Development of metaphor and metonymy comprehension

547
The
British
Psychological
British Journal of Developmental Psychology (2010). 28. 547-563 — . y - - . —
.
<£> 2010 The British Psychological Soaety • H S S S Society
www.bpsiournals.co.uk
Development of metaphor and metonymy
comprehension: Receptive vocabulary and
conceptual knowledge
Gabriella Rundblad'* and Dagmara Annaz^
'Department of Education and Professional Studies, King's College London, UK
^Centre for Brain and Cognitive Development, Birkbeck University of
London, UK
Figurative language, such as metaphor and metonymy are common in our daily
communication. This is one of the first studies to investigate metaphor and metonymy
comprehension using a developmental approach. Forty-five typically developing
individuals participated in a metaphor-metonymy verbal comprehension task
incorporating 20 short picture-stories. Cross-sectional trajectory analyses linking
task performance to either chronological age or receptive vocabulary (mental age, MA)
were used to compare the development of metaphor and metonymy. Results showed
that development of metaphor and metonymy comprehension is strongly linked with
chronological and MA, but metaphor comprehension develops at a slower rate
compared to metonymy. It was also found that participants, across all ages, consistently
showed around 21% better performance on metonymy. The relationship between
metaphor and metonymy comprehension is discussed in terms of linguistic and
cognitive models of figurative language comprehension arguing that metonymy is
cognitively more basic than metaphor.
Figurative language is a ubiquitous part of everyday communication and everyday
life, and is therefore readily used and understood by ordinary speakers and listeners
(e.g., Pollio, Barlow, Fine, & Pollio, 1977). Yet figurative expressions have commonly
been treated as deviations from some literal equivalent (e.g., Corbett, 1990), with
metaphors referred to as manipulations of meaning that children need to grasp in order
to enjoy creative literature and poetry in particular (e.g.. Smith, 1976). Figurative
language studies have also persistently focused on a narrow range offigurativelanguage
devices, mainly metaphors, resulting in an incomplete and potentially skewed picture of
figurative language comprehension and production. In particular, development of
metonymy comprehension and production has fundamentally been ignored.
* Correspondence shouid be addressed to Dr Cabriello Rundbiod, Department of Education and Professional Studies,
King's College London, Waterloo Road, London S£/ 9NH, UK (e-mail: [email protected]).
DOI: 10.1348/0261510O9X45'4373
548
Gabriella Rundblad and Dagmara Annaz
In the current study, we emphasize a developmental approach by constructing crosssectional trajectories of children as young as 5 years of age to 37-year-old adults to
examine development of metaphor and metonymy comprehension. To this end, we will
employ a methodology that highlights developmental change over time (KarmiloffSmith, Thomas, Annaz, Humphreys, Ewing, & Brace, 2004; Thomas et al, 2009). Our
study is of particular relevance due to the limited scope of previous studies that either
equate metaphor •withfigureof speech, or, as in the case of MacKay and Shaw (2004),
exclude metaphors, thus limiting the comparability of the study. This is, to the best of
our knowledge, the first study to compare and contrast the developmental profiles of
metaphor, and metonymy comprehension. Our data reveal for the first time that
comprehension of metaphors and metonyms steadily develops into adulthood. Perhaps
the most striking result of the current study is the consistent difference in individual
performance on the task with metonym scores being around 21% higher than metaphor
performance scores.
In the following sections, we wül briefly review research on figurative language
processing, production, and comprehension, and then consider the development of this
ability in typically developing children and adults. We will first, however, need to
establish the definition and potential relation between metaphors and metonyms.
Metaphor and metonymy
Traditionally, metaphor and metonymy used to be distinguished as relying on similarity
versus contiguity (e.g., UUmann, 1967). We would thus distinguish between yace in
'There is a new^ face on the clock' versus 'There is a new face in the office'; here, the
former is a metaphor where a part of the body is compared to a part of a clock, whereas
the latter is a metonym where the expression face stands for the whole person as
opposed to just a part of the body (i.e., the contiguity is situated in the spatial part-whole
relations of the human body).
Recently, the distinction of conceptual similarity versus conceptual contiguity was
reintroduced in combination with prototype theory. Building on Barcelona (2002, 2003)
and Dirven (2002), Peirsman and Geeraerts (2006) suggest that metonymy is a
prototypically structured concept with the prototypical core of contiguity being t()rmed
by spatial part-whole relations (see also Blank, 2003; Feyaerts, 1999; Seto, 1999). Similarly
for metaphor, Shen (1992) argues that the comparison involved in metaphorical
expressions constructs an ad hoc category where one of the terms (usually the second
one [e.g., lion in Peter is a lion']) is the prototypical member of that category.
In contrast, cognitive models of figurative language, such as conceptual metaphor
theory (CMT), treat figurative language as part of the norm in human communication,
arguing that the human conceptual system is figurative in nature (Lakoff & Johnson,
1980). In CMT, metaphor and metonymy are commonly contrasted as a mapping across
two conceptual domains versus a mapping within a single conceptual domain
(Kövecses, 2002; Lakoff, 1987). Thus, the metonym/«cc^ is based within the domain
HUMAN BEING, whereas the metaphor^ce crosses from this animate domain into the
inanimate domain CLOCK. The domains are structured by an idealized cognitive model
(ICM), and by shifting salience from one concept to another (across domains or within
the same domain), particular properties and aspects of the domain(s) and the ICM can
be highlighted (Croft, 2002). Unfortunately, domains as well as ICM are quite vague
concepts (see Croft & Ouse, 2004; Feyaerts, 1999; Riemer, 2001, for discussions),
leaving the metaphor-metonymy distinction somewhat imprecise.
Deve/of)ment of metaphor and metonymy 549
Whether employing a CMT approach, a prototype/categorization approach or a
traditional contiguity/similarity distinction, one difference does seem to remain
constant; metaphor is more complex than metonymy (i.e., metaphor involves two
rather than one domain, metaphor requires the constrtiction of an ad hoc category, or
similarity between two concepts is hard to assess unless contiguous relations within
each of the two concepts have been determined). We will argue that, from a
developmental point of view, it is therefore highly likely that metonymy understanding
will exceed metaphor comprehension at any point in time throughout childhood to
adulthood. In addition, it is also likely that metonymy develops earlier and faster.
Figurative language processing
Early pragmatic studies postulated an initial literal interpretation that only in the event of
inteqiretation failure would trigger a subsequent search for a figtirative interpretation
(Cîrice, 1975; Searle, 1979). More recent figurative langtiage processing studies have,
however, shown that figurative phrases do not require additional processing time
compared to literal phrases (e.g., McElree & Nordlie, 1999). Ciibbs (1984, 1990, 2002)
and Turner and Katz (1997) have proposed that the processing offigurativeexpressions
is direct and requires neither preceding nor subsequent literal interpretation; a literal
interi')retation can be forced, but only upon completion of the figurative interpretation.
Recent psycholinguivStic parallel models (e.g.. Frisson & Pickering, 1999, 2001;
Cilucksberg, 1991 ; Keysar, 1989; Keysar, Shen, Glucksberg, & Horton, 2000) suggest that
both a literal and afigurativeinterpretation are available and processed at the same time,
and that selection is a matter of familiarity and salience. Frisson and Pickering (2007)
have shown that familiar metonyms are easier to process than unfamiliar ones, though
context can render an unfamiliar metonym familiar. Similarly, interjiretations of literal
versus metaphorical language show no difference in brain activation if amjîle and
relevant context is stipplied (Coulson & Van Petten, 2002; Pynte, Besson, Robichon, &
Poli, 1996). The role of salience has been elaborated upon using Giora's (1997, 2002)
graded salience hypothesis, with results showing that more salient meanings (literal or
figurative) were activated earlier and accessed faster than less salient ones (which also
can be literal or figurative).
Figurative language production
Early studies offigurativelanguage predominantly focused on the production of similes
(e.g., 'It's like a snake' of a winding river, Piaget, 1962) and metaphors (e.g., 'I'm a big
waterfair when sliding down, Carlson & Anisfeld, 1969). At first, data tended to be
extracted from recordings of spontaneous speech. In later studies, various elicitation
methods were used. It should be noted that many studies are confounded by
interchangeable use of metaphors with similes, idiom, and proverbs, treating and
labelling them metaphors. This incongruous use of the term metaphor has probably
aided the disagreement as to when figurative language production actually arises, with
the earliest studies generally concluding that metaphor production emerges as early as
around 18-24 months (e.g.. Billow, 1981; C:arlson & Anisfeld, 1969; Piaget, 1962;
Winner, 1979). Whether these so-called child metaphors' are truly metaphorical has
been hotly debated in light of metaphor and categorization theory advancements (see
Gentner (1988) and Vosniadou (1987), for overviews) and some of the studies have
critically tried to distinguish metaphors from overextensions (e.g.. Winner, 1979).
550
Gabriella Rundblad and Dagmara Annaz
In general, production studies have struggled to confirm that an utterance is a conscious
violation of reality motivated by the perception of a similarity between the two concepts
being compared.
Figurative language comprehension
Similarly to production studies, early comprehension studies presumed that evidence of
analogical linking of concepts based on perceptual similarity indicated presence of
metaphorical understanding. Again, use of different methods and classifications brings
inconsistencies and lack of agreement in findings. For example. Winner and Fngel
(1980) used a matching task where classification of an item in terms of perceptual
similarity to another item was scored as metaphorical, Vosniadou and Ortony (1983)
asked participants to complete similarity statements involving similes, and Billow
(1975) used a sentence explanation task with supporting visual stimuli that mixed
perceptual and ñmctional metaphors with proverbs. Moreover, the expressions tested
were often constructed idiosyncratic metaphorical expressions rather than acttial
natural language expressions.
As explored by Waggoner, Messe, and Palermo (1985), recall and construction of
meaning for both literal and metaphorical expressions is dependent on supporting
contexts, preferably extended and appropriately structured stories. Consequently,
results from studies such as Billow (1975; see also Winner and colleagues (1976)) may
be confounded. Smith (1976) contrasted metaphor comprehension in children aged
11-12 years versus 13-14 years by supplying the children with a metaphor in context
and asking them to explain the context and the target metaphor. Unfortunately, Smith's
contexts generally included other figurative expressions in addition to the target
metaphors. Since task complexity has a marked effect on metaphor comprehension
(Vosniadou, Ortony, Reynolds, & Wilson, 1984), Smith's conclusion (1976, pp. 238) that
metaphor is a new cognitive domain' for 11- to 12-year-olds is highly qtiestionable.
Using short stories with familiar concrete nominal metaphors as targets, Jones and
Stone (1989) found that participants performed better when asked to paraphrase rather
than explain the meaning of the metaphor. They notably used prompts in order to force
participants to disambiguate responses. One drawback with paraphrase studies is,
however, that paraphrasing requires meta-cognitive abilities that generally develop late
in childhood and thus the cognitive demands of such tasks could prevent participants
from scoring well on the task whatever their level of metaphorical comprehension
ability (Vosniadou et al, 1984).
Due to the theoretical, methodological, and linguistic inconsistencies, studies have
not surprisingly yielded inconsistent data regarding the age at which comprehension
emerges. Reanaiysis of results taking these inconsistencies into account would suggest a
three-stage development of metaphor comprehension: exclusively literal interpretations
between 3 and 6 years old, onset of metaphor understanding between 7 and 10 years
old, and an integrated understanding coupled with the ability to verbalize their
understanding from the age of 11 years old. Özcaliskan (2005) used story contexts in a
forced-choice task to test this. She found that age 4 marked the onset of metaphor
comprehension and age 5 the onset of abstract verbal reasoning about metaphorical
mappings. These new findings can, to a large extent, be attributed to the test materials
used. In particular, target words were all metaphorical extensions of verbs based on the
typical metaphorical mappings described in CMT (e.g., Lakoff &Jolinson, 1980) elicited
from natural language.
'
Development of metaphor and metonymy 551
The current study aims to examine the dynamics of developmental profiles of
metaphor and metonymy comprehension rather than taking a static view of figurative
language comprehension, thus developmental trajectories or growth models will be
applied (Annaz, Karmiloff-Smith, & Thomas, 2008; Jarrold & Brock, 2004; KarmiloffSmith el al, 2004; Rice, 2004). Although one would ideally use a longitudinal design to
study development, cross-sectional studies can give an initial approximation of respective
developmental trajectories (Thomas et al, 2009), and these can then be validated by a
longitudinal follow-up. In order to use this design, it is important that the task is sensitive
to developmental change (see Annaz et al, 2008). The task designed for this study is
sensitive across a wide age range, thus ideally suited to this approach. One of the main
benefits of tbis approach is the ability to establish the developmental relations between
diflc-rent experimental tasks/variables, assessing the extent to which performance on one
task/variable predicts performance on another task/variable across development.
Method
Participants
Forty-five typically developing children and adults (23 female, 22 male; age range
5; 3-37; 1 years old; M = 17; 11) participated in the ctirrent study. Details can be found
in Table 1. Child participants were randomly recruited from mainstream schools in
South-East England, and adults from universities and companies in South-East England,
by asking the schools, universities, and companies to forward information sheets and
consent forms. All participants had English as their first language, had normal or
corrected-to-normal vision and hearing, and had no language-learning disabilil).
Table 1. Participants' descriptives and metaphor-metonymy group scores
Statistics
CA (N = 45)
Metaphor
Metonymy
Mean
SD
Minimum/maximum
215
116
63-445
4.8
2.2
I-IO
7.0
2.2
3-10
Note. CA, chronological age.
The experimental protocol was approved by King's College London and Birkbeck,
University of London, College Research Ethics Committees prior to recruitment of
participants. Informed consent (including parental consent where appn)priatc) was
obtained before participation and participants were instructed that they were free to
withdraw at any time for any reason. Nofinancialcompensation was offered.
Background measures
The British Picture Vocabulary Scale II (BPVS) is a receptive vocabulary' test used from
age 3 to 18 (Dunn, Dunn, Whetton, & Burley, 1997). There are a total of 168 trials where
each trial contains one target word and four pictures. Depending on the child's
chronological age (CA), a subset of the 168 trials is administered (ceiling level 18 years
old). In each trial, the child is requested to point to the picture that illustrates the
meaning of the target word. In order to confirm that our participants fall within the
verbal IQ range expected from a typically developing population, all participants up to
the age of 18 years old were tested on the BPVS. The BPVS was administered in a
separate session (Table 2).
552
Gabriella Rundblad and Dagmara Annaz
Table 2. BPVS test age equivalent and metaphor-metonymy subgroup scores
Statistics
CA(N = 24)
BPVS (months)
Metaphor
Metonymy
Mean
SD
Minimum/maximum
119
36
63-185
126
46
66-206
3.1
0.95
1-5
2
3-10
Note. CA, chronological age; BPVS, British Picture Vocabulary Scale.
Material and procedure
Comprehension of metaphor and metonymy was tested using a new experimental task.
Thirteen metaphor and 13 metonymy stories were constructed. All target words were
lexicalized, nominal expressions and were matched on frequency and age of acquisition
using the British National Corpus (frequency counts supplied by www.kilgarriff.co.uk)
and the MRC Psychoiinguistic Database (www.psy.uwa.edu.au/mrcdatabase/
uwa_mrc.htm), with the exception of the metonym Robbie Williams which for
obvious reasons does not feature in these compilations.' The 26 stories were assessed
for comprehensibility through a pilot study with five children aged 5-9 years old.
Thereafter, 10 target stories of each type were selected. A familiarity rating of the
metaphoric and métonymie meanings of the final 20 target words (e.g.. How familiar
are you with FLOOD as in There was a flood (i.e., lots of people) outside the shop)
using afive-pointIikert scale (1 = very unfamiliar and 5 = very familiar) in a sample of
140 UK undergraduate native speakers of English showed no difference in familiarity
between the metaphors and metonyms (metaphors: mode = 5, M = 4.45, SD = 1.01;
metonyms: mode = 5,M = 4.38, SD = 1.11; Mann-Whitney U = 959,397.5,/>= .232
[two-tailed]).
The final test material consisted of 10 metaphors and 10 metonyms incorporated
into 20 short, simple picture-stories (see sample stories in Appendix A). Participants
were tested individually in a quiet room, with participants taking between 20 and
30 min to complete the task. Young child participants were offered a break half-way
through. Each story was read, section by section, by the experimenter, while presenting
the child with one picture per section. The last part of each narrative contains the target
metaphor/metonym accompanied by a picture of one of the characters and the
participant is asked to report what that character sees, i.e., to describe the scene, using
the question: What does Character X see?'. In order to ensure that the participant
understood what they had to do at the end of each story and to implicitly instruct the
participants that they needed to respond more elaborately than just repeating the target
word, a warm up story containing the hyponym/jei was used. In this story one of the
characters gets a pet, and consequently, the participants are forced to decide what pet
stands for (i.e., what kind of pet it is) when describing the scene. In addition, a total of
five story-specific prompts were constructed (examples are given in Appendix B). In the
event that the participant did not respond, three prompts (A1-A3) were available,
whereas if the participant repeated the target word or failed to disambiguate the target
word, two prompts could be used (B1-B2). For some stories, it could be argued that
the final prompt available (see prompts A3 and B2 in Appendix 2) is too revealing as
Robbie Williams is probably Britain's most famous pop singer (e.g., he was ranked as the 9th most famous person in the world
by British under 10-year-olds in 2005 according to Luton First (www.lutonfirstcom)).
Deve/opment of metaphor and metonymy 553
the relative pronoun used clearly relates to the intended figurative meaning (e.g., the
pronoun who in Who is the flood' where//ooi/ means a lot of people). However, our
participants' responses show that participants who interpreted the target word literally
usually did not reflect on the pronoun. Those participants who did notice the pronoun
commented on it, yet were persistent in their literal interpretation.
Each narrative was constructed such that the figurative meaning of the target word
was more salient, whereas the literal meaning was possible but highly unlikely. Thus in
the sample stories, Robbie Williams could possibly be visiting and the river could
potentially have flooded the museum; however, it is far more likely that Kate found her
Robbie Williams CD in the lounge and that people were queuing outside the museum.
The narrative for each story is short: the average number of clauses per story is 14.45
(SD: 2.56) and the average number of words per clause is 6.48 (SD: 0.99). Each narrative
is accompanied by three to four simple, hand drawn and coloured pictures, allowing
each story to be divided into three to four sections in order to limit memor>' demands
and aid comprehension of the story. The order of the stories was pseudivrandomized
avoiding more than two stories with the same type offigurativelanguage in a row, and
the order of presentation was the same for all participants.
Scoring
For the purposes of quantitative analysis, responses were classified as either literal or
figurative interpretations of the target word. Only in those cases where the response
clearly demonstrated comprehension of the intended figurative meaning for the target
word did the participant score a pass and was classified score 1. In cases where the
participant did not respond at all, repeated the target word or gave an idiosyncratic
answer, the story-specific prompts were used forcing the participant to supply a
figurative or literal interpretation of the story end; thus making sure that the participant
could not fail the test due to lack of comprehension. Literal interpretations of the target
expression were scored 0.
In order to ensure reliability in coding of the participants' responses, each testing
session was videotaped for subsec|uent scoring and analysis. Two independent niters
coded all testing sessions, with an inter-rater reliability score of .917 (C^ohen's kappa;
p < .001) indicating an outstanding level of coding stability.
Results
Participants could achieve a maximum score of 10 correct on metaphor and metonymy
respectively. Figure 1 depicts the developmental trajectories linking performance with
CA for the participants on metaphor and metonym comprehension, respectively.
Analyses of covariance within the general linear model were used in the current data
.study. Cook's distance ftmction was used to identify potential outliers that could
influence regression. There were no outliers in the current data set. The data were
analysed as follows: (i) we first examined metaphor and metonymy developmental
trajectories separately. Each developmental trajectory was modelled by a linear function
relating individuals' accuracy scores to CA; (ii) then both trajectories were directly
compared in terms of onset and rate of development; (iii)finally,metaphor-metonymy
performance of younger participants was plotted against mental ages (MAs) from the
BPVS to explore whether performance is in line with a given standardized measure (see
Thomas et al (2009), for a similar approach).
554
Cabrielta Rundblad and Dagmara Annaz
Developmental trajectories of metaphor and metonymy
12
0
10
•o
8 8
<»
i 6
0
oo
0
0 o
•
< • ; / < » • •
(0
I 4
2
50
•
fl2 = 0.57^ •
o
^ ^
^ K»
o o
•
IJV^
•
•
3.—'
^ß,'^m
F^ = 0.69
••
u
.
0»
o
•
•
o Metonymy
• Metaphor
100 150 200 250 300 350 400 450 500
Chronological age (in months)
Figure I. Developmental trajectories (expressed in accuracy scores) on metaphor and metonymy
stories plotted according to the CA of the participants (months). Maximum scores participants could
achieve was 10 on each task.
Developmental trajectories of metaphors and metonyms: CA
Performance scores were calculated for each participant and the percentage values on
metaphor and metonymy were 48 and 70%, respectively. There was no floor effect on
either of the tests but around 13% of individuals achieved highest scores on the
metonymy task. Details of group characteristics are presented in Table 1. Initial
inspection of the data revealed that participants produced reliable linear trajectories
with large R^ values (metaphors: R^ = .69; metonyms: R^ = .57), thus explaining a
significant proportion of the model variance. Linear trajectories revealed that the
participants' accuracy performance on metaphors and metonyms improved reliably
with increasing CA (metaphors: f(l,44) = 95.29, p < .001, t^ = .689; metonyms:
F(l,44) = 58.04, p < .001, Ti/^ = .574).
Comparison of metaphor and metonymy trajectories: CA
A comparison of metaphor and metonymy developmental trajectories indicated that
overall performance significantly improved with CA (main effect of age:
/='(1,43)= 119.02, p<.OOl, -q^^ = .735). Next, we compared the trajectories for
performance at onset (i.e., at the earliest age participants were tested, 63 months) and
the rate of increase in performance relative to CA. There was a significant main effect of
task(F(l,43) = 24.99,/? = .001, Tq^^ = .368) indicating a reliably higher performance on
metonymy by youngest participants tested. However, there was no interaction between
age and task (F(l,43) = 42, p = .519, y]j, = .010) suggesting that performance on
metaphors and metonyms increases at similar rate of development with increased CA.
The relative difference between data points on metaphor and metonymy by individual
participants was examined and it was shown that individuals performed on average 21%
higher on metonyms in comparison to metaphors.
In order to be able to compare metaphor-metonymy performance with C;A and MA,
we generated a subsection of the full trajectories that would be overlapping CAs with
MAranges(see Figure 2).
Deve/opment of metaphor and metonymy 555
Metaphors and metonyms subtrajectories: CA
Analyses of subtrajectories (63-185 months old) revealed that participants produced
reliable linear trajectories with relatively large R^ valties (metaphors: R^ = .44;
metonyms: R^ = .46). Similar to analysis offtiUmetaphor and metonymy trajectories,
overall performance significantly improved with CA (main effect of age:
F(l,25} = 30.94,p < .001, 'r\j, = .584). However, in contrast to full-trajectories results,
there was no significant main effect of task (f(l,22) = 0.001, p = .970, iq^ = .001)
indicating similar onset of performance, and there was an interaction between age and
task (/•Xl,23) = 5.87,p = .024, T)^, = .211) suggesting that perfomiance on metaphors
improves at significantly slower rate than mettinyms in relation to increasing CA. These
differences in results are most likely influenced by ceiling scores (13%) on metonymy,
and produced artefacts in trajectories that mask real difference in parameters such as
onset or rate/growth over time.
Developmental trajectories of metaphors and metonyms: BPVS
We then examined whether there was a reliable relationship between metaphormetonymy trajectories and MA measures based on the BPVS language test to see
whether the behavioural scores are in line with the developmental state of other
aspects of the cognitive system. Due to limited test age for BPVS (upper limit is 18
years old), we analysed the developmental trajectory of 24 individuals (age range:
63-185 months old). As expected, in the typically developing population (since MA is
in line with CA: t= -2.28, p = .033), scores increased systematically with verbal
mental age (VMA) from the BPVS task. Initial inspection of the data revealed that
participants produced reliable linear trajectories witb relatively large R~ valties
(metaphors: R^ = .30; metonyms: R~ = .39), albeit lower values in comparison to
subscale-CA trajectories. Linear trajectories revealed that the participants' accuracy on
both metaphors and metonyms improved reliably with increasing MA (metaphors:
/^(l, 23)= 12.13, p=.OO2, T^| ^ = .355; metonyms: /'(I, 23) = 14.43, p < .001,
T\p — .396). A comparison of metaphor and metonymy developmental trajectories
indicated that overall performance significantly improved with VMA (main effect of
age: /-'(1,23) = 19.45,p < .001, T]^ = .469). Similarly to CA subtrajectories, there was
no main effect of task (F(l,23) = 0.7l6,p = .407, r\f, = .032) indicating no difference
it! performance between metaphor and metonymy at onset but a marginally significant
interaction between age and task (/'(I, 23) = 4.11, p = .055, r\jj = .157) stiggesting
that rate of performance on metaphors is slower in comparison to metonyms when
relative to VMA.
(a) 12
(b) 12
10
10
8
6
4
4
2
2
100
150
Chronoiogical age (in months)
Metaphor
8
4.- »
6
0
50
Metonymy
200
50
100
150
200
BPVS (in months)
Figure 2. Developmental subtrajectories (expressed in accuracy scores) on metaphor and metonymy
stories plotted according to (a) CA of the participants (months) and (b) BPVS MA.
556
Gabrielta Rundblad and Dagmara Annaz
Discussion
Due to theoretical, methodological, and linguistic inconsistencies, early studies have
yielded conflicting data regarding the age at which metaphor comprehension emerges.
In addition, studies have tended to limit their focus to a narrow range of figurative
language devices, leaving common devices such as metonymy largely unexplored. (;:MT
as well as traditional and recent similarity versus contiguity accounts of metaphor and
metonymy seem to agree that metonymy is cognitively more basic than metaphor icf
Lakoff, 1987; Peirsman & Geeraerts, 2006). The present study aimed to compare and
contrast the developmental profiles of metaphor and metonymy comprehension,
hypothesizing significantly better performance, and potentially also earlier onset and
faster rate of development, for metonymy.
Performance on a new metaphor-metonymy verbal comprehension task incorporating 20 short picture-stories revealed a developmental change on both metaphor and
metonymy comprehension with CA. For younger participants, trajectories linking
comprehension with receptive vocabulary ability similarly showed an increase in
performance with increasing VMA. Although performance variability was slightly larger
for CA, both CA and receptive vocabulary ability proved highly reliable predictors of
metaphor and metonymy performance.
These results confirm for the first time that comprehension of metaphor
and metonymy steadily improves throughout childhood to adulthood. They
also validate the task as a sensitive measure of developing language abilities in a
typically developing population. As we predicted, analyses divulged a consistent
difference in individual performance on the task, confirming a difficulty effect with
metonymy comprehension being superior to metaphor comprehension by approximately 21%. The advantage of metonymy over metaphor is consistent with a previous
study that reported faster processing times for metonymy (Klepousniotou &
Baum, 2007).
Our results are in line with the greater cognitive distance' between concepts
in metaphor (Pauwels, 1999, pp. 256) that seems to be the defining difference between
metaphor and metonymy. Similarly, Coulson and Van Petten (2005) suggest that
metaphor processing difficulty is associated with a varying degree of cognitive
constraints due to the complexity of the mappings across domains and the conceptual
integration that needs to take place. It could be argued that superiority on metonymy is
an artefact of lexicalization. However, our task contained both lexicalized metaphors
and metonyms matched on frequency and age of acquisition.
Categorization studies have shown that hierarchical classifications, such as
part-whole emerge very early on (e.g., Rosch, Mervis, Gay, Boyes-Braem, & Johnson,
1976). Combined with metonymy being easier and cognitively more basic, it
seems plausible to suggest that developmental onset of metonymy comprehension
precedes onset for metaphor comprehension. We tested this assumption by comparing
the trajectories for performance of our youngest participants (i.e., 63 months).
The results showed a significantly better performance for metonymy at onset (i.e., the
level of performance at the point at which we began measurement), which could
be interpreted to mean that children understand metonyms earlier than metaphors. We
investigated this further by generating a subsection of the full developmental
trajectories for participants up to the age of 15 years old, but interestingly, this effect
disappeared. In order to explain this ostensible discrepancy, we first need to look at the
rate of development for metaphor and metonymy comprehension.
Development of metaphor and metonymy 557
In contrast to ftill trajectories where ceiling scores on metonymy distorted results,
subtrajectories for younger participants revealed tbat metonymy understanding
improves at a significantly faster rate than metaphor understanding, in relation to
both increasing CA and increasing MA. The faster rate of metonymy development paired
with its ceiling scores could also explain the discrepancy for onset. In our subsample,
metonymy developed so much faster already at the point of onset that we found ceiling
scores from around the age of 12 years old. It is likely these ceiling scores 'cap' the
increasing difference in comprehension scores and thus tip' theftiUtrajectory to the
extent that a main effect for task at onset is measured.
The current results suggest that metaphor and metonymy start to develop at similar
CA. However, the current task was designed for school-age children who already
perform successftilly on metonymy (minimum: 3 of 10). Thus, it remains inconclusive
whether onset for these twofigurativedevices overlaps. To clarify this, we would need
to devise a test that is suitable for preschool children.
While there may or may not be a difference in onset, there is clearly a difference in
rate of development. This difference seems indicative of different cognitive, neural
and/or environmental requirements for metaphor and metonymy. One of the most
striking features of metonymy is its conformity to a limited number of cognitive patterns
that each metonym traditionally subscribes to (Lakoff, 1987; Lakoff & Johnson, 1980).
Frisson and Pickering (2007) claim that these patterns (or rules) allow speakers to
construct novel métonymie senses without difficulty. Interestingly, most subtypes of
referential metonymies are attested almost universally (Brdar-Szabó & Brdar, 2003). We
would, therefore, expect that once speakers have acquired and learnt to apply
métonymie patterns, performance rates will increase rapidly and speakers will at a
conspicuously early point be able to achieve very high performance on typical
metonyms. The neural basis for the métonymie patterns remains to be explored, but
given its role in .semantic language comprehension particularly on a sentence level
(Kircher, Leube, Erb, Grodd, & Rapp, 2007), the anterior part of the left inferior frontal
gyms will most likely play a prominent role.
Similarly to metonymy, Lakoff (1987) has established recurrent cognitive patterns for
metaphor, though these are less universal than those established for metonymy.
Metaphor processing has been widely studied in event-related potential and functional
magnetic resonance studies. Recently, Stringaris, Medford, Giampietro, Brammer, and
David (2007, pp. 150) found activation of the left thalamus that could indicate specific
processing needs for metaphor due to its 'ad hoc concept construction and openendedness'. In sum, posing specific neural processing needs that require a greater
degree of environmental input, metaphor comprehension is likely to develop at a slower
rate, and potentially even later than metonymy.
Figurative language comprehension is of clinical relevance in developmental
disorders and neuropsychiatry (e.g., autism: Dennis, Lazenby, & Lockyer, 2001;
schizophrenia: de Bonis, Epelbaum, De Vez, & Feline, 1997). Moreover, in order to
further our understanding of the cognitive constraints in metaphor and metonymy
comprehension in the typically developing population, it is crucial to use crosssyndrome comparisons (Karmiloff-Smith, 1998). Preliminary analysis of a related study
on figurative language comprehension in Williams Syndrome tentatively show similar
performance for this group to typically developing participants on metonymy when
constructed according to MA on the BPVS, whereas metaphor comprehension is
severely affected. Our metaphor-metonymy cross-sectional study that traces
performance from school-age to adulthood allows us to conclude that the timing of
558
Gabrieiia Rundbtad and Dagmara Annaz
developmental change is open to exogenous influences which are of great importance
for fttture investigations.
Acknowledgements
We would like to thank all our participants, especially the children and their teachers for their
continuing collaboration in our research. Special thanks to Anna Ferdenzi and Jo Van Herwegen
for their assistance with recruitment and data collection. We are also grateful to Annette KarmiloffSmith, Jill Hohenstein, and Michael Thomas for fruitful discussions. This research was supported
by a KCL Seedcorn grant to G. R.
References
Annaz, D., Karmiloff-Smith, A., & Thomas. M. S. C. (2008). The importance of tracing developmental
trajectories for clinical child neuropsychology. In J. Reed & J. Warner Rogers (Eds.), Child
neuropsychology: Concepts, theory and practice (pp. 7-18). Maiden: Wiley-Blackwell.
Barcelona, A. (2002). Clarifying and applying the notions of metaphor and metonymy within
cognitive linguistics: An update. In R. Dirven & R. Pörings (Eds.), Metaphor and metonymy in
comparison and contrast (pp. 207-278). Berlin/New York: Mouton de Gruyter.
Barcelona, A. (2003). On the plausibility of claiming a métonymie motivation for conceptual
metaphor. In A. Barcelona (Ed.), Metaphor and metonymy at the crossroads: A cognitive
perspective (pp. 31-58). Berlin/New York: Mouton de Gruyter.
Billow, R. M. (1975). A cognitive developmental study of metaphor comprehension.
Deveioptnental Psychology, 11, 415-423.
Billow, R. M. (1981). Observing spontaneous metaphor in children./owrw«/ of Experimental
Child Psychology/, 31, 430-445.
Brdar-Szabó, R., & Brdar, M. (2003). Referential metonymy across languages: What can cognitive
linguistics and contrastive linguistics learn from each other? International Journal of English
Studies, 3, S5-IO5.
Blank, A. (2003). Words and concepts in time: Towards diacbronic cognitive onomasiology.
In R. Eckardt, K. von Heusinger, & C. Schwarze (Eds.), Words in time. Diachronie semantics
from different points of view (pp. 37-65). Berlin/New York: Mouton de Gruyter.
Carlson, P., & Anisfeld, M. (1969). Some observations on the linguistics competence of a two-yearold child. Child Development, 40, 565-575.
Corbett, E. P. J. (1990). Classical rhetoric for the modem student, ^ew York: Oxford University
Press.
Coulson, S., & Van Petten, C. (2002). Conceptual integration and metaphor: An event related
study Memory & Cognition, 30, 958-968.
Croft, W. (2002). The role of domains in the interpretation of metaphors and metonymies.
In R. Dirven & R. Pörings (Eds.), Metaphor and metonymy in comparison and contrast
(pp. 161-205). Berlin: Mouton de Gruyter.
Croft, W, & Cnise, D. A. (2004). Cognitive linguistics. Cambridge: Cambridge University Press,
de Bonis, M., Epelbaum, C, Deffez, V, & Feline, A. (1997). The comprehension of metaphors in
schizophrenia. Psychopathology, 30, 149-154.
Dennis, M., Lazenby, A. L., & Lockyer, L. (2001). Inferential language in high-function children with
autism./oMrwa/ of autism and developmental disorders, 3, 47-54.
Dirven, R. (2002). Metonymy and metaphor: Different mental strategies of conceptualisation.
In R. Dirven & R. Pörings (Eds.), Metaphor and Metonymy in Comparison and Contrast
(pp. 75-111). Berlin/New York: Mouton de Gruyter.
Dunn, L., Dunn, L., Wlietton, C, & Buriey, J. (1997). British Picture Vocabulary Scale II.
Windsor: NFER-Nelson.
Development of metaphor and metonymy 559
Feyaerts, K. (1999). Metonymie hierarchies. Tlie conceptualisation of stupidity in (ierman
idiomatic expressions. In K.-U. Panther & G. Radden (Eds.), Metonymy in language and
thought (pp. 309-332). Amsterdam/Philadelphia. PA: John Benjamins.
Fris.son, S., & Pickering. M. J. (1999). The processing of metonymy: Evidence from eye
movements, foumal of Experimental Psychology: Leaming, Memory and Cognition, 25,
1366-1383.
Frisson. S.. & Pickering. M. J. (2001). (Obtaining afigurativeinitrprfUition of a word: Support for
underspecification. Metaphor and Symbol, 16, 149-171.
I'risson, S., & Pickering, M. J. (2007). The processing of familiar and novel senses of a word:
Why reading Dickens is easy but reading Needham can be hard. Language and Cognitive
Processes, 22, 595-613.
Gentner. D. (1988). Metaphor as structure mapping: The relational shift. Child Development, 59,
47-59.
(iibbs, R. W. (1984). Literal meaning and psychological theory. Cognitive Science, 8, 275-304.
Gibbs, R. W. (1990). Comprehendingfigurativereferential descriptions, foumal of Experimental
Psychology: I.eaming, Memory and Cognition, 10, 56-66.
Gibbs, R. W. (2002). A new look at literal meaning in understanding what is said and implicated.
Journal of Pragmatics, 34, 457-486.
(iiora. R. (1997). Understanding figurative and literal language: The graded .salience hypothesis.
Cognitive Linguistics, 8, 183-206.
Giora, R. (2002). Literal v.s.figurativelanguage: Different or equal? yoi/n;«/ of Pragmatics, 34,
487-506.
Glucksberg, S. (1991). Beyond literal meaning: The psychology of allusion. Psychological Science,
2, 146-152.
(irice, H. P (1975). Logic and conversation. In P Cole & J. Morgan (Eds.), Speech acts (pp. 41-58).
New York: Academic Press,
larrold, C:., & Brock. J. (2004). To match or not to match? Methodological issues in autism-related
resenKh. foumal of Autism and Developmental Disorders, 34, 81-86.
Jones, J.. & Stone, C. A. (1989). Metaphor comprehension by language learning di.sablcd and
normally achieving adolescent boys. Learning Disability Quarterly, 12, 251-260.
Karmiloff Smith. A. (1998). Development itself is the key to understanding developmental
disorders. Trends in Cognitive Science, 2, 389-398.
Karmiloff-Smith, A., Thomas, M, Annaz, D., Humphreys, K., Ewing, S., &Brace, N. (2(X)4). Exploring
the Williams Syndrome face processing debate: Tlic importance of building developmental
tnijectories.yoíz/v/í// of Child Psychology and Psychiatry, 45, 1258-1274.
Keysar, B. (1989). On the functional equivalence of literal and metaphorical interpretations in
discourse./oMmrt/ of Memoiy and Language, 28, 375-385.
Kcysar, B., Shen, Y L, Glucksbcrg, S., & Horton, W. S. (2000). Conventional language: How
metaphorical is it? foumal of Memory and Language, 43, 576-593.
Kircher. T T. J., Leube, D. T, Erb, M., Grodd, W, & Rapp, A. M. (2007). Neural correlates of
mctapbor processing in schizophrenia. Neurolmage, 34, 281-289.
Klcpousniotou. E.. & Baiim, S. R. (2007). Clarifying further the ambiguity advantage effect in WORI
recognition: Effects of aging and left-lieniisphert damage on the processing of homonymy and
polysemy. Brain and Language, 103, 8-249.
Kövecses, Z. (2002). Metaphor: A practical introduction. Oxford: Oxford University Press.
Likoff, G. (1987). Women,fireand dangerous things: What categories reveal about the mind.
Chicago, IL: Universit)' of CHiicago Press.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: University of Chicago Press.
MacKay, G., & Shaw, A. (2(M)4). A comparative study offigurativelanguage in children with autistic
spectrum disorders. Child Language Teaching and Therapy, 20, 13-32.
McElree, B., & Nordlie, J. (1999). literal andfigurativeinterpretations are computed in equal time.
Psychtmomic Bulletin & Review, 6, 486-494.
560
Gabrieiia Rundblad and Dagmara Annaz
Ozçaliskan, S. (2005). On learning to draw the distinction between physical and metaphorical
motion: Is metaphor an early emerging cognitive and linguistic capacity? Journal of Child
Language, 32, 291-318.
Pauwels, P. (1999). Putting metonymy in its place. In K. Panther & G. Radden (Eds.), Metonymy in
language and thought (pp. 255-272). Philadelphia, PA: John Benjamins.
Peirsman, Y, & Geeraerts, D. (2006). Metonymy as a prototypical category. Cognitive Linguistics,
17, 269-316.
Piaget, J. (1962). Play, dreams and imitation in childhood. New York: Norton.
PoUio, H., Barlow, J., Fine, H., & Pollio, M. (1977). Psychology and the poetics of grotvth:
Eigurative language in psychotherapy and education. Hillsdale, NJ: Erbaum.
Pynte, J., Besson, M., Robichon, F. H., & Poli, J. (1996). Tbe timecourse of metaphor
comprehension, an event-related potential study Brain and Language, 55, 293-316.
Rice, M. L. (2004). Growth models of developmental language disorders. In M. L. Rice &
S. E Waren (Eds.), Developmental language disorders: Erom phenotypes to etioiogies
(pp. 207-240). Mabwah, NJ: Erlbaum.
Riemer, N. (2001). Remetonymizing metaphor: Hypercategories in semantic extension. Cognitive
Linguistics, 12, 379-401.
Rosch, E., Mervis, C. B., Gay, W. D., Boyes-Braem, P., & Johnson, D. N. (1976). Basic objects in
natural categories. Cognitive Psychology, 8, 382-439.
Searle, J. R. (1979). Expression and Meaning. Cambridge: Cambridge University Press.
Seto, K. (1999). Distinguishing metonymy from synecdoche. In K.-U. Panther & G. Radden (Eds.),
Metonymy in ianguage and thought (pp. 91-120). Amsterdam and Philadelphia, PA:
John Benjamins.
Shen, Y (1992). Metaphor and categories. Poetics Today, 13, 771-794.
Smith, J. W. A. (1976). Children's comprehension of metaphor: A piagetian interpretation.
Language & Speech, 19, 235-243.
Stringaris, A. K., Medford, N. C, Giampietro, V, Brammer, M. J., & David, A. S. (2007). Deriving
meaning: Distinct neural mechanisms for metaphoric, literal, and non-meaningful sentences.
Brain and Language, 100, 150-162.
Thomas, M. S. C, Annaz, D., Ansari, D., Scerif, G., Jarrold, C, & Karmiloff-Smith, A. (2009). Using
developmental trajectories to understand genetic disorders, yowrw«/ of Speech Language and
Hearing Research, 52, 336-358.
Turner, N. E., & Katz, A. (1997). Evidence for the availability of conventional and of literal meaning
during the comprehension of proverbs. Pragmatics and Cognition, 5, 203-237.
Ullmann, S. (1967). The principles of semantics. Oxford: Blackwell.
Vosniadou, S. (1987). Contextual and linguistic factors in children's comprehension of nonliteral
language. Metaphor and Symbolic Activity, 2, 1-11.
Vosniadou, S., & Ortony, A. (1983). The emergence of the literal-metaphorical-anomalous
distinction in young children. Chiid Development, 54, 154-161.
Vosniadou, S., Ortony, A., Reynolds, R. E., & Wilson, P T. (1984). Sources of difficulty in
the yoimg child's understanding of metaphorical language. Child Development, 55,
1588-1606.
Waggoner, J. E., Messe, M. J., & Palermo, D. S. (1985). Grasping the meaning of metaphor: Story
recall and comprehension. Chiid Deveiopment, 56, 1156-1166.
Winner, E. (1979). New names for old things: The emergence of metaphoric language.yowm«/ of
Chiid Language, 6, 469-491.
Winner, E., Engel, M., & Gardner, H. (1980). Misunderstanding metaphor: What's the problem.
Journal of Experimental Chiid Psychoiogy, 30, 22-32.
Winner, E., Rosenstiel, A. K., & Gardner, H. (1976). The development of metaphoric
understanding. Developmental Psychology, 12, 289-297.
Received 2 March 2009; revised version received 7 May 2009
Development of metaphor and metonymy 561
Appendix A
An example of a metaphor story: flood (meaning 'lots of people)
1. Stuart works at a museum. The museum is in
2. It is a small museum and not so many people
the middle of town near a big river.
come to the museum.
t
m^
4. It is Monday morning and Stuart is at home.
3. Stuart's boss wants more people to come to
Suddenly, the phone rings - it is Stuart's boss.
the museum. So Stuart prepares a very special
Stuart's boss says. "You did it Stuart! There is a
exhibition. Stuart's boss tells lots and lots of
flood outside the museum." Stuart runs to the
people about Stuart's exhibition.
museum to look. What does Stuart see?
562
Gabrieiia Rundblad and Dagmara Annaz
An example of a metonymy story: Robbie Williams (meaning 'CD')
•
I
m.
I. Kate and Anne are listening to music in
2. Kate wants to play her favourite song to
Kate's room. Kate has a lot of CDs with songs
Anne. Kate looks for the CD with the song on.
on.
3. But Kate cannot find the CD.
4. After a while, Kate calls: "Come and look
She says. "Maybe my favourite CD is in another
Anne! I found Robbie Williams in the lounge".
room". Kate goes to look for the CD in the other
Anne goes to the lounge to look.
rooms. Anne stays in Kate's room.
What does Anne see?
Development of metaphor and metonymy 563
Appendix B
An example of prompts for a metaphor story :y7ooi/(meaning 'lots of people')
PROMPT Al
PROMPT Bl
Stuart's boss says: 'You did it Stuart! There
Tell me about the museum and the flood!
is a flood outside the museum.' Stuart runs to
the museum to look. What does Stuart see'?
PROMPT B2
Who is the flood'?
PROMPT A2
Stuart's boss says: 'There is a flood outside
the museum.' Tell me about the museum and
the flood!
PROMPT A3
Who is the flood?
An example of prompts for a metonymy story: Robbie Williams (meaning 'CD' )
PROMPT A1
PROMPT Bl
Kate calls: 'Come and look Anne! I found
Tell me about Kate and Robbie Williams!
Robbie Williams in the lounge'. Anne goes to
the lounge to look. What does Anne see?
PROMPT B2
What is Robbie Williams?
PROMPT A2
Kate calls: 'I found Robbie Williams in the
lounge.' Tell me about Kate and Robbie
Williams!
PROMPT A3
What is Robbie Williams?
Copyright of British Journal of Developmental Psychology is the property of British Psychological Society and
its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.