation process that is associated with written language, particularly
the language of the press (cf. Moon 2001). At this point, we can only speculate that in spoken language the actual topics of the NPs of isomorphic
idioms appear as implicit referents in the speech-context. Therefore, they
do not have to be specifically indicated. More extensive research on how
the different idiom-variation strategies interact with the communicative
demands of different genres and usage-context could therefore refine my
cognitive-linguistic account in socio-pragmatic terms.
4. Approaching idiom representation and variation from a cognitive-linguistic
perspective, this study has centrally focused on the conceptual motivations
for and constraints on idiom-variation. Idiomatic constructions, however,
are not only characterised by their internal semantic organisation relative to conceptual representations, but also by their discourse-functional
properties: idioms are used in certain discursive contexts to fulfil specific
communicative purposes. In other words, idioms carry the intrinsic creative potential to derive a great deal of weak implicatures from the literal
scene that they convey. These conceptual and emotional entailments structure their ideational, interpersonal as well as textual coherence. The present
study has primarily defined the discursive relevance of idioms in terms of
the concretising and evaluating cognitive-modelling-function. For future
research, it would be desirable to extend cognitive-linguistic analyses of
idiom-variability to examine other and more context-specific discoursefunctional properties. For instance, in the BNC idioms such as walk a
tightrope were found to be used with singular NPs only. However, the conceptual approach predicts that the NP tightrope could also be used in the
plural since it conveys the countable, phrase-induced figurative sense ‘difficulty, difficult situation.’ Is the predominance of singular forms therefore
caused by the specific discourse-functional properties of this construction?
The corpus-data indicate that this might be the case. More specifically,
the usage-tokens revealed that the construction is predominantly used to
concretise specific and singular problem-situations that are of immedi-
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.6 (324-380)
 Idiomatic Creativity
ate interest in the usage-context, rather than to point to various general
difficulties. Such potential discourse-functional constraints on systematic
idiom-variation cannot be explained on the basis of underlying conceptual backgrounds only. Further (cognitive-)pragmatic research is therefore
needed in this area.
5. My cognitive-linguistic model of idiom representation accommodates the
fact that many idioms cannot be motivated by speakers. For instance,
come a cropper is a highly opaque idiomatic construction. Accordingly, it
is not possible to ascribe a meaningful semantic structure to this idiom
on the basis of the present account. Nevertheless, several variants could
be found for it, most of them featuring intensifying adjectival premodifiers (come an appalling cropper, come a very nasty cropper, etc.). These
variants clearly cannot be explained as an effect of manipulating the idiom as a cognitive micro-model. As a consequence, one must concede that
some restricted patterns of idiom variation are probably memorised, rather
than creatively produced by speakers. For the present example, one can
speculate that the cognitive grammars of the respective speakers include a
constructional schema from which the variants can be readily constructed:
[come an intensifier/adjective of extent cropper]. This would speak
for the presence of institutionalised variational patterns, rather than creatively produced variants. Such an account of idiom-variation would be in
accordance with Langacker’s usage-based conception of language, which
regards constructional schemas as digested performance. To investigate
this problem further, corpus-based approaches could be combined with
psycholinguistic methods. The corpus investigation would have to explore
to what extent the variants of a given idiom are frozen within a specific
constructional schema. The corpus-based data could then be used in psycholinguistic experiments to check comprehension times for the variants
as compared to the standard configurations. If the variants are themselves
stored as entrenched constructional schemas, then they should be processed as fast as the unvaried base-forms. This leads me to the psycholinguistic implications of my analysis.
b. Desiderata for psycholinguistic research
1. A first requirement for psycholinguistic analysis concerns individual variability with regard to idiom representation and idiom processing. Since
idiom-transparency is a cognitive phenomenon, speakers must be expected
to vary in terms of how they understand, store and process different idioms
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.7 (380-431)
Chapter 8. Conclusions and outlook 
in their minds. A given idiom cannot be expected to be equally represented
and motivated for all speakers. For instance, I have claimed that idioms
such as have had one’s chips can only be motivated if it is possible for a
speaker to relate this expression to the relevant gambling-domain. Thus,
since idiom variation seems to be dependent on idiom representation, different idioms cannot be unequivocally regarded as being equally variable
for all speakers. To explore this problem of individual processing differences more extensively, more sophisticated psycholinguistic experiments
on idiom variation and representation are needed. The present model of
idioms as complex activation-sets could offer a classificatory starting point
for future psycholinguistic tests.
2. Thus, a further implication for future psycholinguistic research concerns
the heterogeneous nature of idiom transparency and idiom variation.
Many psycholinguistic studies have based their idiom processing and idiom motivation experiments on very rough idiom classifications only. Usually, these classifications are based on binary oppositions such as motivated
vs. unmotivated or decomposable vs. non-decomposable idioms. However,
my model of idiomatic activation-sets has highlighted the fact that semantic and structural distinctions between different idiomatic constructions
are far more subtle and fine-grained. In Section 4.6, I therefore proposed
a more detailed classification of alternative transparency-types reflected by
idioms. It would be interesting to see, if this cognitive-linguistic classification has clear-cut psycholinguistic correlates. It would, for instance, be
interesting to find out if qualitatively different imagery is evoked by literally
compositional in comparison with literally non-compositional idioms. In
other words, imagery experiments could be further fine-tuned with a view
to a more detailed classification of idiom-transparency. Moreover, it should
be scrutinised if different (re)motivation strategies are applied to the different subtypes. Are isomorphic idioms, for instance, primarily motivated
via their phrase-induced figurative senses rather than their literal scenes
as a whole? Do isomorphic idioms therefore demand different motivation
strategies in comparison with motivated but non-analysable idioms? And
to what extent do these differences influence idiom processing with a view
to a hybrid processing model?
3. The same desideratum can be put forward for the psycholinguistic analysis
of idiom variation. While a number of psycholinguistic models have been
proposed to approach the phenomenon of idiom variation (see Sections
2.2.4 and 2.2.5), these models have paid very little attention to alternative idiom-variation principles such as topic indication or literal-scene
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.8 (431-473)
 Idiomatic Creativity
manipulation (as well as their interactions). I have claimed that these principles differ with regard to how the idiomatic activation-set is manipulated
relative to the context-specific topic. But are these qualitative differences
also reflected in processing terms? Do topic indication and literal-scene
manipulation depend on different psycholinguistic processing routes? Do
relatively inconspicuous forms of idiom alteration take less processing effort than strikingly creative, ambiguous wordplays? These are questions
that have hardly been addressed by psycholinguistic research. However,
psycholinguistic investigation into this area could provide very important insight into the postulated heterogeneity of idiom processing. It could
break new paths towards an experimentally motivated classification of the
dynamically spreading activation-potentials through idiom activation-sets
in alternative usage-constellations. Corresponding psycholinguistic comprehension tests (e.g., eye-tracking experiments) could proceed from the
corpus-data that I have presented here.
4. With a view to idiom processing, it is further necessary to scrutinise the
precise interaction of conceptual and linguistic processing in idiom production and comprehension. I have claimed that all forms of idiom variation that appear on the lexicogrammatical surface are mediated by the
motivated (or unmotivated) conceptual association of the literal with the
idiomatic scene. Is it therefore necessary to postulate distinct linguistic and
conceptual subprocesses or stages of idiom manipulation or are conceptual
and linguistic parsing virtually inseparable? From a cognitive-linguistic
perspective, which conceives linguistic processing in connectionist terms
as parallel processing through intertwined conceptual and symbolic networks, the latter conception seems more realistic. With a view to variation
data, simultaneous or interactive idiom-processing models indeed appear
to be better equipped than modular step-by-step models to capture the
complex mental manipulation-patterns. This said, psycholinguistic testing
through priming or eye-tracking experiments or neuro-linguistic experimentation of brain-area activation could provide important evidence to
bring more light to this fairly obscure area of idiom processing, in particular, as well as linguistic processing, in general.
5. A more fine-grained psycholinguistic model of idiom production and
comprehension would also make it possible to substantiate the role of the
postulated idiom-variation constraints on variability. If idiom processing
depends on the simultaneous processing of conceptual and linguistic substructures this raises the question when, where and how the recognisability,
compatibility and grammaticality constraints affect the processing of sys-
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.9 (473-533)
Chapter 8. Conclusions and outlook 
tematic and unsystematic variants. To my knowledge, there is, for instance,
no psycholinguistic study that has tested the influence of the recognisability constraint on idiom comprehension. In Sections 6.5.1. and 7.6.,
I have proposed that the combined evocation of conceptual and lexical
clues supports the recognition of idiom variants. Moreover, I have claimed
that motivated idioms reflect stronger conceptual than lexical fixedness,
whereas opaque idioms are more dependent on their stable lexical realisation. To what extent do these alternative salient-idiomatic-substructures
support the comprehension of idiom variants? Can the suggested difference between conceptual and lexical backing with motivated and opaque
idioms be verified psycholinguistically? Again, psycholinguistic evidence
could be very helpful for refinements of the theoretical model.
6. A further psycholinguistic problem concerns the precise role of conceptual
metaphor in idiom processing. In Chapter 5, it was shown that conceptual
metaphors play a quintessential role for idiom motivation: they make it
possible to organise the internal semantic structure of idioms. But what is
the precise contribution of conceptual metaphors to idiom processing? In
Section 7.5, I have claimed that conceptual metaphors cannot be attributed
a prominently active role with instances of topic indication that disrupt
the literal-scene. In contrast, conceptual metaphors were claimed to be
of great influence for elaborate patterns of literal scene manipulation. A
qualitative psycholinguistic study on idiom interpretation strategies could
help to clarify this problem. Subjects could be asked to explain how they
make sense of a given metaphorically-motivated idiom and corresponding idiom-variants (e.g. topic-indicating and literal-scene manipulating
variants of jump on the bandwagon as discussed in Section 7.5). Alternatively, neuro-linguistic studies could reveal if subjects show different brainactivation patterns while processing such variants. A study on the precise
role of conceptual metaphors for idiom processing could thus bring further light to hybrid-processing models and clear-up the Glucksberg-Lakoff
controversy on idiom and metaphor processing.
7. Finally, this study has restricted its analytical focus to systematic lexicogrammatical alterations of idioms. However, both in spoken and written
language non-intentional erroneous alterations of the base-form can be
encountered. Word-finding difficulties in idiom-production constitute a
fascinating area of erroneous idiom use. Non-intentional and hoc substitutions such as wet one’s appetite for whet one’s appetite or dull as dishwater
for dull as ditchwater indicate that the formal structure of an idiom is
associated with a complex conceptual scene. In my view, such erroneous
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.10 (533-551)
 Idiomatic Creativity
lexical substitutions should therefore be explicable relative to the conceptual backgrounds that are associated with both the idiomatic and the literal
meaning. The systematic analysis of such erroneous idiom-variants could
therefore provide interesting insights for the cognitive-linguistic modelling
of idiom representation. Moreover, idiomatic slips-of-the-tongue and contaminations provide further valuable material to understand idiom production. For instance, contaminations such as shoot daggers at someone
(look daggers at someone + shoot a glance at someone) indicate that the
two quasi-synonymous idioms are closely associated constructions and
that they can therefore be co-activated as alternative symbolic units to
code a given target conceptualisation. In other words, the formal blend
points to external network relationships between different synonymous
idioms and suggests that both constructions can be directly stipulated in
response to the task of coding the target conceptualisation. The present
model of idioms as mental activation-sets does not specify external network relationships between different idioms (unless they are motivated by
the same conceptual backgrounds). To account for such systematic links
between alternative idioms, the architecture of the mental activation-set
model would have to be extended to include different types of external
network relationships (cf. Dobrovol’skij 1995). In my view, such an elaboration of my model would make to possible to account for idiomatic
slips-of-the-tongue and contaminations in a more systematic way. Moreover, idiomatic tongue-slips could further clarify the relationship between
linguistic and conceptual processing in idiom production. This said, it is a
great methodological challenge to collect a representative database of such
erroneous variants.
To conclude this book, I would like to refer back to the speculative metaphorical hypothesis with which I have introduced it. Idioms have been claimed to
have a Cheshire-cat-like nature. In the same way as Lewis Carroll’s comic figure, they can adopt alternative shapes when they appear as variants in actual
discourse without giving up their integrity as entrenched symbolic units. The
mental parameters that underlie these variable occurrences can be systematised by the help of a cognitive-linguistic approach to idiom representation and
variation. Nevertheless, these insights should be taken with a necessary pinch
of modesty. The heterogeneous and complex nature of the class of linguistic
phenomena defined as idioms still makes it difficult to grasp the discursive behaviour of these expressions in a predictable way. In this sense, idioms have so
far defied a fully waterproof scientific description. In other words, in the same
JB[v.20020404] Prn:24/02/2006; 13:21
F: HCP1708.tex / p.11 (551-555)
Chapter 8. Conclusions and outlook 
way as the Cheshire cat, they grin back at linguists and still leave them amazed
at the intricate complexity of natural language and its underlying cognitive
correlates.
JB[v.20020404] Prn:23/01/2006; 16:10
F: HCP17NO.tex / p.1 (48-131)
Notes
. Note that authors like Sinclair (1987, 1991), Pawley and Syder (1983) or Mel’čuk (1995,
1998) argue that phraseological constructions are predominant in linguistic production.
. The BNC is used as the main database for instances of idiom variation. All examples
are quoted by indicating the text-number and line-number in brackets, e.g. (K4T: 127).
Wherever passages in these quotes are highlighted in italics, these emphases are mine.
. A neccessary condition for this model to work is that the idiom is potentially ambiguous
between its literal and its idiomatic reading. Consequently, only idioms with a grammatical literal meaning can be handled by this framework. Obviously, the same is true for
Weinreich’s account.
. The idiom-list hypothesis is also in accordance with Searle’s (1993) pragmatic model
of metaphor and figurative-language comprehension. According to Searle’s model, it is the
recognition of a deviant literal meaning that triggers the search for a figurative meaning:
“Where the utterance is defective if taken literally, look for an utterance meaning that differs from sentence meaning” (Searle 1993: 103). This statement implies a two-step model of
metaphor comprehension. First, the literal sentence meaning of the metaphorical statement
has to be processed. When this meaning does not offer a contextually-appropriate utterance
meaning, the actual utterance meaning has to be inferred in a second step. In pragmatic theories, metaphors are thus reduced to deviant uses of language whose actual meaning has to
be creatively derived from the literal sentence meaning through an application of pragmatic
rules. This demands a great amount of inferential processing.
. It is important to emphasise that Gibbs has modified his views on idiom processing
considerably in his later studies.
. Note that Burger (1973) does not claim that all idioms are semantically unanalysable and
unmotivated. In the contrary, he proposes a sophisticated framework to describe the internal semantic structure of idioms (see Burger 1973: 16–17, Chapter 2). On the basis of this
framework, he differentiates alternative degrees of semantic motivation. However, Burger
does not assume that the different degrees of semantic analysability for different idioms can
be directly correlated with their syntactic behaviour. Rather, it is a fine-grained contrast “der
jedoch für das normale Funktioneren der Ketten vermutlich irrelevant ist” (Burger 1973: 17).
. This view is supported by Leverato’s psycholinguistic study on the development of figurative and idiomatic competence, see Levorato (1993: 107–113).
. It is important to emphasise that the idea of semantic analysability is not only promoted
by Nunberg et al. Using alternative terms such as ‘decomposition,’ ‘isomorphism’ or ‘homomorphism’ this concept has been described by various phraseologists: Burger (1973),
JB[v.20020404] Prn:23/01/2006; 16:10
F: HCP17NO.tex / p.2 (131-210)
 Idiomatic Creativity
Burger et al. (1982), Dobrovol’skij (1988, 1995, 1997), Geeraerts and Bakema (1993), Geeraerts (1995), Keil (1997), Koller (1977), Mel’čuk (1995), Nunberg (1978). I have introduced
the notion on the basis of Nunberg et al.’s study because it was most influential for other
views on idiom variability in the Anglo-American research context.
. Again, it is important to emphasise that the literal compositionality of idioms must be
clearly distinguished form top-down analysability based on figurative-literal isomorphism
(phrase-induced polysemy, semantic decomposition). Idioms that possess a compositional
literal level (e.g., kick the bucket) do not have to be analysable. Conversely, some analysable
idioms do not have to be literally compositional (e.g., blow the gaff).
. To avoid potential terminological confusion, it is necessary to emphasise that this interpretation of idiomatic compositionality cannot be read as a counterposition to the traditional conception of non-compositionality, which focuses on the bottom-up derivation of
the idiomatic meaning by adding the literal meanings of the constituents.
. Note that more limited versions of simultaneous processing models are also present in
Swinney and Cutler (1979); Estill and Kemper (1982).
. The notion of motivation is used by various authors (see, e.g., Burger 1973: 26; Burger
et al. 1982: 4; Dobrovol’skij 1995: 41–45; Barkema 1996: 140) to describe the possibility of
remotivating the idiomatic meaning of fully or partially non-compositional idiomatic constructions on the basis of their literal meaning. As Dobrovol’skij (1995: 41) points out, the
concept has a long tradition in the analysis of idioms and was already present in the work of
Vinogradov (1947), a pioneer in the field of phraseology.
. The etymological explanation for the metaphorical transfer of red herring to its contemporary meaning runs as follows: fugitives used to throw heavily spiced (and therefore red)
herrings behind their track to distract the bloodhounds from their trail.
. This criticism is motivated by a more general controversy in metaphor research.
This Lakoff-Glucksberg controversy is centred about the role of entrenched conceptual
metaphors for the comprehension of metaphorical expressions. For more details see Glucksberg et al. (1993), Lakoff (1993a: 236–7), Glucksberg and McGlone (1999).
. In traditional terms, this difference between internal domain structures and conceptual
hierarchies is captured by the distinction between meronymy and taxonymy (Cruse 1986:
Chapters 6 and 7).
. Note that Lakoff (1993a: 206) uses the less precise formula love is a journey to refer to
this metaphorical mapping. In my view, this formula is too general because the conceptual
metaphor refers to the development of the relationship of the loving couple rather than to
the emotion of ‘love’ as such.
. The terms “vehicle,” “tenor,” and “grounds” are used by Richards (1965). In cognitive
linguistics “source” and “target” are the corresponding, more frequently used concepts (see
Lakoff & Johnson 1980). “Topic” was introduced by Leech (1969) to substitute Richard’s
term “tenor.” Using this terminology, I follow Goatly (1997: 9).
. The Invariance Principle captures constraints on metaphorical correspondences (see
Lakoff 1990; Turner 1990, 1993; Lakoff 1993a: 216). The principle states that metaphorical
grounding does not necessarily have to be situated in conceptual knowledge, but that it is often related to lower-level cognitive structures: the ultimate realm for the recognition of simi-
JB[v.20020404] Prn:23/01/2006; 16:10
F: HCP17NO.tex / p.3 (210-275)
Notes 
larity is preconceptual image-schematic structure. The principle states that image-schematic
structure inherent in the target domain cannot be violated and limits the possibilities for
mappings automatically; if there is no further conceptual link, image-schemas motivate the
conceptual association between the source and the target. Moreover, metonymic grounding
occurs if two conceptual domains are commonly associated in concrete experience (see Taylor 1995: 138; Barcelona 2000b; Radden 2000). Metonymy–based metaphors such as more is
up or time is space often give the superficial impression of creating conceptual similarities.
. In this sense Turner and Fauconnier implicitly follow Black’s “interaction view of
metaphor” (see Black 1993: 19).
. Although blended metaphors question the unidirectionality hypothesis developed in
Lakoff and Johnson’s standard theory of metaphor, the recognition of conceptual blending
does not displace the notion of conceptual metaphor. Rather conceptual-blending theory and conceptual-metaphor theory focus on this process of semantic extension from
slightly different perspectives. While conceptual metaphors are general and systematic patterns of semantic extension across superordinate conceptual domains, conceptual blending
focuses on the online creation of complex metaphorical inferences (cf. Turner & Fauconnier 2000: 135). Such inferences are commonly accompanied by the construction of blends,
where informational structure from the source and target domains is coordinated.
. Since domains are organised in conceptual networks, and since such networks slightly
differ from person to person, domains do not have clear-cut boundaries. As a result, it is
not always possible to distinguish metaphor from metonymy. Various cognitive linguists
(Goossens 1990; Radden 2000; Ruiz de Mendoza Ibáñez 2000) therefore suggest to regard
the distinction between metaphor and metonymy as scalar, rather than absolute.
. By using this notion, Langacker places himself in a tradition of attempts that try to
explain the function of verb arguments and their grammatical relations by means of semantically based roles. The notion of ‘role archetype’ thus corresponds to the cognate concepts:
‘actants,’ ‘participants,’ ‘semantic roles,’ ‘theta-roles’ (cf. Ungerer & Schmid 1996: 173).
. Of course, the coding of conceptualisations by symbolic structures is often less straightforward than this model for prototypical transitive clauses implies. However, Langacker
(1991a: Chapter 8) proposes elegant solutions to deal with variational transitive coding as
well as other clause types.
. It is important to emphasise that Langacker, unlike the generative paradigm, does not
assume a clear-cut distinction between linguistic creativity and other forms of cognitive
creativity. Rather, linguistic creativity is the effect of general cognitive creativity applied to
linguistic units in the process of coding, see Langacker (1987: 71–72).
. Given the limitations of the present analysis, I must ignore the complex question of how
and to what degree the entrenchment of a given idiomatic configuration is distributed over
the members of a linguistic community including their sociolects and registers.
. Although not yet being homonymous, the extension may be on its way to become so.
New technological inventions in the computer mouse-frame are not conceptualised with
reference to the rodent frame. Thus, one speaks of a cordless mouse rather than a *tailless
mouse. In my view, such technological inventions (e.g. mice without cords, ergonomic mice,
JB[v.20020404] Prn:23/01/2006; 16:10
F: HCP17NO.tex / p.4 (275-352)
 Idiomatic Creativity
mice with scrolling keys, and the use of touchpads instead of mice) must be regarded as
important factors for the weakening of the metaphorical ground – the similar shape.
. The plural of mouse2 is mice rather than mouses. This indirect linguistic indicator speaks
for the preserved recognition of the metaphorical motivation underlying the use of mouse2 .
. The model simplifies the formal structure of idioms by including only two core lexical
items [A] and [B]. I have slightly adapted Geeraerts’s model to my purposes. For instance,
the syntagmatic dimension is reinterpreted in terms of Langacker’s notion of ‘integration’
(cf. Figure 3.6).
. As their diachronic development indicates, two of these constructions became opaque
due to the elimination of constituents: trip the light fantastic (toes), shoot (into) the breeze
(see OED).
. Note that apart from these processes of semantic extension, idioms may also incorporate
other forms of indirectness like irony (a pretty kettle of fish) or hyperbole (not be able to
organise a piss-up in a brewery, not set the Thames on fire). Since these forms of figuration are
more marginal, they are ignored in the present study (see Dietz 1999 for a comprehensive
overview).
. Idioms containing terms that describe body-parts are called “somatisms” or “somatic
idioms” (Somatismen); (see also Durco 1994: 65; Dobrovolskji 1988: 42).
. Idioms like act the fool/giddy goat, a busy bee or a wandering Jew contain stereotypical – and often politically incorrect, racist! – knowledge about animals, persons or other
cultures. Deviating from Dobrovol’skij and Piirainen (1997: 117), I regard such idiomatic
stereotypes as emblematic. Goats work as emblems for stupidity, bees for diligence and
Jews for restlessness.
. According to Weinreich, such expressions cannot even be regarded as “proper idioms,”
because they are not ambiguous between their literal and their idiomatic meaning. In his
view “ambiguity is an essential characterisitc of true idioms” (Weinreich 1969: 44).
. I would like to emphasise that the class of blended metaphorical idioms does not fully
correspond to the class of semantically ill-formed idioms. As will be shown below, semantic
ill-formedness can be caused by other factors such as the occurrence of unique constituents
(e.g. ‘kith’ and kin), or lexico-syntactic ill-formedness (e.g., trip the light fantastic).
. Burger et al. (1982: 29–30) distinguish four subtypes for this process: literalisation (“Literalisierung”), remotivation (“Remotivierung”), polysemantisation (“Polysemantisierung”),
and ‘take it literally’ (“wörtlich nehmen”). Since the details underlying these distinctions are
not of direct relevance for the present study, these processes are not scrutinised in further
detail here.
. As Sabban (1998: 68) points out, ambiguation must be regarded as a general rhetorical
strategy that is not only restricted to idiomatic constructions.
. Of course, the simple present is used after the subordinating conjunction when. In this
context the habitual interpretation of the simple present is irrelevant.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.1 (47-157)
References
Abeillé, A. (1995). The flexibility of French idioms: A representation with Lexicalized Tree
Adjoining Grammar. In M. Everaert, E. van der Linden, A. Schenk, & R. Schreuder
(Eds.), Idioms: Structural and psychological perspectives (pp. 15–42). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Allerton, D. J. (2002). Stretched verb constructions in English. London: Routledge.
Anders, H. (1995). Never say die: Englische Idiome um den Tod und das Sterben. Frankfurt
a.M.: Peter Lang.
Arnold, D. & L. Sadler (1989). Non-compositionality and translation. In J. Peckham
(Ed.), Recent developments and applications of natural language processing (pp. 23–55).
London: Kogan Page.
Aston, G. & L. Burnard (1998). The BNC handbook: Exploring the British national corpus
with SARA. Edinburgh: Edinburgh University Press.
Baranov, A. N. & D. Dobrovol’skij (1991). Kognitive Modellierung in der Phraseologie: Zum
Problem der aktuellen Bedeutung. Beiträge zur Erforschung der deutschen Sprache, 10,
112–123.
Baranov, A. N. & D. Dobrovol’skij (1996). Cognitive modeling of actual meaning in the field
of phraseology. Journal of Pragmatics, 25, 409–429.
Barcelona, A. (2000a). Introduction: The cognitive theory of metaphor and metonymy. In
A. Barcelona (Ed.), Metaphor and metonymy at the crossroads: A cognitive perspective
(pp. 1–28). Berlin: Mouton de Gruyter.
Barcelona, A. (2000b). On the plausibility of claiming a metonymic motivation for
conceptual metaphor. In A. Barcelona (Ed.), Metaphor and metonymy at the crossroads:
A cognitive perspective (pp. 31–58). Berlin: Mouton de Gruyter.
Barkema, H. (1996). Idiomaticity and terminology: A multi-dimensional descriptive model.
Studia Linguistica, 50(2), 125–160.
Barz, I. (1986). Probleme der phraseologischen Modifikation. DaF, 6, 321–326.
Bates, E., J. L. Elman, M. H. Johnson, A. Karmiloff-Smith, D. Parisi, & K. Plunkett (1998).
Innateness and emergentism. In W. Bechtel & G. Graham (Eds.), A companion to
cognitive science (pp. 590–601). Oxford: Blackwell.
Bauer, L. (1983). English word-formation. Cambridge: Cambridge University Press.
Bechtel, W. & A. Abrahamsen (1991). Connectionism and the mind: An introduction to
parallel processing in networks. Cambridge: Basil Blackwell.
Berlin, B., D. Breedlove, & P. Raven (1974). Principles of Tzeltal plant classification. New York:
Academic Press.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.2 (157-252)
 Idiomatic Creativity
Black, M. (1993). More about metaphor. In A. Ortony (Ed.), Metaphor and thought (pp.
19–41). Cambridge: Cambridge University Press.
Blakemore, D. (1992). Understanding utterances: An introduction to pragmatics. Oxford:
Blackwell.
Blasko, D. G. & C. M. Connine (1993). Effect of familiarity and aptness on the processing
of metaphor. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19,
295–308.
Bobrow, S. A. & S. M. Bell (1973). On catching on to idiomatic expressions. Memory and
Cognition, 1, 343–346.
Botelho da Silva, T. & A. Cutler (1993). Ill-formedness and transformability in Portuguese
idioms. In C. Cacciari & P. Tabossi (Eds.), Idioms: Processing, structure, and interpretation (pp. 129–143). Hillsdale, NJ: Lawrence Erlbaum Associates.
Burger, H. (1973). Idiomatik des Deutschen. Tübingen: Niemeyer.
Burger, H. (1989). ‘Bildhaft, übertragen, metaphorisch’: Zur Konfusion um die
semantischen Merkmale von Phraseologismen. In G. Gréciano (Ed.), EUROPHRAS
‘88. Phraséologie contrastive. Actes du colloque international Klingenthal-Strasbourg, 12–
16 mai 1988 (pp. 17–29). Strassbourg: USHS.
Burger, H. (1998a). Phraseologie: eine Einführung am Beispiel des Deutschen. Berlin: Erich
Schmidt Verlag.
Burger, H. (1998b). Idiom and metaphor: Their relation in theory and text. In P.
Ďurčo (Ed.), EUROPHRAS ‘97. Phraseology and paremiology (pp. 30–36). Bratislava:
Akadémia PZ.
Burger, H. (2000). Konzepte von “Variation” in der Phraseologie. In A. Häcki Buhofer (Ed.),
Vom Umgang mit sprachlicher Variation. Soziolinguistik, Dialektologie, Methoden und
Wissenschaftsgechichte (pp. 35–51). Tübingen: Francke Verlag.
Burger, H., A. Buhofer, & A. Sialm (1982). Handbuch der Phraseologie. Berlin: Walter de
Gruyter.
Cacciari, C. (1993). The place of idioms in a literal and metaphorical world. In C. Cacciari &
P. Tabossi (Eds.), Idioms: Processing, structure, and interpretation (pp. 27–55). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Cacciari, C. & P. Tabossi (1988). The comprehension of idioms. Journal of Memory and
Language, 27, 668–683.
Cacciari, C. & P. Tabossi (Eds.). (1993). Idioms: Processing, structure, and interpretation.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Cacciari, C. & S. Glucksberg (1991). Understanding idiomatic expressions: The contribution
of word meanings. In G. B. Simpson (Ed.), Understanding word and sentence (pp. 217–
240). Amsterdam: North-Holland.
Cacciari, C. & S. Glucksberg (1995). Imagining idiomatic expressions: Literal or figurative
meanings? In M. Everaert, E. van der Linden, A. Schenk, & R. Schreuder (Eds.), Idioms:
Structural and psychological perspectives (pp. 43–56). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Carroll, L. (1896/1993). Alice in wonderland and Through the looking-glass. Wordsworth
Classics.
Carston, R. (2002). Thoughts and utterances. The pragmatics of explicit communication.
Oxford: Blackwell.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.3 (252-370)
References 
Carter, R. (2004). Language and creativity: The art of common talk. London: Routledge.
Čermák, F. (1988). On the substance of idioms. Folia Linguistica, 22(3–4), 313–338.
Čermák, F. (1994). Idiomatics. In P. A. Luelsdorff (Ed.), Prague School of Structural and
Functional Linguistics (pp. 185–195). Amsterdam: John Benjamins.
Chafe, W. L. (1968). Idiomaticity as an anomaly in the Chosmkyan paradigm. Foundations
of Language, 4, 109–127.
Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: The MIT Press.
Chomsky, N. (1971). Review of Skinner. New York Review of Books, December 30, 1971.
Chomsky, N. (1975). Reflections on language. New York: Pantheon.
Chomsky, N. (1980). Rules and representations. New York, NY: Columbia University Press.
Coffey, S. (2001). Disturbing the form-meaning nucleus of multiword units: Data and issues.
Studi italiani di linguistica teorica e applicata, 2, 215–28.
Colombo, L. (1993). The comprehension of ambiguous idioms in context. In C. Cacciari
& P. Tabossi (Eds.), Idioms: Processing, structure, and interpretation (pp. 163–200).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Coulmas, F. (1979). On the sociolinguistic relevance of routine formulae. Journal of
Pragmatics, 3, 238–266.
Coulmas, F. (1981). Introduction: Conversational routine. In F. Coulmas (Ed.), Conversational routine: Explorations in standardized communication situations and prepatterned speech (pp. 1–17). The Hague: Mouton.
Cowie, A. P. (1998). Introduction. In A. P. Cowie (Ed.), Phraseology: Theory, analysis, and
applications (1–20). Oxford: Clarendon Press.
Cowie, A. P., R. Mackin, & I. R. McCaig (1993). Oxford Dictionary of English Idioms. Oxford:
Oxford University Press.
Croft, W. (1991). Syntactic categories and grammatical relations: The cognitive organisation of
information. Chicago, IL: University of Chicago Press.
Croft, W. (1993). The role of domains in the interpretation of metaphors and metonymies.
Cognitive Linguistics, 4, 335–370.
Croft, W. (1998). The structure of events and the structure of language. In M. Tomasello
(Ed.), The new psychology of language. Cognitive and functional approaches to language
structure (pp. 67–92). Mahwah, NJ: Lawrence Erlbaum Associates.
Croft, W. (2000). Explaining language change. An evolutionary approach [Longman
Linguistics Library]. London: Longman.
Croft, W. (2001). Radical construction grammar: Syntactic theory in typological perspective.
Oxford: Oxford University Press.
Cruse, D. A. (1986). Lexical semantics. Cambridge: Cambridge University Press.
Cutler, A. (1982). Idioms: The colder the older. Linguistic Inquiry, 13, 317–320.
Davidson, J. (1991). Natural creation and the formative mind. Dorset: Shaftesbury.
Dietz, H. (1999). Rhetorik in der Phraseologie. Zur Bedeutung rhetorischer Stilelemente im
idiomatischen Wortschatz des Deutschen. Tübingen: Niemeyer.
Dobrovol’skji, D. (1988). Phraseologie als Objekt der Universalienlinguistik. Leipzig: VEB
Verlag Enzyklopädie.
Dobrovol’skji, D. (1995). Kognitive Aspekte der Idiom Semantik: Studien zum Thesaurus
deutscher Idiome. Tübingen: Gunter Narr.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.4 (370-476)
 Idiomatic Creativity
Dobrovol’skji, D. (1997). Idiome im mentalen Lexikon: Ziele und Methoden der kognitivbasierten Phraseologieforschung. Trier: Wissenschaftlicher Verlag Trier.
Dobrovol’skji, D. (1999a). Zu semantischen und pragmatischen Effekten kreativer IdiomModifikationen. Nouveaux cahiers d’allemand, 17, 363–374.
Dobrovol’skji, D. (1999b). Haben transformationelle Defekte der Idiomstruktur
semantische Ursachen? In B. Fernàndez, N. I. Behr, & C. Rozier (Eds.), Phraseme und
typisierte Rede (pp. 25–38). Tübingen: Stauffenburg Verlag.
Dobrovol’skji, D. (2000). Ist die Semantik von Idiomen nichtkompositionell? In S.
Beckmann, P. P. König, & G. Wolf (Eds.), Sprachspiel und Bedeutung. Festschrift für
Franz Hundsnurscher zum 65. Geburtstag (pp. 113–124). Tübingen: Niemeyer.
Dobrovol’skji, D. (2004). Semantische Teilbarkeit der Idiomstruktur: zu operationalen
Kriterien. In Ch. Palm Meister (Ed.), EUROPHRAS 2000 (pp. 61–68). Tübingen:
Stauffenburg Verlag.
Dobrovol’skij, D. & E. Piirainen (1997). Symbole in Sprache und Kultur: Studien zur
Phraseologie aus kultursemiotischer Perspektive. Bochum: Brockmeyer.
Ďurčo, P. (1994). Probleme der allgemeinen und kontrastiven Phraseologie. Am Beispiel
Deutsch und Slowakisch. Heidelberg: Julius Groos Verlag.
Elman, J. (1998). Connectionism, artificial life, and dynamical systems. In W. Bechtel &
G. Graham (Eds.), A companion to cognitive science (pp. 488–505). Oxford: Blackwell.
Ernst, Th. (1981). Grist for the linguistic mill: idioms and ‘extra’ adjectives. Journal of
Linguistic Research, 1(3), 51–68.
Estill, R. B. & S. Kemper (1982). Interpreting idioms. Journal of Psycholinguistic Research, 11,
559–568.
Fauconnier, G. (1994). Mental spaces: Aspects of meaning construction in natural language.
Cambridge: Cambridge University Press.
Fauconnier, G. (1997). Mappings in thought and language. Cambridge: Cambridge
University Press.
Fauconnier, G. & M. Turner (1998). Conceptual integration networks. Cognitive Science,
22(2), 133–187.
Fellbaum, Ch. (1993). The determiner in English idioms. In C. Cacciari & P. Tabossi (Eds.),
Idioms: Processing, structure, and interpretation (pp. 271–295). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Fernando, C. (1996). Idioms and idiomaticity. Oxford: Oxford University Press.
Fernando, C. & R. Flavell (1981). On idiom: Critical views and perspectives [Exeter Linguistic
Studies, 5]. Exeter: University of Exeter Press.
Ferris, C. (1992). The meaning of syntax. A study in the adjectives of English. Harlow, Essex:
Longman.
Feyaerts, K. (1999). Metonymic hierarchies: The conceptualization of stupidity in German
idiomatic expressions. In K.-U. Panther & G. Radden (Eds.), Metonymy in language and
thought [Human Cognitive Processing, 4] (pp. 309–332). Amsterdam: John Benjamins.
Fillmore, Ch. J. (1975). An alternative to checklist theories of meaning. In H. Cogen, G.
Thurgood, & K. Whistler (Eds.), Proceedings of the Berkeley linguistic society (pp. 123–
131). Berkeley, CA: Berkeley Linguistics Society.
Fillmore, Ch. J. (1985). Frames and the semantics of understanding. Quaderni di Semantica,
VI(2), 222–254.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.5 (476-581)
References 
Fillmore, Ch. J. & B. T. Atkins (1992). Toward a frame-based lexicon: The semantics of RISK
and its neighbors. In A. Lehrer & E. Kittay (Eds.), Frames, fields, and contrasts (pp.
75–102). Hillsdale, NJ: Lawrence Erlbaum.
Fillmore, Ch. J., P. Kay, & M. C. O’Connor (1988). Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64(3), 501–538.
Fleischer, W. (1982). Phraseologie der deutschen Gegenwartssprache. Leipzig: VEB Verlag
Enzyklopädie.
Flores d’ Arcais, G. B. (1993). The comprehension and semantic interpretation of idioms.
In C. Cacciari & P. Tabossi (Eds.), Idioms: Processing, structure, and interpretation (pp.
79–99). Hillsdale, NJ: Lawrence Erlbaum Associates.
Fraser, B. (1970). Idioms within a transformational grammar. Foundations of Language, 6,
22–42.
Gazdar, G., E. Klein, G. K. Pullum, & I. Sag (1985). Generalized phrase structure grammar.
Oxford: Blackwell.
Geeraerts, D. (1995). Specialization and reinterpretation in idioms. In M. Everaert, E. van
der Linden, A. Schenk, & R. Schreuder (Eds.), Idioms: Structural and psychological
perspectives (pp. 57–73). Hillsdale, NJ: Lawrence Erlbaum Associates.
Geeraerts, D. & P. Bakema (1993). De prismatische semantiek van idiomen en composita.
Leuvense Bijdragen, 82, 185–226.
Gernsbacher, M. A. & R. W. Robertson (1999). The role of suppression in figurative language
comprehension. Journal of Pragmatics, 31, 1619–1630.
Gibbs, R. W. (1980). Spilling the beans on understanding and memory for idioms in
conversation. Discourse Processes, 9, 17–30.
Gibbs, R. W. (1985). On the process of understanding idioms. Journal of Psycholinguistic
Research, 14(5), 465–472.
Gibbs, R. W. (1986). Skating on thin ice: Literal meaning and understanding idioms in
conversation. Memory and Cognition, 8, 149–156.
Gibbs, R. W. (1989). Understanding and literal meaning. Cognitive Science, 13, 243–251.
Gibbs, R. W. (1990). Psycholinguistic studies on the conceptual basis of idiomaticity.
Gibbs, R. W. (1992). What do idioms really mean. Journal of Memory and Language, 31,
485–506.
Gibbs, R. W. (1993). Why idioms are not dead metaphors. In C. Cacciari & P. Tabossi (Eds.),
Idioms: Processing, structure, and interpretation (pp. 57–77). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Gibbs, R. W. (1994). The poetics of mind. Cambridge: Cambridge University Press.
Gibbs, R. W. (1995). Idiomaticity and human cognition. In M. Everaert, E. van der Linden,
A. Schenk, & R. Schreuder (Eds.), Idioms: Structural and psychological perspectives (pp.
Gibbs, R. W. & G. P. Gonzales (1985). Syntactic frozenness in processing and remembering
idioms. Cognition, 20, 243–259.
Gibbs, R. W., N. Nayak, J. Bolton, & M. Keppel (1989). Speakers’ assumptions about the
lexical flexibility of idioms. Memory and Cognition 17, 58–68.
Gibbs, R. W., N. Nayak, & C. Cutting (1989). How to kick the bucket and not decompose:
Analyzability and idiom processing. Journal of Memory and Language, 28, 576–593.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.6 (581-685)
 Idiomatic Creativity
Gibbs, R. W. & N. Nayak (1989). Psycholinguistic studies on the syntactic behaviour of
idioms. Cognitive Psychology, 21, 100–138.
Gibbs, R. W. & J. O’Brien (1990). Idioms and mental imagery. Cognition, 36(1), 35–68.
Gibbs, R. W. & H. L. Colston (1995). The cognitive psychological reality of image schemas
and their transformations. Cognitive Linguistics, 6, 347–378.
Giora, R. & O. Fein (1999). On understanding familiar and less-familiar figurative language.
Journal of Pragmatics, 31, 1601–1618.
Glass, A. L. (1983). The comprehension of idioms. Journal of Psycholinguistic Research, 12,
429–442.
Gläser, R. (1986). Phraseologie der englischen Sprache. Tübingen: Niemeyer.
Gläser, R. (1988). The grading of idiomaticity as a presupposition for a taxonomy of idioms.
In W. Hüllen & R. Schulze (Eds.), Understanding the lexicon (pp. 264–279). Tübingen:
Max Niemeyer.
Gläser, R. (1998). The stylistic potential of phraseological units in the light of genre analysis.
In A. P. Cowie (Ed.), Phraseology: Theory, analysis, and applications (pp. 125–143).
Oxford: Clarendon Press.
Glucksberg, S. (1993). Idiom meanings and allusional content. In C. Cacciari & P. Tabossi
(Eds.), Idioms: Processing, structure, and interpretation (pp. 3–26). Hillsdale, NJ:
Glucksberg, S. (2001). Understanding figurative language. New York, NY: Oxford University
Press.
Glucksberg, S., M. Brown, & M. S. McGlone (1993). Conceptual metaphors are not
automatically accessed during idiom comprehension. Memory and Cognition, 21, 711–
719.
Glucksberg, S. & B. Keysar (1993). How metaphors work. In A. Ortony (Ed.), Metaphor and
thought (pp. 401–424). Cambridge: Cambridge University Press.
Glucksberg, S. & M. S. McGlone (1999). When love is not a journey: What metaphors mean.
Journal of Pragmatics, 31, 1541–1558.
Goatly, A. (1997). The language of metaphors. London: Routledge.
Goldberg, A. (1995). Constructions: A construction grammar approach to argument structure.
Chicago, IL: Chicago University Press.
Goossens, L. (1990). Metaphtonymy: The interaction of metaphor and metonymy in
expressions for linguistic action. Cognitive Linguistics, 1, 3223–3240.
Gréciano, G. (1987). Idiom und sprachspielerische Textkonstitution. In J. Korhonen (Ed.),
Beiträge zur allgemeinen und germanistischen Phraseologieforschung. Internationales
Symposium in Oulu. 13.–15. Juni 1986 (pp. 193–206). Oulu, University.
Häcki Buhofer, A. (1989). Psycholinguistische Aspekte in der Bildhaftigkeit von
Phraseologismen. In G. Gréciano (Ed.), EUROPHRAS ‘88. Phraséologie contrastive.
Actes du colloque international Klingenthal-Strasbourg, 12–16 mai 1988 (pp. 165–175).
Strassbourg: USHS.
Häcki Buhofer, A. (1996). Theorie und Methodologie einer psycholinguistisch basierten
Bedeutungsbeschreibung: Zur Bandbreite phraseologischer Bedeutungen. In E.
Weigand & F. Hundsnurscher (Eds.), Lexical structures and language use. Proceedings
of the international conference on lexicology and lexical semantics, Münster, September
13–15, 1994 (pp. 387–398). Tübingen: Niemeyer.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.7 (685-800)
References 
Häcki Buhofer, A. (1999). Psycholinguistik der Phraseologie. In N. Fernández Bravo, I. Behr,
& C. Rozier (Eds.), Phraseme und typisierte Rede (pp. 63–76). Tübingen: Stauffenburg
Verlag.
Halliday, M.A.K. (1978). Language as social semiotic. London: Edward Arnold.
Halliday, M.A.K. (1985). An introduction to functional grammar. London: Edward Arnold.
Harras, G. (1983). Handlungssprache und Sprechhandlung. Berlin: Mouton de Gruyter.
Hays, P. R. (2000) From the Jurassic dark: Linguistic relativity as evolutionary necessity. In
M. Pütz & M. H. Verspoor (Eds.), Explorations in linguistic relativity (pp. 159–171).
Amsterdam: John Benjamins.
Hegedüs-Lambert, C. (1995). Phraséologie contrastive: La saisie des cinquante formatifs les
plus productifs en allemand et en français. PhD dissertation, Strassbourg.
Hemmi, A. (1994). ‘Es muss wirksam werben, wer nicht will verderben’: Kontrastive Analyse
von Phraseologismen in Anzeigen-, Radio- und Fernsehwerbung. Bern: Peter Lang.
Hessky, R. (1995). Zum kognitiven Ansatz in der Phraseologie: ‘Aufgewärmter Kohl’
oder ‘eine neue Platte’?. In G. Harras (Ed.), Die Ordnung der Wörter: Kognitive und
lexikalische Strukturen (pp. 289–302). Berlin: Walter de Gruyter.
Hundt, M. (1995). Modellbildung in der Wirtschaftssprache: zur Geschichte der Institutionen- und Theoriefachsprachen der Wirtschaft. Tübingen: Niemeyer.
Indurkhya, B. (1992). Metaphor and cognition. Dordrecht: Kluwer Academic Publishers.
Johnson, M. (1987). The body in the mind. The bodily basis of reason and imagination.
Chicago, IL: University of Chicago Press.
Johnson-Laird, P. N. (1983). Mental models. Towards a cognitive science of language, inference,
and consciousness. Cambridge: Cambridge University Press.
Katz, J. J. (1973). Compositionality, idiomaticity, and lexical substitution. In S. Anderson
& P. Kiparsky (Eds.), A Festschrift for Morris Halle (pp. 357–376). New York: Holt,
Rinehart, and Winston.
Katz, J. J. & J. A. Fodor (1963). The structure of a semantic theory. Language, 39, 170–210.
Katz, J. J. & P. M. Postal (1963). Semantic interpretation of idioms and sentences containing
them. Quarterly Progress Report, 70, MIT Research Laboratory of Electronics, 275–282.
Kay, P. & Ch. J. Fillmore (1999). Grammatical constructions and linguistic generalizations:
The What’s X doing Y? construction. Language, 75(1), 1–33.
Kemmer, S. & M. Barlow (2000). Introduction: A usage-based conception of language.
In M. Barlow & S. Kemmer (Eds.), Usage-based models of language (pp. vii–xxviii).
Stanford, CA: CSLI Publications.
Keil, M. (1997). Wort für Wort: Repräsentation und Verarbeitung verbaler Phraseologimen
(Phraseo-Lex). Tübingen: Max Niemeyer Verlag.
Keysar, B. & B. M. Bly (1999). Swimming against the current: Do idioms reflect conceptual
structure? Journal of Pragmatics, 31, 1559–1576.
Kövecses, Z. & P. Szabó (1996). Idioms: A view from cognitive semantics. Applied Linguistics,
17(3), 326–355.
Koller, W. (1977). Redensarten, Linguistische Aspekte. Vorkommensanalysen, Sprachspiel.
Tübingen: Niemeyer.
Korhonen, J. (Ed.). (1992). Phraseologie und Wortbild – Aspekte der Lexikonerweiterung.
Finnisch-deutsche sprachwissenschaftliche Konferenz, 5.–6. Dezember 1990 in Berlin.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.8 (800-901)
 Idiomatic Creativity
Kroenenfeld, D. B. (2000). Language and thought – collective tools for individual use. In
M. Pütz & M. H. Verspoor (Eds.), Explorations in linguistic relativity (pp. 197–223).
Kuiper, K. & M. Everaert (2004). Limits on the phrase structural configurations of phrasal
lexical items. In Ch. Palm Meister (Ed.), EUROPHRAS 2000 (pp. 257–262). Tübingen:
Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind.
Lakoff, G. (1990) The invariance hypothesis: Is abstract reason based on image-schemas?
Lakoff, G. (1993a). The contemporary theory of metaphor. In A. Ortony (Ed.), Metaphor
and thought (pp. 202–251). Cambridge: Cambridge University Press.
Lakoff, G. (1993b). The syntax of metaphorical semantic roles. In J. Pustejovsky (Ed.),
Semantics and the lexicon [Studies in Linguistics and Philosophy, 49] (pp. 27–36).
Dordrecht: Kluwer Academic Publishers.
Lakoff, G. & M. Johnson (1980). Metaphors we live by. Chicago, IL: University of Chicago
Press.
Lakoff, G. & M. Johnson (1999). Philosophy in the flesh: The embodied mind and its challenge
to western thought. New York, NY: Basic Books.
Lakoff, G. & M. Turner (1989). More than cool reason: A field guide to poetic metaphor.
Lamb, S. M. (1999). Pathways of the brain: The neurocognitive basis of language. Amsterdam:
John Benjamins.
Lamb, S. M. (2000). Neuro-cognitive structure in the interplay of language and thought.
In M. Pütz & M. H. Verspoor (Eds.), Explorations in linguistic relativity (pp. 173–196).
Langacker, R. W. (1987). Foundations of cognitive grammar. Volume I: Theoretical
prerequisites. Stanford, CA: Stanford University Press.
Langacker, R. W. (1991a). Foundations of cognitive grammar. Volume II: Descriptive
application. Stanford, CA: Stanford University Press.
Langacker, R. W. (1991b). Concept, image, and symbol. The cognitive basis of grammar. Berlin:
Mouton de Gruyter.
Langacker, R. W. (1993). Reference-point constructions. Cognitive Linguistics, 4, 1–38.
Langacker, R. W. (1997). The contextual basis of cognitive semantics. In J. Nuyts
& E. Pedersen (Eds.), Language and conceptualization (pp. 229–252). Cambridge:
Cambridge University Press.
Langacker, R. W. (1998). Conceptualization, symbolization, and grammar. In M. Tomasello
(Ed.), The new psychology of language. Cognitive and functional approaches to language
structure (pp. 1–39). Mahwah, NJ: Lawrence Erlbaum Associates.
Langacker, R. W. (2000a). A dynamic usage-based model. In M. Barlow & S. Kemmer (Eds.),
Usage-based models of language (pp. 1–64). Stanford, CA: CSLI Publications.
Langacker, R. W. (2000b). Why a mind is necessary: Conceptualization, grammar, and
linguistic semantics. In L. Albertazzi (Ed.), Meaning and cognition: A multidisciplinary
approach (pp. 25–38). Amsterdam: John Benjamins.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.9 (901-1007)
References
Langlotz, A. (1998). A Cognitive Semantic approach to contrastive phraseology – The
motivation of idioms by conceptual metaphors: A contrastive analysis of English and
German idioms denoting mental fitness and weakness. Basel: Lizentiatsarbeit.
Langlotz, A. (2001a). Ambiguity and the creative use of idioms in commercial print
advertisements. In A. Hamm & P. Buccellato (Eds.), Linguistic ambiguity [RANAM 34]
(pp. 111–127). Strasbourg: Université Marc Bloch, Services des périodiques.
Langlotz, A. (2001b). Cognitive principles of idiom variation – idioms as complex linguistic
categories. Studi Italiani di Linguistica Teorica e Applicata, 2, 289–302.
Langlotz, A. (2004a). Conceptual blending as a principle of idiom structure and idiom
variation. In Ch. Palm Meister (Ed.), EUROPHRAS 2000 (pp. 263–272). Tübingen:
Langlotz, A. (2004b). What are metaphors? In D. J. Allerton, N. Nesselhauf, & P. Skandera
(Eds.), Phraseological units: Basic concepts and their application [ICSELL, 8] (pp. 37–51).
Basel: Schwabe.
Langlotz, A. (2005). Are constructions the basic units of grammar? – The phraseological
turn in linguistic theory. In A. Hamm (Ed.), Language chunks and linguistic units
[RANAM 38] (pp. 45–63). Strasbourg: Université Marc Bloch, Services des périodiques.
Langlotz, A. (forthcoming). A cognitive linguistic analysis of lexical substitution in idioms.
In Proceedings of EUROPHRAS 2004.
Langlotz, A. (under review). Occasional adnominal idiom modification – a cognitive
linguistic approach. IJES (International Journal of English Studies).
Leech, G. N. (1969). A linguistic guide to English poetry. Harlow: Longman.
Levelt, W. J. M. (1999). Producing spoken language: A blueprint of the speaker. In C. M.
Brown & P. Hagoort (Eds.), The neurocognition of language (pp. 83–122). Oxford:
Oxford University Press.
Levorato, M. C. (1993). The acquisition of idioms and the development of figurative
competence. In C. Cacciari & P. Tabossi (Eds.), Idioms: Processing, structure, and
interpretation (pp. 101–128). Hillsdale, NJ: Lawrence Erlbaum Associates.
Makkai, A. (1972). Idiom structure in English. The Hague: Mouton.
McArthur, T. (1992). The Oxford companion to the English language. Oxford: Oxford
University Press.
McGlone, M. S., S. Glucksberg, & C. Cacciari (1994). Semantic productivity and idiom
comprehension. Discourse Processes, 17, 167–190.
Mel’čuk, I. (1995). Phrasemes in language and phraseology in linguistics. In M. Everaert, E.
van der Linden, A. Schenk, & R. Schreuder (Eds.), Idioms: Structural and psychological
Mel’čuk, I. (1998). Collocations and lexical functions. In A. P. Cowie (Ed.), Phraseology:
Theory, analysis, and applications (pp. 23–53). Oxford: Clarendon Press.
Mena Martínez, F. M. (2002). La desautomizatión de las paremias inglesas por sustitución:
Un estudio cognitivo. PhD dissertation, University of Murcia, Spain.
Minugh, D. (1999). “You people use such weird expressions”: The frequency of idioms in
newspaper CDs as corpora. In J. M. Kirk (Ed.), Corpora galore: Analyses and techniques
in describing English. Amsterdam: Rodopi.

JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.10 (1007-1110)
 Idiomatic Creativity
Minugh, D. (2001). “You believe that, and I’ll throw in the Golden Gate for free”: Using
corpora in search for the mother lode of idioms. Studi Italiani di Linguistica Teorica e
Applicata, 2, 243–256.
Moon, R. (1998). Fixed expressions and idioms in English. A corpus-based approach. Oxford:
Clarendon Press.
Moon, R. (2001). The distribution of idioms in English. Studi Italiani di Linguistica Teorica
e Applicata, 2, 229–242.
Napoli, D. (1988). Subjects and external arguments, clauses and non-clauses. Linguistics and
Philosophy, 11, 323–345.
Nayak, N. P. & R. W. Gibbs (1990). Conceptual knowledge in the interpretation of idioms.
Journal of Experimental Psychology: General, 119, 315–330.
Newmeyer, F. J. (1974). The regularity of idiom behavior. Lingua, 34, 327–342.
Nicolas, T. (1995). Semantics of idiom modification. In M. Everaert, E. van der Linden,
Nordén, M. (1994). Logische Beziehungsnetze und Inferenzprozeduren. Zu einer semantischkognitiven Theorie der verbalen Idiome im Deutschen. Stockholm: AlmqvistandWiksell.
Nunberg, G. (1978). The pragmatics of reference. Bloomington, IN: Indiana University
Linguistics.
Nunberg, G., I. A. Sag, & Th. Wasow (1994). Idioms. Language, 70(3), 491–538.
O’Grady, W., M. Dobrovolsky, & M. Aronoff (1997). Contemporary Linguistics: An
introduction. Boston: Bedford/St.Martin’s.
Ortony, A., D. L. Schallert, R. E. Reynolds & S. J. Antos (1978). Interpreting metaphors and
idioms: Some effects of the context on comprehension. Journal of Verbal Learning and
Verbal Behaviour, 17, 465–477.
Orwell, G. (1946). Politics and the English language. Reprinted 1962 in Inside the Whale and
Other Essays. Harmondsworth: Penguin.
Palmer, F. R. (1981). Semantics. Cambridge: Cambridge University Press.
Pawley, A. & F. H. Syder (1983). Two puzzles for linguistic theory: Nativelike selection
and nativelike fluency. In J. C. Richards & R. W. Schmidt (Eds.), Language and
communication (pp. 191–225). London: Longman.
Pilkington, A. (2000). Poetic effects: A relevance theory perspective. Amsterdam: John
Benjamins.
Pilz, K. D. (1978). Phraseologie: Versuch einer interdisziplinären Abgrenzung, Begriffsbestimmung und Systematisierung unter besonderer Berücksichtigung der deutschen Gegenwartssprache, 2 vols. Göppingen: Kümmerle.
Pilz, K. D. (1981). Phraseologie, Redensartforschung. Stuttgart: Metzler.
Quirk, R., S. Greenbaum, G. Leech, & Jan Svartvik (1985). A comprehensive grammar of the
English language. London: Longman.
Radden, G. (2000). How metonymic are metaphors? In A. Barcelona (Ed.), Metaphor and
metonymy at the crossroads: A cognitive perspective (pp. 93–108). Berlin: Mouton de
Gruyter.
Radford, A. (1997). Syntax: A minimalist introduction. Cambridge: Cambridge University
Press.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.11 (1110-1219)
References
Reddy, M. J. (1993). The conduit metaphor: A case of frame conflict in our language
about language. In A. Ortony (Ed.), Metaphor and thought (pp. 164–201). Cambridge:
Cambridge University Press.
Richards, I. A. (1965). The philosophy of rhetoric. Oxford: Oxford University Press.
Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition
and categorization (pp. 27–48). Hillsdale, NJ: Lawrence Erlbaum Associates.
Rosch, E., C. B. Mervis, W. D. Gray, D. M. Johnson, & P. Boyes-Braem (1976). Basic objects
in natural categories. Cognitive Psychology, 8, 382–439.
Rothkegel, A. (1973). Feste Syntagmen. Grundlagen, Strukturbeschreibung und automatische
Analyse. Tübingen: Max Niemeyer.
Ruiz de Mendoza Ibáñez, F. J. (2000). The role of mappings and domains in understanding
metonymy. In A. Barcelona (Ed.), Metaphor and metonymy at the crossroads: A cognitive
perspective (pp. 109–132). Berlin: Mouton de Gruyter.
Rumelhart, D. E. (1993). Some problems with the notion of literal meanings. In A. Ortony
(Ed.), Metaphor and thought (pp. 71–82). Cambridge: Cambridge University Press.
Sabban, A. (1998). Okkasionelle Variationen sprachlicher Schematismen: eine Analyse
französischer und deutscher Presse- und Werbetexte. Tübingen: Narr.
Salminen, O. (1993). Aspekte der Sprachkreativität. Zu Grundlagen einer Theorie der
sprachlichen Performanz. Helsinki: Suomalainen Tiedeakatemia.
Schenk, A. (1995). The syntactic behaviour of idioms. In M. Everaert, E. van der Linden,
Schank, R. C. & R. P. Abelson (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ:
Schwarz, M. (1992). Kognitive Semantiktheorie und neuropsychologische Realität: Repräsentationale und prozedurale Aspekte der semantischen Kompetenz. Tübingen: Niemeyer.
Schwarz, M. (1996). Einführung in die Kognitive Linguistik. Tübingen: Francke Verlag.
Schweigert, W. A. (1992). The muddy water of idiom comprehension begin to settle. In M.
Everaert, E. van der Linden, A. Schenk, & R. Schreuder (Eds.), Proceedings of idioms
(pp. 111–118). Tilburg: Katolieke Universiteit Brabant.
Searle, J. (1993). Metaphor. In A. Ortony (Ed.), Metaphor and thought (pp. 83–111).
Cambridge: Cambridge University Press.
Shannon, C. E. & W. Weaver (1949). The mathematical theory of communication. Urbana:
University of Illinois Press.
Sinclair, J. (1987). Collocation: A progress report. In R. Steele & T. Threadgold (Eds.),
Language topics: An international collection of papers by colleagues, students and admirers
of Professor Michael Halliday to honour him on his retirement, 2 (pp. 319–331).
Sinclair, J. (1991). Corpus, concordance, collocation. Oxford: Oxford University Press.
Skandera, P. (2004). What are idioms. In D. J. Allerton, N. Nesselhauf, & P. Skandera (Eds.),
Phraseological units: Basic concepts and their application [ICSELL, 8] (pp. 22–36). Basel:
Schwabe.
Smith, N. (1999). Chomsky: Ideas and ideals. Cambridge: Cambridge University Press.
Sperber, D. & D. Wilson (1986). Relevance: Communication and cognition. Oxford: Blackwell.

JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.12 (1219-1327)
 Idiomatic Creativity
Sperber, D. & D. Wilson (1995). Relevance: Communication and cognition (2nd edition).
Oxford: Blackwell.
Stein, G. (1979). Studies in the function of the passive. Tübingen: Narr.
Strässler, J. (1982). Idioms in English: A pragmatic analysis. Tübingen: Gunter Narr Verlag.
Swinney, D. A. & A. Cutler (1979). The access and processing of idiomatic expressions.
Journal of Verbal Learning and Verbal Behaviour, 18, 523–534.
Tabossi, P. & F. Zardon (1993). The activation of idiomatic meaning in spoken language
comprehension. In C. Cacciari & P. Tabossi (Eds.), Idioms: Processing, structure, and
interpretation (pp. 145–162). Hillsdale, NJ: Lawrence Erlbaum Associates.
Tabossi, P. & F. Zardon (1995). The activation of idiomatic meaning. In M. Everaert, E.
van der Linden, A. Schenk, & R. Schreuder (Eds.), Idioms: Structural and psychological
Talmy, L. (1988). Force dynamics in language and cognition. Cognitive Science, 12, 49–100.
Talmy, L. (2000). Toward a cognitive semantics, 2 vols. Cambridge: MIT Press.
Tannen, D. (1989). Talking voices: Repetition, dialogue and imagery in conversational discourse. Cambridge: Cambridge University Press.
Taylor, J. R. (1995). Linguistic categorization. Prototypes in linguistic theory. Oxford:
Clarendon Press.
Thun, H. (1978). Probleme der Phraseologie. Untersuchungen zur wiederholten Rede
mit Beispielen aus dem Französischen, Italienischen, Spanischen und Rumänischen.
Titone, D. A. & C. M. Connine (1999). On the compositional and noncompositional nature
of idiomatic expressions. Journal of Pragmatics, 31, 1655–1674.
Tucker, G. (2001). ‘Getting our heads around it’: Semantic and syntactic tension in the
transitivity analysis of metaphorically derived multi-word verbs. Studi Italiani di
Linguistica Teorica e Applicata, 2, 303–314.
Turner, M. (1990). Aspects of the invariance hypothesis (commentary). Cognitive Linguistics,
1, 247–255.
Turner, M. (1993). An image-schematic constraint on metaphor. In R. A. Geiger & B.
Rudzka-Ostyn (Eds.), Conceptualizations and mental processing in language (pp. 292–
306). Berlin: Mouton de Gruyter.
Turner, M. & G. Fauconnier (2000). Metaphor, metonymy, and binding. In A. Barcelona
(Ed.), Metaphor and metonymy at the crossroads: A cognitive perspective (pp. 133–145).
Berlin: Mouton de Gruyter.
Ungerer, F. & H. Schmid (1996). An Introduction to cognitive linguistics. London: Longman.
Van de Voort, M. & W. Vonk (1995). You don’t die immediately when you kick and empty
bucket: a processing view on semantic and syntactic characteristics of idioms. In M.
Everaert, E. van der Linden, A. Schenk, & R. Schreuder (Eds.), Idioms: Structural and
psychological perspectives (pp. 284–299). Hillsdale, NJ: Lawrence Erlbaum Associates.
Veisbergs, A. (2001). Occasional transformations of idioms: Wordplay in interpreting. Studi
Italiani di Linguistica Teorica e Applicata, 2, 257–272.
Verhagen, M. (1990). Support verbs in Disneyland. Eurotra-Essex Internal Memoranda No.
20. Wivenhoe: Essex University Linguistics Department.
JB[v.20020404] Prn:20/02/2006; 15:22
F: HCP17RE.tex / p.13 (1327-1377)
References
Vinogradov, V. V. (1947). Ob osnovnych tipach frazeologičeskich edinic v sovremenonnom
russkom jazyke. In V. V. Vinogradov, 1977, Izbrannye trudy. Leksikologija i leksikografija.
Moskva: Nauka.
Warrren, B. (1984). Classifying adjectives. Göteborg: Acta Universitatis Gothoburgensis.
Wasow, T., I. Sag, & G. Nunberg (1983). Idioms: An interim report. In S. Hattori & K. Inoue
(Eds.), Proceeding of the XIIIth international congress of linguistics (pp. 102–115). Tokyo:
Comité International Permanent des Linguistes.
Weinreich, U. (1969). Problems in the analysis of idioms. In J. Puhvel (Ed.), Substance and
structure of language (pp. 23–81). Berkeley, Los Angeles CA: University of California
Press.
Dictionaries and databases
British National Corpus – World Edition (2000). Humanities Computing Unit of Oxford
University on behalf of the BNC Consortium. Oxford.
Collins Cobuild. Dictionary of Idioms (1995), R. E. Moon (Ed.). London: Harper Collins.
Collins Cobuild English Language Dictionary (1987), J. Sinclair (gen. ed.). London: Pons.
Longman Dictionary of Contemporary English (1987). Berlin: Langenscheidt, Longman.
Longman Dictionary of English Idioms (1979). Th. Hill Long (Ed.). London: Longman.
Oxford English Dictionary Online. ww.oed.com

JB[v.20020404] Prn:14/03/2006; 10:57
F: HCP17AI.tex / p.1 (48-188)
Author index
A
Abeillé , , , 
Allerton 
Anders 
Arnold & Sadler 
Aston & Burnard 
B
Baranov & Dobrovol’skij 
Barcelona , 
Barkema , , , , 
Barz 
Bates et al. , , 
Bauer 
Bechtel & Abrahamsen 
Bell & Bobrow 
Berlin et al. 
Black 
Blakemore 
Blasko & Connine 
Botelho da Silva & Cutler
, , 
Burger , , , , , , ,
, , , , , ,
, , , –, ,
, , , , , 
Burger et al. , , –,
, , , , , ,
–, , , 
C
Cacciari , , , 
Cacciari & Glucksberg ,
, –
Cacciari & Tabossi , 
Carroll , , 
Carston 
Carter , , , , ,
, 
Čermák 
Chafe , , , , 
Chomsky , , , , 
Coffey 
Colombo 
Coulmas 
Cowie 
Cowie et al. 
Croft , , –, , ,
, , 
Cruse , 
Cutler 
D
Davidson 
Dietz 
Dobrovol’skij & Piirainen
, –, , 
Dobrovol’skji , , , ,
, , –, , , ,
, , , , , ,
, , , , , 
Ďurčo 
E
Elman 
Ernst , , , 
Estill & Kemper , 
F
Fauconnier , 
Fauconnier & Turner , 
Fellbaum , , , ,
, 
Fernando –, 
Fernando & Flavell 
Ferris 
Feyaerts 
Fillmore 
Fillmore & Atkins 
Fillmore et al. 
Fleischer , 
Flores d’ Arcais 
Fraser , , , , 
G
Gazdar et al. , 
Geeraerts , , , ,
, , , 
Geeraerts & Bakema 
Gernsbacher & Robertson

Gibbs , , , , , ,
, , , , , 
Gibbs & Colston 
Gibbs & Gonzales 
Gibbs & Nayak 
Gibbs & O’Brien , , 
Gibbs et al. 
Giora & Fein , 
Gläser , , 
Glass 
Glucksberg , –, ,
, , , , , 
Glucksberg & Keysar 
Glucksberg & McGlone ,

Glucksberg et al. , 
Goatly , 
Goldberg 
Goossens , , 
Gréciano 
H
Häcki Buhofer , 
Halliday , , 
Harras 
Hays , 
Hegedüs-Lambert 
JB[v.20020404] Prn:14/03/2006; 10:57
F: HCP17AI.tex / p.2 (188-337)
 Author index
Hemmi 
Hessky 
Hundt 
I
Indurkhya , , , , 
J
Johnson , , 
Johnson-Laird 
K
Kövecses & Szabó 
Katz , –
Katz & Postal , 
Kay & Fillmore 
Keil , , , 
Kemmer & Barlow , ,

Keysar & Bly , 
Koller , 
Korhonen 
Kroenenfeld , 
Kuiper & Everaert 
L
Lakoff , , , , , ,
, , –, , , ,
, , , , 
Lakoff & Johnson , , ,
, , , , , , 
Lakoff & Turner , , ,
, , 
Lamb , , –, , ,
, , , 
Langacker , , –,
–, , , , –,
–, –, , , ,
, , , , , ,
, , , , , ,

Langlotz , , , , ,
, , , , , 
Leech 
Levelt 
Levorato 
M
Makkai 
McArthur 
McGlone et al. , , , 
Mel’čuk , , , , , ,
, , , 
Mena Martínez , 
Minugh 
Moon , , , , , ,
, –, , , ,
, , , , , 
N
Napoli 
Nayak & Gibbs , 
Newmeyer , , , 
Nicolas –, , , ,
, , 
Nordén 
Nunberg , , , 
Nunberg et al. , –, ,
, , , , , , ,
, , , 
O
Ortony et al. , 
Orwell 
P
Pawley & Syder , 
Pilkington , , 
Pilz 
Q
Quirk et al. , –,
, , , , , ,

R
Radden 
Radford 
Reddy 
Richards 
Rosch , 
Rosch et al. 
Rothkegel 
Ruiz de Mendoza Ibáñez

Rumelhart 
S
Sabban , , , , ,
, , , , 
Salminen 
Schank & Abelson 
Schenk , , , , , ,
, , , , , 
Schwarz –
Schweigert , 
Searle , 
Shannon & Weaver 
Sinclair , , 
Skandera 
Smith 
Sperber & Wilson –,
, , , , , ,

Stein 
Strässler 
Swinney & Cutler , 
T
Tabossi & Zardon , 
Talmy , 
Tannen 
Taylor , , 
Thun 
Titone & Connine , , 
Tucker 
Turner , 
Turner & Fauconnier ,
, 
U
Ungerer & Schmid 
V
Van de Voort & Vonk 
Veisbergs 
Verhagen 
Vinogradov 
W
Warrren 
Wasow et al. , , 
Weinreich , –, –,
, 
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.1 (47-140)
Subject index
A
abstraction (cognitive) , , , , , 
see also schematicity vs. specificity
activation probabilities , –
activation-set 
see also idiomatic activation-set
linguistic (symbolic) –, , , –
adnominal modification , –, , –,

see also premodification, postmodification
alternative symbolisation , 
alteration , 
ambiguation , , , , –, 
analysability
see also isomorphism
constructional –, 
cognitive view of 
idiomatic , –, –, 
anaphora
see idiomatic anaphora
antonymy , –
apposition
and premodifier , 
and of complement –, , –,
, 
article
see also article variation
definite –, –
indefinite (zero) –, –
article variation , , , , , ,
–, –, 
and constructionally-idiosyncratic idioms

and existential constructions –
and isomorphism/analysability –,
–, –
and opaque idioms , 
and unique situational reference –
archetypes (conceptual) , , 
archetype metaphors –, –, –
asyntactic idiom
see constructional idiosyncrasy
autonomous syntactic (rules, transformations,
processing) , , –, , , , ,
, , 
B
base (conceptual) 
base form –
basic level (concept) , 
and coding , , , 
basis/context of activity/development is
a path-metaphor 
billiard-ball model 
blending (conceptual) –, , 
see also idiom motivation
and idiom variation , , , 
blended metaphorical idiom , , , 
BNC (British National Corpus) –,
–, 
bottom-up (composition, processing) , ,
, , –, –, , 
see also top-down
building-block view of (language, meaning)
, , , 
C
categorisation –, 
and linguistic coding –, 
and idiomatic coding 
cause for effect-metonymy , , –
Cheshire cat –, , , 
Chomskyan paradigm , –, , , 
coding –
decisions , –
idiomatic –, 
parameters , –, 
problem of , , –
Cognitive Grammar –, 
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.2 (140-221)
 Subject index
cognitive environment –, , , 
mutually-manifest –, , 
cognitive micro-models , –, ,
–, , , –, , 
explanatory 
symptomatic 
ontologising 
descriptive 
comparison (cognitive) –
compatibility (constraint) , , , ,
, , –
complex scene –
see also literal scene, figurative scene
and idioms –, , , –, 
background and foreground of –
compositeness 
cognitive view of –
composition –, 
see also compositionality
scaffolding function of , 
compositionality
see also non-compositionality, literal
compositionality
and idioms –, –, –, ,
–, , 
principle of , 
cognitive view of 
computation (cognitive) , 
concept –
conceptual backing 
conceptual core
see figurative core
conceptualisation –, 
see also target conceptualisation
bipartite 
coding of –
conceptual metaphor –, 
and idiom motivation , –, –,
, –, –, 
and idiom production –, –,
, 
composite (see metaphorical composition)
cognitive functions of , 
conceptual priming –
configuration hypothesis , , –, , ,
, –, 
conjunction variation , –, 
connectionism –, 
controversial data –
constituents (lexical) 
garden-path –
key (see key)
schematic , 
semantic autonomy of (see semantic
autonomy)
construal (of conceptual content) , –,
–, 
constructional adaptation –, , ,

constructional idiosyncrasy –, , ,
–, , , , –, 
constructional schema , –, –
and idiom structure –, , , 
and idiom variation –, , , 
contamination
see formal blend
content requirement 
contextual effects –
and idiom variation , , , ,
–, 
contiguity (conceptual) , , , –, 
control is physical grasp-metaphor ,
, 
coordination , 
core vocabulary , 
corpus data –, –, 
cranberry (collocation, constituents, morphs)
, , –, , –
creativity , 
see also linguistic creativity, idiomatic
creativity
creativity principle (open-choice principle)
–
D
dead metaphor theory 
dealing with an issue is
consumption-metaphor –
decomposable idioms –, , 
decomposition hypothesis –, –, ,

dephraseological derivation 
designation , 
determiner variation
see article variation
development of a configuration is
life-metaphor –
development of a love relationship is a
journey-metaphor –
development is processing of a
machine-metaphor 
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.3 (221-312)
Subject index 
developing configuration is a
vehicle-metaphor –
direct-access hypothesis , 
domain , 
see also source domain, target domain
hybrid –, , , 
spf (success, progress, failure) –
E
elaboration , , 
elaborative distance –
emblem –, 
entrenchment (cognitive) , , –
erroneous variant , , , , ,
–
etymological basis –, , –, 
event (cognitive) 
autonomous vs. peripherally-connected 
event structure-metaphor , 
evocational autonomy , 
experientialist realism (experientialism) –
external modification (of NP) –, –
see also premodification
F
failure is down-metaphor 
failure is death-metaphor 
figurative competence , –, , 
figurative core –, –
figurative scene , 
figurative-literal isomorphism , , ,
–, 
see also analysability, isomorphism
fixedness , , , –, 
focal adjustments , –
force-dynamic relation , 
and idioms , , , , , , 
formal blend , –, 
frame , 
fronting
see topicalisation
frozenness , , , , , , 
cognitive view of –, –
constructional 
functionality (constraint) –, 
G/H
Glucksberg-Lakoff controversy , , 
grammaticality (constraint) –, –
ground (metaphorical) , , , , ,

see also link
grouping , , 
homonymisation –
I
idealised cognitive model (ICM) , 
for success, progress and failure (spf)
–
ideational function –, , , 
idioms
as semantic units –, –, 
as long words –, –
cognitive functionality of , , –,
, , , , 
definition of 
generative treatment of –, –, , 
opaque (see opaque idioms)
orthodox view of 
compositional view of 
idiomatic activation-set , , –, ,
–, –, , , 
idiomatically combining expression –, ,

idiomatic anaphora –, –
idiomatic cataphora –, , , 
idiomatic creativity –, –
cline of –, –, 
intrinsic , , , , 
variational , 
idiomaticity 
idiomatic meaning , , , –
adaptation of , , –, , 
idiomatic phrases –
idiomatic scene
see figurative scene
idiom comparison rule 
idiom comprehension –, –, , ,
–
idiom-external relativisation 
idiom-internal relativisation 
idiom-list hypothesis 
idiom list model –
idiom motivation , –, –, –,
, , –, 
and blending –, , 
and conceptual metaphor , –,
–, , –, –
and emblems , , –
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.4 (312-392)
 Subject index
and metonymy –, –, , ,
, –
and metaphtonymy , , , ,
–
constituental , , , –
global , , –
Lakoff ’s model of –, , 
idiom principle –, 
idiom processing
see idiom production, idiom
comprehension
idiom production , –, –,
–
idiom standard
see standard
idiom transparency , –, –, ,
–
types of –, 
idiom variant 
idiom variation
constraints (see variation constraints)
formal –
principles of (see variation principles)
problems of –
semantic –
systematic vs. non-systematic , ,
–, , –
idiom-variation grammar –, , ,
, , 
ill-formed idiom
see constructional idiosyncrasy, blended
metaphorical idiom
image-metaphor , , , 
imagery –
image schema –, 
implicated conclusion , , 
implicated premise , 
implicatures –
strong , , –, –
weak , , –, , –,
–, 
inheritance hierarchy , 
intimacy effects , –
integration
conceptual (see blending)
syntagmatic (linguistic) , , –, 
intensification , –, , –, 
interaction of metaphor, metonymy, blending,
emblem –, , 
interactive processing models , , 
internal modification (of NP) –, –
see also premodification
interplay of levels of meaning , –,

instability/insecurity is water-metaphor
, , 
instability/uncertainty is air-metaphor
, 
instantiation 
institutionalisation , , 
institutionalised variant 
intensity of activity is heat-metaphor 
intentional vs. non-intentional variants ,

isomorphism –, –, 
see also analysability, figurative-literal
isomorphism
K
key (constituent) , –, , , , ,
–, 
kinetic idioms , 
L
latently isomorphic idiom 
lexical idiom –
lexical representation hypothesis , , ,

lexical substitution , , , –
and isomorphism –
vs. appositive premodification and
postmodification 
lexicalisation –, , 
lexicalised figurative sense , –, 
linguistic creativity –, –, , 
see also idiomatic creativity
as pattern re-forming , 
cline of 
link (metaphorical) , , –, , 
see also ground
literal compositionality , , –, ,

literally non-compositional idiom –, ,
, 
literal meaning , , –
literal scene , , –, –, –
see also literal-scene manipulation
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.5 (392-484)
Subject index 
literal-scene manipulation , –, 
and passivisation –
and adnominal modification –,
–
and lexical substitution –
literal-scene setting , 
O
occasional variant , –, , 
one-shot metaphor
see image-metaphor
opaque idiom , , , –, , –,
–, –
M
meaning adaptation , –
see also idiomatic meaning
metaphor
conceptual (see conceptual metaphor)
dead , 
live 
sleeping 
pragmatic view of –, , 
metaphorical coherence , , 
metaphorical composition –, –
metaphorical correspondences , –
and idiom motivation , , , ,
–, 
metaphtonymy –
metonymy (conceptual) –, 
mental space theory 
mind (as a dynamical network) –
multi-layered cognitive network –
multiple metaphor motivation –
P
paradigmatic dimension
see prismatic model of idiom semantics
parasitic elaboration , , , , 
part for whole-metonymy , –
partially compositional idiom , , –
passivisation , , –, , –,
–, 
and isomorphic idioms –
and latently isomorphic idioms 
and opaque idioms –
supporting and constraining factors
–
pattern of figuration , , , , –,
–, 
perspectivisation –, , –
perspectival variants
see perspectivisation
phrase idiom –, 
phrase-induced figurative sense , , ,
–, 
phrase-induced polysemy model (PIP) –
poetic (stylistic) effects –, , , 
polysemous idioms 
positive/negative development is
balance/imbalance-metaphor 
postmodification 
by of-complement , , –, ,
–, 
by prepositional phrase –
by relative clause –, –
partitive 
pre-coded (construction, expression, etc.)
–, –, –, , 
predication –, 
of idiom –
scope of , , 
premodification 
by adjective –, , –, –,
–
by noun –
appositional 
prismatic model of idiom semantics 
N
network
view of grammar, , 
view of idiom –, , 
view of conceptual metaphors –, ,
–
neutral use (of idiom) , 
non-ambiguity (constraint) , 
non-compositionality , , , –, 
non-conspicuity (constraint) , 
number –
number variation (pluralisation) –,
–, , , –
and isomorphism , –
and mass nouns –
and opaque idioms 
and motivation –
and constructionally-idiosyncratic idioms

JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.6 (484-566)
 Subject index
paradigmatic dimension of –, 
syntagmatic dimension of –, –
profile , 
profiling , 
and idioms , –, , –
problems are obstacles-metaphor 
problems/impediments are (oppressive)
forces-metaphor , –, 
progress is forward movement in
space-metaphor , –, –
projection , , , 
proverbiality –, –, 
pseudo variant , , , 
psycholinguistic models –, –, –,
–, –, –
punning , , –, 
pun variant , , , , , , ,
, , 
purposeful activity is boxing-metaphor
, 
purposeful activity is a
chase/hunt-metaphor 
purposeful activity is a
fight/battle-metaphor –
purposeful activity is a game-metaphor
–
purposeful activity is a race-metaphor
–
purposeful activity/development is
swimming-metaphor 
R
recognisability (constraint) –, ,
–
relevance theory –, , –, , 
restricted collocability 
restricted collocation 
routine, routinisation
cognitive , , , 
linguistic , , –, , , , 
metaphorical 
S
salient idiomatic substructure –, 
scaffolding (conceptual) , , , –,
, , 
scene 
see also literal scene, figurative scene
schema –, 
schematicity (vs. specificity) , , 
and idioms , , –, 
schematic reformulations , 
script , 
security/stability is solid
ground-metaphor 
selection , , 
segmentation , 
semantic autonomy , 
semantic extension –, , , , –
semi-idiom , 
sensorimotor data set –, –
simultaneous processing models –, ,
–, , , 
situational reference (of definite article) ,
–
somatic idioms , , , 
source domain (definition of) –
for spf-idioms –
specification , , –
spf-idioms (definition of) –
standard (of categorisation) , –
idiomatic –, –
linguistic –
metonymic 
states/situations are containers ,
–
strong communication , , 
stylistic effects
see poetic effects
stylistic markedness , 
substructure (symbolic, conceptual) 
and idiom structure , , –,
–, , –, 
success is up-metaphor 
symbolisation 
symbolic units , , 
continuum of , 
(grammar as) structured inventory of 
meaning of –
synonyms , , , , –, 
syntagmatic dimension
see prismatic model of idiom semantics
systematic variation (variant)
see idiom variation
T
target (of categorisation) , –
idiomatic –
linguistic –
metonymic 
JB[v.20020404] Prn:14/03/2006; 11:04
F: HCP17SI.tex / p.7 (566-616)
Subject index 
target conceptualisation –, 
target domain (definition of) –
for spf-idioms –
taxonomy –, , , , 
top-down
analysis , , –, 
projection , 
topic (metaphorical) , , , 
topic indication , , –, , 
and adnominal modification –,
–
and lexical substitution –
topic indicator , , , , , –
topic-related literal-scene manipulation ,
–, 
topicalisation , , , , , , , 
transformational deficiency –, –, ,

transitive construction , 
and passive –
coding of –
U
unit (definition of) 
usage-based (model of language) –, ,
, , , 
usage-event –
idiomatic –
usual variant , –, –, 
V
variability (definition of) 
cognitive view of –
variation constraints –, –
variation potential , –, 
variation principles –, , , –
vehicle (metaphorical) , , , , ,

voice –
W
weak communication , , , , 
wordplay , ––, –, –,
, , , , , , 
In the series Human Cognitive Processing the following titles have been published thus far or are
scheduled for publication:
17 LANGLOTZ, Andreas: Idiomatic Creativity. A cognitive-linguistic model of idiom-representation and idiomvariation in English. 2006. xi, 325 pp.
16 TSUR, Reuven: ‘Kubla Khan’ – Poetic Structure, Hypnotic Quality and Cognitive Style. A study in mental, vocal
and critical performance. 2006. xii, 252 pp.
15 LUCHJENBROERS, June (ed.): Cognitive Linguistics Investigations. Across languages, fields and philosophical
boundaries. xiv, 328 pp. + index. Expected June 2006
14 ITKONEN, Esa: Analogy as Structure and Process. Approaches in linguistics, cognitive psychology and
philosophy of science. 2005. xiv, 249 pp.
13 PRANDI, Michele: The Building Blocks of Meaning. Ideas for a philosophical grammar. 2004. xviii, 521 pp.
12 EVANS, Vyvyan: The Structure of Time. Language, meaning and temporal cognition. 2004. x, 286 pp.
11 SHELLEY, Cameron: Multiple Analogies in Science and Philosophy. 2003. xvi, 168 pp.
10 SKOUSEN, Royal, Deryle LONSDALE and Dilworth B. PARKINSON (eds.): Analogical Modeling. An
exemplar-based approach to language. 2002. x, 417 pp.
9 GRAUMANN, Carl Friedrich and Werner KALLMEYER (eds.): Perspective and Perspectivation in Discourse.
2002. vi, 401 pp.
8 SANDERS, Ted J.M., Joost SCHILPEROORD and Wilbert SPOOREN (eds.): Text Representation. Linguistic
and psycholinguistic aspects. 2001. viii, 364 pp.
7 SCHLESINGER, Izchak M., Tamar KEREN-PORTNOY and Tamar PARUSH: The Structure of Arguments.
2001. xx, 264 pp.
6 FORTESCUE, Michael: Pattern and Process. A Whiteheadian perspective on linguistics. 2001. viii, 312 pp.
5 NUYTS, Jan: Epistemic Modality, Language, and Conceptualization. A cognitive-pragmatic perspective. 2001.
xx, 429 pp.
4 PANTHER, Klaus-Uwe and Günter RADDEN (eds.): Metonymy in Language and Thought. 1999. vii, 410 pp.
3 FUCHS, Catherine and Stéphane ROBERT (eds.): Language Diversity and Cognitive Representations. 1999.
x, 229 pp.
2 COOPER, David L.: Linguistic Attractors. The cognitive dynamics of language acquisition and change. 1999.
xv, 375 pp.
1 YU, Ning: The Contemporary Theory of Metaphor. A perspective from Chinese. 1998. x, 278 pp.

Idiomatic Creativity - Perpustakaan STKIP Kusuma Negara

Transcription

Similar documents

This is only a preview. The actual page count of this

Unloved, over-loved or just misunderstood?

Lesson Share - Onestopenglish

HCID - sept 2011 - Namahn

Something to say

Hazework for Std 6M ICT Using Microsoft Words and please refer to

Graphical Methods for Describing Data 6` 57

ESLIN BSL