Glotta — Zeitschrift für Griechische und lateinische Sprache

Transcription

GLOTTA. Zeitschrift für griechische und lateinische Sprache
Begründet von Paul Kretschmer und Franz Skutsch
Herausgegeben von Michael Meier-Brügger und Gerhard Meiser
Im Beirat: Antonio Lillo Alcaraz (Murcia), Thomas Lindner (Salzburg), Witold
Manczak (Kraków), Georges-Jean Pinault (Paris), Brent Vine (Los Angeles),
Rudolf Wachter (Basel/Lausanne) und Andreas Willi (Oxford)
Manuskriptsendungen werden nur nach vorheriger Anfrage mit kurzer Zusammenfassung des Inhalts an folgende Anschriften erbeten:
Prof. Dr. Michael Meier-Brügger, Freie Universität Berlin, Seminar für Vergleichende und Indogermanische Sprachwissenschaft, Habelschwerdter Allee 45,
D-14195 Berlin; E-Mail: [email protected]
Prof. Dr. Gerhard Meiser, Martin-Luther-Universität Halle-Wittenberg, Orientalisches Institut, Seminar für Indogermanistik und Allgemeine Sprachwissenschaft,
Heinrich-und-Thomas-Mann-Str. 26, D-06108 Halle/Saale;
E-Mail: [email protected]
Manuskripte müssen die deutlich lesbare Anschrift des Verfassers tragen.
Sekretariat und Layout: Christiane Gambôa, Martin-Luther-Universität HalleWittenberg, Orientalisches Institut, Seminar für Indogermanistik und Allgemeine
Sprachwissenschaft, Heinrich-und-Thomas-Mann-Str. 26, D-06108 Halle/Saale;
E-Mail: [email protected]
Verlag: Vandenhoeck & Ruprecht GmbH & Co. KG, Theaterstr. 13, D-37073
Göttingen
Anzeigenverkauf: Ulrike Vockenberg
Internet: www.v-r.de
E-Mail: [email protected] (für Bestellungen und Abo-Verwaltung)
Das Werk und seine Teile sind urheberrechtlich geschützt. Jede Verwertung in
anderen als den gesetzlich zugelassenen Fällen bedarf der vorherigen schriftlichen Einwilligung des Verlages. Hinweis zu § 52a UrhG: Weder das Werk noch
seine Teile dürfen ohne vorherige schriftliche Einwilligung des Verlages öffentlich
gemacht werden. Dies gilt auch bei einer entsprechenden Nutzung für Lehr- und
Unterrichtszwecke.
Druck und Bindung: O Hubert & Co. GmbH & Co. KG, Robert-Bosch-Breite 6,
D-37079 Göttingen
© 2013 Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen
Printed in Germany
ISSN 0017-1298
ISSN 0017-1298 (E-Journal)
ipabo_66.249.78.36
Inhalt
Rutger J. Allan, Exploring Modality’s Semantic Space
Grammaticalization, Subjectification and the case of ὀφείλω . . . . . . . . . 1
Piotr Berdowski, In search of the lexical meaning of the Latin terms
cetarius and cetaria. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Sara Chiarini, Eine Anmerkung zu μελάνδετος. . . . . . . . . . . . . . . . . . . . . . . 62
Annemieke Drummen, A constructionist approach to the potential optative
in classical Greek drama. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Tristano Gargiulo, Un costrutto discusso: participio pro verbo di modo
finito? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Tom Keeline, Orthographicum quoddam: reccido . . . . . . . . . . . . . . . . . . . . 126
Martin Joachim Kümmel, Zur Verwendung des Perfektfuturs im Altlatein. . 130
Antonio Lillo, Ionic and Attic: on temporal constructions in Thucydides,
Xenophon and Plato . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Anika Nicolosi, Etimo, uso e significato di μενοεικής. . . . . . . . . . . . . . . . . . 170
Rafał Rosół, Ein linguistisches Problem mit Konkubinen. . . . . . . . . . . . . . . 178
Olga Spevak, La dislocation à droite en latin. . . . . . . . . . . . . . . . . . . . . . . . . 195
Jesús de la Villa, Πάλαι / τὸ πάλαι, πρίν / τὸ πρίν, παραχρῆμα /
τὸ παραχρῆμα . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Anssi Voitila, Deontic meaning of the auxiliary verb construction μέλλω +
INF in the Septuagint and Hellenistic Greek . . . . . . . . . . . . . . . . . . . . . . 242
Martin L. West, λυκάβας, λυκηγενής, ἀμφιλύκη . . . . . . . . . . . . . . . . . . . . . . 253
Benedicte Nielsen Whitehead, Negitanummius? Suggestion for an
emendation of Apuleius’ Metamorphoses 10.21.8. . . . . . . . . . . . . . . . . . 265
Klaus-Jürgen Grundner, Stichwörterverzeichnis . . . . . . . . . . . . . . . . . . . . . . 276
Exploring Modality’s Semantic Space
Grammaticalization, Subjectification and
the case of ὀφείλω
By RUTGER J. ALLAN, Amsterdam
Unsere Sprache kann man ansehen als eine alte Stadt: Ein
Gewinkel von Gäßchen und Plätzen, alten und neuen
Häusern, und Häusern mit Zubauten aus verschiedenen
Zeiten; und dies umgeben von einer Menge neuer Vororte
mit geraden und regelmäßigen Straßen und mit einförmigen
Häusern. (Wittgenstein, Philosophische Untersuchungen 18)
1. Introduction1
The semantic domain of modality is notoriously difficult to
define. It is associated with notions such as obligation, necessity, probability, possibility, permission, impossibility, etc., and
with grammatical categories such as indicative, imperative,
subjunctive, and optative. Modal meanings are expressed by
grammatical moods as well as lexical items such as modal
auxiliaries, modal adverbs and modal particles. Pinning down
the relationships among these multifarious expressions is bound
to be a complex enterprise. Most commonly, modality is defined
as a semantic category that expresses the attitude of the speaker
towards the content of his utterance (e.g. Bybee, Perkins &
Pagliuca 1994, 176). This definition, however, is too broad to
demarcate modality from other related semantic domains such
as tense and aspect in a clear and entirely satisfactory fashion.
Therefore, according to Bybee, Perkins & Pagliuca, ‘it may be
impossible to come up with a succinct characterization of the
notional domain of modality and the part of it that is expressed
__________
1
I wish to thank Gerard Boter and Albert Rijksbaron for their valuable
comments on an earlier version of this paper.
Glotta 89, 1–46, ISSN 0017–1298
© Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen 2013
ipabo_66.249.78.36
2
Rutger J. Allan
grammatically’. Instead, they claim that mood and modality can
be best understood as ‘a set of diachronically related functions’
(Bybee, Perkins & Pagliuca 1994: 176). This means, in other
words, that a real understanding of the semantic domain of
modality can only emerge from an analysis of these diachronic
relations. In this paper, I would like to examine whether such a
fundamentally diachronical approach will also provide a better
understanding of modality in Ancient Greek. A key role in a
diachronical analysis of modality is reserved for the notion of
semantic map. A semantic map can be thought of as a representation (a ‘road map’) of the network of diachronic relations between various polysemic meanings.2 On this approach, diachronic semantic developments are used as indicators of the way in
which a complex network of family resemblances is structured.
If meaning A historically extends to meaning B, it is assumed
that meanings A en B are also similar from a a synchronical
point of view. This approach is based on the assumption that
diachronic changes in the distribution of a linguistic form follow
connected paths in conceptual space (Croft 2001, 105).3
This paper will consist of two parts. In the first part, I will
present a case study regarding the verb ὀφείλω ‘owe, must’ and
its diachronic development. In the second part, an attempt will
be made to draw a map representing the semantic space of
verbal modality in Greek. In this map, all types of Greek verbal
modal expressions (full lexical verbs, auxiliaries, moods) are
located with reference to a number of major semantic dimensions such as deontic vs. epistemic modality, event-oriented modality vs. speaker-oriented modality and scale of modal
strength. These notions will be described in more detail below.
__________
2
Successful examples of the semantic map approach to modality are the
typologically-oriented studies of Bybee, Perkins & Pagliuca (1994) and Van
den Auwera & Plungian (1998). For an overview of this approach, see De
Haan (2006). In Allan (2003), I have taken a similar approach to analyse the
polysemic structure of the middle voice in Ancient Greek.
3
This assumption is essentially a diachronic version of Croft’s Semantic
Map Connectivity Hypothesis: ‘[L]anguage-specific and construction-specific
grammatical categories should map onto connected regions in conceptual
space’ (Croft 2001, 96).
3
In addition, the map will indicate historical relationships between the various modal expressions. The most suitable verb to
illustrate the way in which diachronic semantic developments
can be used as evidence for modality’s semantic map in Ancient
Greek is ὀφείλω. Ὀφείλω is unique in that it, throughout the
history of Greek, shows almost all conceivable semantic
changes in the domain of modality. It develops from a lexical
verb ‘owe’ to a deontic modal auxiliary ‘must’. It acquires an
epistemic meaning. Its past tense ὤφελ(λ)ον came to be used in
counterfactual assertions ‘should (have)’, and, finally, the verb
developed a function similar to a mood as it came to mark
counterfactual wishes ‘if only...’. This wide range of diachronically related meanings of ὀφείλω, make the verb the best
vehicle for a cartographic exploration of modality’s semantic
space. Before I turn to ὀφείλω, I will elaborate on a number of
widely-recognized semantic categories which can be thought of
as constituting the various dimensions of modality’s semantic
space: (1) deontic vs. epistemic modality, (2) event-oriented
modality vs. speaker-oriented modality and (3) the scale of
modal strength. Next, I will introduce two concepts that play a
central role in my approach of the semantics of modality. These
are grammaticalization and subjectification.
Before embarking on an exploration of Greek modality it is
important to discuss a number of major modal categories. A
crucial distinction within the domain of modality is that between
deontic and epistemic modality. Deontic modality is concerned
with notions such as obligation and permission (Palmer 2001, 9–
10, Traugott & Dasher 2002, 105–6, Nuyts 2006, 4–5).
‘[D]eontic modality is concerned with the necessity or possibility of acts performed by morally responsible agents’, as Lyons
(1977, 823) puts it. Another characteristic of deontic modality
noted by Lyons is that it ‘typically proceeds, or derives, from
some source or cause. (...) It may be some person or institution
to whose authority he [i.e the person which is obliged to perform
some act, RJA] submits’. Epistemic modality is defined by
Bybee, Perkins & Pagliuca (1994) as follows: ‘Epistemic
ipabo_66.249.78.36
4
Rutger J. Allan
modality applies to assertions and indicates the extent to which
the speaker is committed to the truth of the proposition’ (see
also Palmer 2001, 8–9, Traugott & Dasher 2002, 106–7, Nuyts
2006, 6). Unlike deontic modality, epistemic modality is
concerned with beliefs and knowledge.
Another important distinction which is often mentioned in the
literature on modality is the idea of a scale of modal strength
(Traugott & Dasher 2002, 117–8). Modal expressions are
associated with a certain degree (‘strength’) of obligation or
likelihood. A common distinction made in the domain of
modality is that between modal expressions which indicate
necessity and those which are associated with possibility.
Modals which can convey both deontic and epistemic meanings
typically show a relationship between the degree of obligation
and the degree of likelihood which they express. Thus, Bybee,
Perkins & Pagliuca (1994, 195), note on the basis of crosslinguistic evidence that modal auxiliaries which, in their deontic
reading, express strong obligation typically convey inferred
certainty in their epistemic reading. With regard to Ancient
Greek, I would like to propose a distinction between four
degrees of modal strength:4 (1) REALIS: This type of modality
relates to actual events, i.e. events that are presented as having
actually occurred in the past or as actually occurring in the
present (Langacker 1991, 242–3). Realis represents the maximum degree of modal strength.5 (2) NECESSITY: deontic necessity (obligation) involves an strong external (social) condition which compels an agent to complete an action (Bybee,
Perkins & Pagliuca 1994, 177). The epistemic counterpart of
obligation in Ancient Greek relates to the region on the scale of
modal strength between reality and mere possibility. The
__________
4
A similar scale of modality in Greek has been proposed by Méndez
Dosuna 1999, 343–4. See also Tichy 2006, 304–5 (Indo-Iranian and Greek).
5
Realis is typically expressed by the indicative in Greek, especially in
main clause assertions. It be noted, however, that the indicative does not
always signal realis modality. Non-realis uses of the indicative are: the
counterfactual uses of the secondary indicative and the neutral use of the
indicative in conditional protases.
5
speaker conceives the realization of the event as necessary or at
least ‒ implying a weaker commitment ‒ probable. This type of
modality is called projected reality by Langacker (1991, 278):
the momentum of the course of events is conceived as strong
enough such that the future course of reality can be projected
with confidence. Only an extraordinary or unforeseeable event
could prevent projected events from being realized. (3)
POSSIBILITY: In the case of deontic possibility an agent is
granted permission to perform an action. Epistemic possibility
indicates that the proposition is possibly true. (4) COUNTERFACTUALITY: Counterfactual modality pertains to propositions which the speakers knows to be not true, i.e. events that
have not been/are not being realized.
The last major distinction in the domain of modality concerns
the orientation of the modal expression. Event-oriented modality
relates to ‘the existence of internal or external conditions on an
agent with respect to the completion of the action (...)’ (Bybee,
Perkins & Pagliuca 1994, 177).6 With speaker-oriented modality, the internal or external conditions of the agent are
(presented as) irrelevant or non-existent. Only the will of the
speaker or his/her personal evaluation of reality is referred to.7
The contrast between event- and speaker-oriented modality
should be thought of as a semantic continuum (Narrog 2005,
687). It can be seen as a difference between more objective and
more subjective modality (Traugott & Dasher 2002, 133–6).
Event-oriented modality appeals to objective (‘sociophysical’)
properties, norms and regularities, whereas speaker-oriented
__________
6
Bybee, Perkins & Pagliuca actually use the term agent-oriented, which
is less felicitous because it implies that the participant has the semantic role
of agent (Van den Auwera & Plungian 1998, 83). The term event-oriented,
furthermore, makes clear that the source of modal force may also be located
in the external conditions of the grammatical subject.
7
Bybee, Perkins & Pagliuca (1994, 179) appear to distinguish only
deontic types of speaker-oriented modality (imperative, prohibitive, optative,
hortative, admonitive, permissive). In my view, also epistemic modal expressions can be speaker-oriented in that they may be grounded purely on the
subjective beliefs of the speaker without reference to objective knowledge of
the state of affairs.
ipabo_66.249.78.36
6
Rutger J. Allan
modality is based solely on the subjective will, beliefs and
opinions of the speaker.
Crucial to the analysis of semantic change in the domain of
modality is grammaticalization. Grammaticalization is ‘the process whereby lexical items and constructions come in certain
linguistic contexts to serve grammatical functions, and, once
grammaticalized, continue to develop new grammatical functions’ (Hopper & Traugott 2003, xv). The study of processes of
grammaticalization has significantly improved our understanding of historical developments in the domain of modality
by revealing a number of wide-spread tendencies. For example,
there is a significant cross-linguistic tendency for deontic and
agent-oriented meanings to develop into epistemic ones, for
agent-oriented meaning to change into speaker-oriented meanings,
and for participant-internal meanings to develop into participantexternal meanings.8 The work on grammaticalization has also
revealed the fundamentally gradual character of grammatical
change. In grammaticalization theory, the image of a cline is
used to characterize this process. The particular cline which is
most relevant for the grammaticalization of modals is the cline
full verb > auxiliary > verbal clitic > verbal affix. The various
stages on this cline should not be taken as discrete categories but
rather as parts of a continuum.9 It should be stressed that
linguistic items do not necessarily go through all stages of the
cline. The grammaticalization process may well start or –
unpredictably – come to a halt in the middle of the cline. In
addition to the verb-to-affix-cline, there is another type of
grammaticalization relevant to the Greek modal system. This
type regards the obligatory use of the moods in certain types of
subordinate clauses. The development of subordinate uses of the
moods is usually the last stage of the grammaticalization chain
__________
8
See e.g. Traugott (1989), Traugott and Dasher (2002), Traugott and
Hopper (2003), Heine, Claudi & Hünnemeyer (1991), Bybee, Perkins &
Pagliuca (1994), Van den Auwera & Plungian (1998) and Narrog (2005).
9
See, for example, Hopper & Traugott (2003, 111). A gradual view on
(the grammaticalization of) auxiliaries can also be found in Kuteva (2001)
and Anderson (2006), two important recent studies on auxiliation.
7
(Bybee, Perkins & Pagliuca 1994, 213, De Haan 2006, 33,
Nuyts 2006, 8). This means that, potentially, a process of grammaticalization within the domain of modality may follow a path
from a full lexical content-verb through a modal auxiliary, to a
grammatical (inflectional) mood, which eventually is used
obligatorily in certain types of subordinate clauses.
A notion which is closely connected with grammaticalization
is subjectification. Subjectification is described by Traugott as a
process whereby ‘(...) meanings become increasingly based in
the speaker’s subjective belief state/attitude toward the proposition’ (Traugott 1989, 35); According to Langacker, ‘Subjectification is a semantic shift or extension in which an entity
originally construed objectively comes to receive a more subjective construal’ (Langacker 1991, 215). The notions subjectivity
and objectivity, as they are used by Langacker, pertain to the
construal relation between a conceptualizer and his conception,
in other words, between the subject and object of the conception. An entity is construed more subjectively when it is limited
to the subject role, while it is construed more objectively when it
is limited to the object role of the conception. In diachronic
change, subjectification exhibits unidirectionality, that is, the
meaning of a given form will show a tendency to develop from
more objective to more subjective, rather than vice versa. An
example of a semantic shift involving subjectification is the
development of the English while from a connective marking an
objective temporal relationship ‘during the time that’ to a
concessive conjunction ‘although’ signalling (unexpected) contrasts between propositions. This concessive meaning can be
seen as more dependent on the subjective attitude of the speaker.
It goes beyond the scope of this paper to give a detailed
discussion of the complex notion of subjectification. I will,
therefore, focus only on the relevance of subjectification for a
description of the semantic development of modal verbs.10
__________
10
Important discussions and applications of the notion of subjectification
(in its Langackerian formulation) can be found in Langacker (1990), Langacker (1991, 215–220), Langacker (2000, 297–309), Athanasiadou, Canakis
ipabo_66.249.78.36
8
Rutger J. Allan
The semantic development from the original lexical content
verbs to modal auxiliaries, and finally to grounding predication
(equivalent with grammatical mood) can be seen as a form of
subjectification (see Langacker 1991, 270–1, 2000, 307–9): the
speaker becomes gradually more involved in the conceptualization of the state of affairs. Subjectification in the domain of
modality primarily relates to the location of the source of modal
force (locus of potency). In the process of subjectification, the
source of potency gradually moves from its initial position in the
clausal subject to a position in the speaker: ‘the modal force
anchored by the subject becomes progressively more and more
aligned onto the conceptualization relation anchored by the
speaker’ (Achard 1998, 163). In the original ‘premodal’ lexical
verb, the clausal subject is the so-called locus or source of
potency, that is, the subject possesses a physical or mental force
that, when unleashed, tends to bring about the realization of a
certain event. Obviously, this event is potential rather than
actual. Examples of ‘premodal’ lexical verbs are: You want/
know how/are able to swim. The speaker is minimally involved
in the conceptualization of the state of affairs.
In the next stage of development, the verb typically changes
into a deontic modal auxiliary (see Langacker 1991, 270–1,
2000, 308). Now, the source of potency is no longer identified
with the grammatical subject; in other words, there is an
attenuation of subject control. In this stage, the source of
potency has shifted from the physical-mental domain to the
domain of social interaction. The source of potency may be
located in a social norm, a specific or generalized authority, or
the speaker. As the source of potency is not necessarily any
specific individual it is often highly diffuse and difficult to
identify. Consider, for example: You have to do your homework.
The source of potency may be the speaker (: - because I say so),
__________
& Cornillie (2006). Traugott has developed a different, but related, idea of
subjectification (see e.g. Traugott 1989, Traugott & Dasher 2002). In this
paper, I will be mainly concerned with Langacker’s conception of subjectification.
9
but may also be another instance such as a parent, the teacher, a
general social norm, school regulations, an internal sense of
duty, etc. In other words, the source of potency is often conceptualized implicitly and ‘offstage’. The transfer of the source of
potency from the objective physical domain to a more diffuse
social domain makes that the speaker is more involved in the
conceptualization of the event. This transfer can therefore be
considered as a process of subjectification: the role of the
conceptualizing subject in the conception of the source of
potency has become more prominent.11 Not only the source of
potency can be rather diffuse, but also target of potency, that is,
the individual at which the modal force is directed. The
grammatical subject is not necessarily the target of potency any
more. For example, in an expression such as This fence must be
painted by tomorrow the modal force ‘is simply directed toward
realization of the event, to be apprehended by anyone who might
be in a position to respond to it’ (Langacker 2000, 308).
As for epistemic modal auxiliaries, they are diffuse both in
regard to the source and to the target of potency (Langacker
2000, 309). The potency involved in epistemic modals are
inherent, according to Langacker, in the evolutionary momentum of reality as it is assessed by the conceptualizer. The
conceptualizer (speaker) assesses what the likelihood is that
reality evolves in such a way that the target event is realized.
The conceptualizer makes this assessment on the basis of how
reality has been evolving up through the present. In Langacker’s
words: ‘(...) the conceptualizer carries out a mental extrapolation
of ongoing reality, projecting into the future, and senses the
degree of force impelling this mental extrapolation in the
envisaged direction, or the degree of resistance encountered in
projecting it through the target’ (Langacker 2000, 309). This
__________
11
It should be noted that subjectification is not a form of replacement of
an objectively construed entity by a subjectively construed one. Rather, the
subjective component is already immanent in the objective conception, and it
remains when the latter fades away (Langacker 2000, 298). In other words,
subjectification is a process of semantic ‘bleaching’.
ipabo_66.249.78.36
10
Rutger J. Allan
means that the conception of a force (or, source of potency) is
not inherent in the physical or social domain any more, but is
located in the mental activity of the conceptualizer. This force,
therefore, is subjectively construed in a strong sense. As a consequence, the development from deontic modal auxiliaries to
epistemic ones can also be seen as a process of subjectification.
Finally, modals can become grammaticalized into grounding
predications. Grounding is defined by Langacker as follows:
‘An entity is grounded when its location is specified relative to
the speaker and hearer and their spheres of knowledge. For
verbs, tense and mood ground an entity epistemically; (...)’
(Langacker 1987, 489). In other words, a grounding predication
(tense or mood) locates the state of affairs in a particular
epistemic region. Grammatical moods are typically grounding
predications.12 An essential conceptual difference between typical modal auxiliaries and grounding predications concerns the
locus of potency. Auxiliaries typically have their locus of potency in the external (social or physical) circumstances of the
subject. Modal grounding predications have their source of
potency in the ground, more specifically, in the speaker. The
speaker is thus maximally involved in the conceptualization of
the state of affairs. Grounding predications are speaker-oriented
modal expressions. In Ancient Greek, the grammatical mood
inflections, i.e. the indicative, imperative, subjunctive and optative, can be identified as modal grounding predications.13
__________
12
In English, according to Langacker, also typical modal auxiliaries such
as can, may, must, will are grounding predications. The German modal
auxiliaries können, müssen, mögen, wollen (which are less grammaticalized)
are denied the status of grounding predications by Langacker (1991, 271).
Mortelmans (2006) rightly takes a more gradual view of this distinction.
13
Cf. Achard (1998, 2002), who argues that mood (indicative, subjunctive, conditional) functions as a clausal grounding operator in French.
11
2. Grammaticalization and subjectification: the case of ὀφείλω
In this section, I will describe in more detail the four major
diachronic stages through which ὀφείλω developed from a
purely lexical verb denoting an objective property (Stage I:
‘owe’), via an event-oriented modality referring to a socialpsychological norm (Stage II: ‘must’ and Stage III: ‘should
(have)’), to a speaker-oriented form of modality, expressing a
wish (Stage IV: ‘if only’).14 The semantic developments will be
analyzed in terms of subjectification: the development from
objective meanings to more subjective, speaker-based, meanings
(subjectification) and in terms of increasing grammaticalization.
In this connection, special attention will be payed to the use of
verbal aspect (both with respect to the auxiliary ὀφείλω and to
the dependent infinitive) and to changes in word order.
Stage I: Lexical verb ‘owe a debt’
It is clear that the verb originally meant owing a debt.15 This
meaning is well-attested (7 instances) in the Homeric epics. For
example,
(1) πολέσιν γὰϱ Ἐπειοὶ χϱεῖος ὄφειλον (Λ 688)
16
For to many the Epeians owed a debt.
The verb used with the sense ‘owe’ is a lexical word. Its
meaning refers solely to an aspect of the objective psychophysical world. This meaning is not modal, but, because it
develops into a modal verb, I call it premodal (cf. Van der
Auwera & Plungian 1998, Traugott and Dasher 2002). In its
original sense ‘owe’, ὀφείλω has an object noun (typically
__________
14
See also Ruiz Yamuza (2008a).
This meaning is already attested for Mycenaean Greek. For the etymological background of the verb, I refer to Chantraine, DE, Slings (1975), De
Lamberterie (1992) and Ruiz Yamuza (2008a).
16
The translations (sometimes adapted) are by Lattimore.
15
ipabo_66.249.78.36
12
Rutger J. Allan
χϱεῖος) and can be passivized.17 Whether or not it could also be
combined with an infinitive (e.g. referring to an act of paying) is
impossible to say for lack of evidence. But even if it could not
be construed with an infinitive, ὀφείλω would still be associated
implicitly with an activity: owing a debt implies, after all, that
one is obliged to pay it back. The subject of the verb ὀφείλω (the
one owing the debt) is identical to the subject of the implicit
action of paying back. This subject can be said to be the locus of
potency, that is, it is under the influence of a mental force which
is directed at the realization of the action in the sense that it is
his/her duty to pay back the debt. This action is, of course, not
actual but potential. This (implicit) reference to a potential
action, at which a force residing in the subject is directed, makes
the originally premodal verb a suitable candidate for evolving
into a modal verb.
Stage IIa: deontic necessity (obligation) ‘must, be obliged to’
The meaning ‘owe (a debt)’ grammaticalizes into a modal
auxiliary by a process of semantic bleaching.18 The preceding
act (typically the act of borrowing) which created the debt, and
the idea of repaying gradually disappear from the lexical
meaning of the verb. The element of obligation, present in the
original meaning ‘to owe’ develops into the abstract element of
deontic necessity ‘must’. As Lehmann graphically puts it:
‘Grammaticalization rips off the lexical features until only the
grammatical features are left’ (Lehmann 2002, 115). Because
the source of modality is connected to the physical, psychological or social circumstances of the grammatical subject, ὀφείλω
in Stage II can be viewed as an event-oriented modality. The
source of modality cannot be identified with the grammatical
__________
17
Five times with object χϱεῖος/χϱέως ‘debt’ (Λ 686, 688, 698, γ 367, φ
17); two times with ζωάγϱια and μοιχάγϱια (θ 332, 462).
18
In a similar way, the English modal verb ought was originally the past
tense of owe. German sollen, English shall developed from a verb skallan
‘owe’. Cf. also Latin debere.
13
subject anymore (as in Stage I), but it is transferred from the
objective psycho-physical world to the more subjectively
construed, social domain. As a consequence of this transfer, the
role of the conceptualizing subject (speaker) has become more
prominent. In other words, besides grammaticalization a process
of subjectification has taken place.
This use of ὀφείλω is not yet attested in Homer. However, it
must have existed already before Homer’s time since, as we
shall see, the development to Stage III is inconceivable without
Stage II. An example from Herodotus is the following:19
(2) [Herodotus about what the Argives say:]
ἐγὼ δὲ ὀφείλω λέγειν τὰ λεγόμενα, πείθεσθαι μὲν οὐ
παντάπασι ὀφείλω (Hdt. 7.152.3)
I must say what is told, but I don’t have to believe it at all.
The transition from the premodal use (stage I) to its auxiliary
use (Stage II) might have been brought about by a reinterpretation of a construction of the premodal verb with an infinitive,
possibly along the following lines. First, a final-consecutive
infinitive is added to the verb plus object: [ὤφειλον χϱέως]
[ἀποστήσασθαι] ‘[I owed a debt] [to pay back]’. Then, the
object of the main verb is reinterpreted as the object of the
infinitive: [ὤφειλον] [χϱέως ἀποστήσασθαι] ‘[I owed > had to]
[pay back a debt]’. In this construction the main verb became
reanalyzed as a verb expressing deontic necessity.20
__________
19
Other classical examples are: Hdt. 7.16.γ.1, E. Alc. 682, 712. In Pi. N.
2.6, the verb is used in an impersonal construction ὀφείλει δ’(...) δϱέπεσθαι
κάλλιστον ἄωτον ‘it is necessary that he will pick the finest bloom’.
20
Reanalysis may well be the most important mechanism in grammaticalization (see Traugott & Hopper 2003, 39–58). For the role of constructions
in grammaticalization, see Traugott (2003).
ipabo_66.249.78.36
14
Rutger J. Allan
Stage IIb epistemic necessity ‘must, be certain to’
In Classical Greek, the verb also develops an epistemic meaning.
The evidence for this use is however not very abundant. Only
one case that I have come across is convincing:
(3) [Xerxes to Artabanus:]
Εἰ δὲ ἐϱίζων πϱὸς πᾶν τὸ λεγόμενον μὴ τὸ βέβαιον ἀποδέξεις,
σφάλλεσθαι ὀφείλεις ἐν αὐτοῖσι ὁμοίως καὶ ὁ ὑπεναντία
τούτοισι λέξας (Hdt. 7.50.2)
But if you quarrel with all that is said and cannot put forth a secure
position, you must be mistaken just as he who holds the contrary
opinion.
The major semantic factor determining an epistemic reading (vs.
a deontic one) concerns the location of the modal force (see also
Heine 1995, 35). The force is not inherent in the physical or
social world any more, but it is located in the subjective mental
activity of speaker. The transition from deontic to epistemic
meaning, therefore, can be analysed as a form of subjectification. Xerxes is led to conclude that Artabanus is mistaken on the
basis of the available evidence given by the protasis. There is no
‘element of will’ present that has an interest in the (non-)realization of the event. Apart from the absence of a socio-physical
modal force in the world, there are a number of secondary
semantic properties pointing to an epistemic reading (1)
σφάλλεσθαι has no controlling agent, (2) σφάλλεσθαι is static
(‘be wrong/mistaken’), (3) σφάλλεσθαι does not occur later than
the time referred to by the modal verb. (In other words, the
infinitive is not dynamic, but declarative: the present stem
indicates that the state of affairs σφάλλεσθαι is simultaneous
with ὀφείλεις), (4) σφάλλεσθαι is factual.21
__________
21
For the properties associated with the distinction between deontic and
epistemic meaning, see Heine (1995).
15
The other cases of ὀφείλω that are sometimes interpreted as
being epistemic are in reality deontic. Consider, for example, the
following case:22
(4) [Greeks messengers to Locrians and Phocians:]
οὐ γὰϱ θεὸν εἶναι τὸν ἐπιόντα ἐπὶ τὴν Ἑλλάδα ἀλλ’
ἄνθϱωπον, εἶναι δὲ θνητὸν οὐδένα οὐδὲ ἔσεσθαι τῷ κακὸν ἐξ
ἀϱχῆς γινομένῳ οὐ συνεμείχθη, τοῖσι δὲ μεγίστοισι αὐτῶν
μέγιστα· ὀφείλειν ὦν καὶ τὸν ἐπελαύνοντα, ὡς ἐόντα θνητόν,
ἀπὸ τῆς δόξης πεσεῖν [ἄν] (Hdt. 7.203.2)
For it was not a god but a human being who was invading Hellas,
and there was not and there would never be any mortal to whom not
a mixture of evil was given at birth, and to the greatest men the
largest share. Therefore, also the one marching against them was
bound to fall from pride, because he was mortal.
This case should not be seen as epistemic because the modal
force behind ὀφείλειν is located in the socio-physical world, that
is, external to the speaker. More specifically, the modal force is
constituted by the cosmic order which determines that the
greatest men receive the largest portion of evil. According to
this ‘law’, the invader had to fall from pride. Two additional
semantic features are typical of a deontic meaning: the Aktionsart of πεσεῖν is dynamic (as opposed to static), and it refers to an
event which is posterior to the time referred to by ὀφείλειν. The
infinitive πεσεῖν is therefore a dynamic infinitive (as opposed to
declarative). On an epistemic reading, the aorist πεσεῖν (as a
declarative infinitive) has to be interpreted as anterior: ‘he must
have fallen’. The particle ἄν found in the mss. cannot be correct
since ἄν may only be combined with declarative infinitives
depending on verbs of saying or thinking (see K-G, I, 240).
Only the absence of a clearly identifiable controlling agent (the
source of modility is rather diffuse: ‘the Cosmic Order’, ‘the
__________
22
Powell’s lexicon glosses this use as ‘be certain to’. Also Ruiz Yamuza
(2008a, 79) interprets it as epistemic. She mentions a number of other
examples which are, in my view, not epistemic but deontic since they involve
a social convention or a universal rule of appropriateness and their dependent
infinitives are dynamic.
ipabo_66.249.78.36
16
Rutger J. Allan
Way Things Go’) makes it a less prototypical example of a
deontic expression. Agentivity is not a necessary semantic property for deontic expressions (Heine 1995, 35).
Stage III: counterfactual assertions ‘should (have)’
At Stage III, the past tense of the modal auxiliary expressing an
obligation has acquired a special counterfactual meaning. For
example,
(5)
μῆτεϱ ἐπεί μ’ ἔτεκές γε μινυνθάδιόν πεϱ ἐόντα,
τιμήν πέϱ μοι ὄφελλεν Ὀλύμπιος ἐγγυαλίξαι
Ζεὺς ὑψιβϱεμέτης· νῦν δ’ οὐδέ με τυτθὸν ἔτισεν· (Α 352–4)
Since, my mother, you bore me to be a man with a short life,
therefore Zeus of the loud thunder on Olympos should grant me
honour at least. But now he has given me not even a little.
Achilles complains to his mother Thetis that, although he is
destined not to reach old age, Zeus does not grant him to achieve
honour. The presence of the adverb νῦν in 354, marking as often
the transition to actual reality, makes it clear that we are dealing
with a counterfactual utterance. The source of the obligation is
located in a general norm. A short life should be compensated
with the achievement of great honour. Because the locus of modality is connected to the agent or to his/her physical, psychological or social circumstances, ὤφελλον/ὤφελον in Stage III
is to be viewed as an event-oriented modality.
The following examples are cases of counterfactual assertions
in which ὤφελλον/ὤφελον refers to a situation in the past. The
event expressed by the infinitive is no longer realizable at the
moment of speech.
(6) [Proteus to Menelaos:]
ἀλλὰ μάλ’ ὤφελλες Διί τ’ ἄλλοισί τε θεοῖσι
ῥέξας ἱεϱὰ κάλ’ ἀναβαινέμεν, ὄφϱα τάχιστα
σὴν ἐς πατϱίδ’ ἵκοιο πλέων ἐπὶ οἴνοπα πόντον (δ 472–4)
Well, you really ought to have gone aboard after offering fine victims
to Zeus and other gods, so you’d soonest reach your fatherland as
you sailed the wine-dark sea.
17
Proteus refers to the moment that the Greeks departed from
Troy. The optative ἵκοιο in the final subordinate clause confirms
that the main clause refers to a past state of affairs.
(7) μὴ ὄφελες λίσσεσθαι ἀμύμονα Πηλεΐωνα
μυϱία δῶϱα διδούς· ὃ δ’ ἀγήνωϱ ἐστὶ καὶ ἄλλως (Ι 698–99)
You should not have supplicated the blameless son of Peleus with
innumerable gifts offered. He is a proud man without this.
Obviously, the sending of the embassy to Achilles cannot be
undone. The embassy should not have taken place because after
the embassy Achilles has taken up an even more inconciliatory
attitude against Agamemnon.
(8) [Hera to Zeus:]
οὐκ ὄφελον Τϱώεσσι κοτεσσαμένη κακὰ ῥάψαι; (Σ 367)
Should I, in my wrath, not have woven sorrows for the men of Troy?
Hera cannot undo her evil plans anymore.
(9) [Antilochus to Achilles:]
ἀλλ’ ὤφελεν ἀθανάτοισιν
εὔχεσθαι (Ψ 546–7)
But he should have prayed to the immortal gods.
Antilochus states that Diomedes should have prayed to the gods
during the chariot-race.
Although we lack direct evidence, tense and aspect distinctions probably played a crucial role in the transition between
Stage II and Stage III. It is, therefore, important to discuss
matters of verbal aspect in greater detail. The verb ὀφείλω
initially denoted a stative situation ‘owe, be in debt’. Not
surprisingly, therefore, in the past tense we only find imperfects
(4 times) but no aorists in Homer. This stative semantics
probably persisted when its meaning bleached to an auxiliary
expressing deontic necessity ‘must, have to (+ inf.)’ (Stage II).
When the imperfect tense ὤφελλον/ὤφειλον of the modal
auxiliary (ὀφείλω II) was used, two sources of possible
ambiguity emerged. First, it was ambiguous as to whether the
ipabo_66.249.78.36
18
Rutger J. Allan
event referred to by the infinitive had eventually been realized
or not. This type of ambiguity is a general property of modal
verbs which has also been recognized in relation to other languages (see Bybee 1995). Modal verbs have in common the
semantic property that they do not imply the completion of the
event expressed by the infinitive with which they occur. A
second source of ambiguity which is generally connected with
the past tense of modal verbs relates to the issue whether or not
the modality is still in effect at the moment of speech, that is,
whether the addressee is still obliged to complete the action
referred to by the infinitive (given that it is still realizable):
imperfect ὤφελλον/ὤφειλον ‘I was obliged to do that (- and I
still am)’.23 The next step might have been that an invited
inference emerged implying that the infinitival event did not
take place: ‘I was obliged to do X (- but I did/could not)’. For
example, there may have been a condition that has not been met,
whereby realization has been blocked. This inference eventually
became generalized so that it expressed a counterfactual
assertion (Stage III): ‘I should have done X’. The ambiguity
mentioned above as to whether the modal force is still in effect
at the moment of speech brings about that the past tense can also
be used referring to the moment of speech: ‘I should do X’
(present counterfactual).24 Verbal aspect is also crucial to the
issue of the distribution between the variant forms ὤφελλον and
ὤφελον. Later on I will present evidence that the traditional
__________
23
Note that I assume that the counterfactual uses of ὀφείλω first emerged
in the imperfect form ὤφελλον/ὤφειλον. The semantic ambiguities (I. Infinitival event realized or not? II. Modal force still present at speech moment?)
associated with the imperfect can be linked to the non-completed aspectual
character of the imperfect tense. As an historical parallel of this development
we can point to the emergence of the imperfects ἔδει, ἐξῆν, ἐχρῆν, προσῆκε
(K-G, I, 204–6) as counterfactuals.
24
The English form should, originally the past tense of shall, went
through a similar type of development in that it also can be used to refer to
present contexts (see Bybee 1995). The English verb must even went a step
further. Originally, it was the past tense of moot but it came to be used as a
full-blown present tense.
19
view that ὤφελλον is an imperfect and ὤφελον is an aorist form
is correct (pace Ruijgh).
As we have seen, the modal meaning of the verb ὤφελλον/
ὤφελον at Stage III can be characterized as a counterfactual
assertion. The speaker states that the event designated by the
infinitive has not been realized or is not being realized, although
it should have been or should be. It is important to note that at
this stage there is also an invited inference that the situation of
realization of the infinitival event, according to the speaker, is
desirable. For example, in Α 352–4 the deontic necessity connected to ὄφελλεν primarily resides in Zeus’ moral obligation to
compensate for Achilles’ short-livedness by granting him honour. However, since Zeus’ granting honour would obviously be
favourable to Achilles, the hearer inferences that also Achilles
himself wishes it. It is this type of pragmatic inference that
gradually gives rise to a novel, speaker-oriented, meaning of
ὤφελλον/ὤφελον (Stage IV). The role of pragmatic invited
inference is a central aspect of Traugott’s approach to grammaticalization, the Invited Inferencing Theory of Semantic Change.
According to Traugott change occurs when utterance type implicatures, which arise on the fly in the context of strategic
interaction between speakers and hearers, become conventionnalized (semanticized) as part of the coded meaning. Invited
inferencing also leads to subjectification (Traugott & Dasher
2002, 34–40, Traugott & Hopper 2003, 71–98).
Stage IV: Counterfactual wish ‘If only...’
At this stage the pragmatic implication of desirability has
become generalized. Now ὤφελλον/ὤφελον has the illocutionary function of a wish. The locus of the deontic modality is not
the participant and his/her circumstances, as was the case at
Stage II and III, but only the speaker. In other words, at Stage
IV the verb indicates a speaker-oriented modality. This means
that the source of modality is subjectively construed to a
maximal degree. This stage is, therefore, the final stage of the
ipabo_66.249.78.36
20
Rutger J. Allan
subjectification process. Since the speaker is the source of
modality, ὤφελλον/ὤφελον in counterfactual wishes can be
considered a grounding predication (see Introduction), that is, its
function is identical to the function of a grammatical mood:
ὤφελλον/ὤφελον in counterfactual wishes can be compared with
the use of the optative mood in counterfactual wishes in Homer
and the secondary indicative mood in Classical Greek. In this
respect, the verb constitutes an exception in Ancient Greek since
the other modal grounding predications in Greek, the moods,
appear to be inflectional.25 Examples are:
(10) a. [Odysseus to Agamemnon:]
οὐλόμεν’ αἴθ’ ὤφελλες ἀεικελίου στϱατοῦ ἄλλου
σημαίνειν, μὴ δ’ ἄμμιν ἀνασσέμεν (Ξ 83–4)
Ruinous! I wish you directed some other, unworthy army, and were
not lord over us.
b. [Agamemnon to Nestor:]
ἀλλά σε γῆϱας τείϱει ὁμοίϊον· ὡς ὄφελέν τις
ἀνδϱῶν ἄλλος ἔχειν, σὺ δὲ κουϱοτέϱοισι μετεῖναι (Δ 315–6)
But age weakens you which comes to all; if only some other of the
fighters had your age and you were one of the young men!
In example (10a.) Odysseus expresses the wish that Agamemnon would be leader of another, less worthy, army. In (10b.),
Nestor is not able to engage in spear-fighting any more, since
old age is pressing hard on him. Agamemnon wishes that Nestor
be a young warrior again.
Besides having undergone subjectification, ὀφείλω in Stage
IV has also been affected by grammaticalization. An indication
of progressive grammaticalization is the emergence of a fixed
word order. In Lehmann’s terminology, its syntagmatic variability decreases. ‘The syntagmatic variability of a sign is the
ease with which it can be shifted around in its context’
__________
25
Ὤφε(λ)λον’s morphological behaviour also shows typical features of a
grounding predication: it does not have an infinitive nor different tense forms
such as a present or a future. In this respect, the verb is similar to the English
modal must, which Langacker analyses as a grounding predication (Langacker 1991, 271).
21
(Lehmann 2002, 140). Once ὀφείλω has become an auxiliary, it
begins to show a preference for the first position in the clause
(disregarding clause-initial prepositive words such as relative
pronouns, wish-particles, etc.).
Table (1): Fixation of ὀφείλω’s position in the clause
‘owe’
First position
Other position
0 (0 %)
7 (100 %)
Counterfactual
assertion
4 (57 %)
3 (43 %)
Counterfactual
wish
23 (76 %)
7 (24 %)
In Homer, the verb in its original meaning ‘owe’ (7 occurrences)
never appears in the first position of the clause. In its counterfactual assertive meaning (Stage III), however, the verb occurs
in first position 4 times out of 7 (57 %). In counterfactual wishes
(Stage IV), the verb appears in first position 23 times out of 30
(76 %).26 This preference for the first clausal position implies
that also the relative order of auxiliary and infinitive becomes
fixed: only in two cases out of 37, the infinitive preceeds
ὤφελλον/ὤφελον.
In the following table, all counterfactual uses (both assertions
and wishes) of ὤφελλον/ὤφελον have been collected.
__________
26
In most non-initial cases, ὤφελλον/ὤφελον is located in second position
after a constituent carrying pragmatic Focus.
ipabo_66.249.78.36
22
Rutger J. Allan
27
Table (2): distribution of ὤφελλον and ὤφελον
ὤφελλον
[14x]
ὤφελον
[23x]
Present
Unrealized but
realizable:
[4x] Α 353 [A], Τ 200
[P], Ζ 350 [P], Ξ 84
[P, P]
Unrealized and
unrealizable:
[6x] Κ 117 [P], Α 415
[P], Δ 315 [P, P], Ω
254 [Perfect], α 217
[P], δ 97 [P, P],
Past
Unrealized and unrealizable:
[10x] δ 472 [P], Τ 59 [A], Η
390 [A], Ρ 686 [A], Σ 19 [A],
Χ 481 [A], Ω 764 [A], θ 312
[A], ξ 68 [A], σ 401 [A],
Unrealized and unrealizable:
[17x] Ι 698 [P], Σ 367 [A], Ψ
546 [P], Γ 40 [P,A], Γ 173
[A], Γ 428 [A], Ζ 345 [P], Λ
380 [A], Σ 86 [P, A], Φ 279
[A], Χ 426 [A], β 184 [A], ε
308 [A, A], λ 548 [P], ν 204
[A], ξ 274 [A, A], ω 30 [A]
The aorist ὤφελον is used 17 times referring to a past state of
affairs and 6 times referring to the present; the imperfect
ὤφελλον is used with present reference in 4 cases, and with past
reference in 10 cases. In the following section I will address two
issues regarding verbal aspect. First, I will propose an explanation for the distribution of ὤφελλον and ὤφελον. Next, I will
examine the aspectual form of the dependent infinitive.
The traditional view states that ὤφελλον is an imperfect and
ὤφελον is an aorist form (see Chantraine, DE; Schwyzer, I, 746,
II, 308, 345).28 However, the question how we should explain
their textual distribution still remains unanswered. A difference
in temporal reference does not seem likely. As can be gained
from the table, ὤφελλον and ὤφελον can both refer to present
and past. Explaining the distribution by taking recourse to
metrical convenience, as de Lamberterie (1992) does, is not
__________
27
The underscored places refer to counterfactual assertions (Stage III).
Ruijgh (1971, 301) unconvincingly proposes to take ὤφελον as an
imperfect of an unattested verb *ὀφέλω (see also Monro 1891, 38). Apart
from the fact that we have to assume the existence of a verb ὀφέλω, it is not
very probable that two forms, ὤφελλον and ὤφελον, which behave in an
almost identical way, actually would be derived from two different verbs.
28
23
entirely satisfactory. I will argue, instead, that the occurrence of
ὤφελλον and ὤφελον can be accounted for by means of a
semantic distinction involving the realizibility of the state of
affairs referred to by the infinitive. Counterfactuality pertains to
unrealized (non-factual) states of affairs. However, it is important to keep in mind that unrealized is not identical to unrealizable.29 Let us consider examples (10a.) and (10b.) again.
In (10a.) we find an imperfect ὤφελλες, in (10b.) we find the
aorist ὄφελεν. I would like to propose that the imperfect
indicates that, although the event is not realized at the moment
of speech (due to some blocking condition), it is not precluded
that the event may still be realizable at some time in the future.
This ‘openness’ of the present-referring imperfect can be
ascribed to the inherent non-completed aspect of the present
stem. The aorist form, being associated with completedness,
expresses that the event designated by the infinitive is not
realizable any more at the moment of speaking because the
momentum which would have led to the realization of the event,
is lost. Thus, in (10a.), the imperfect ὤφελλες used by Odysseus
implies that it is not excluded that Agamemnon will be the
commander of another, unworthy army in the future. The aorist
in (10b.) makes it clear that Agamemnon’s wish for Nestor to be
young again is unrealizable, neither at the present moment, nor
anytime in the future. I would like to note that this specific
distinction is only valid in those cases in which ὤφελλον/
ὤφελον refer to the present time. When ὤφελλον/ὤφελον refer
to the past, the state of affairs has always already become unrealizable (see also table [2]). A clear indication that the present__________
29
This subtle but important distinction is often ignored in our grammars.
For example, Kühner-Gerth note with regard to counterfactual wishes that
realization of the event is not possible any more ‘die Erfüllbarkeit des Wunsches gehört der Vergangenheit an’ (K-G, 1, 206). This is incorrect. Whether
or not the desired event is realizable is often left indeterminate because it is
considered irrelevant. With respect to the optative in wishes, Chantraine
(1953, 214) therefore rightly remarks ‘(...) l’optative de souhait chez Homère
(...) s’emploie même si le souhait n’est pas réalisable, ou plutôt la question de
savoir s’il est réalisable ou non ne se pose pas’ (similarly, K-G, I, 228
Anm.1).
ipabo_66.249.78.36
24
Rutger J. Allan
referring imperfect ὤφελλον indeed relates to realizable state of
affairs concerns the fact that four out of five infinitives depending on present-referring ὤφελλον designate controllable events
(πένεσθαι, ἐγγυαλίξαι, σημαίνειν, ἀνασσέμεν). Controllability
points to future realizability. By contrast, none of 8 infinitives
construed with the present-referring ὤφελον is controllable. This
difference can hardly be coincidental. Let us consider the other
examples of imperfect ὤφελλον.
(11) [Achilles to Agamemnon:]
ἄλλοτέ πεϱ καὶ μᾶλλον ὀφέλλετε ταῦτα πένεσθαι,
ὁππότε τις μεταπαυσωλὴ πολέμοιο γένηται
καὶ μένος οὐ τόσον ᾖσιν ἐνὶ στήθεσσιν ἐμοῖσι. (Τ 200–2)
You should busy yourself rather at some other time about these
things, when there is some stopping point in the fighting, at some
time when there is not so much fury inside of my heart.
Agamemnon wants to have the promised gifts brought to
Achilles’ hut, and a boar for a sacrifice and a meal. Achilles,
however, is intended to go to battle to avenge the dead, not
tolerating any delay. The source of the deontic necessity is the
social obligation felt by Achilles to avenge the dead comrades
immediately. The event referred to by the infinitive πένεσθαι is
not to be realized in the present, according to Achilles. But he
presents it as still realizable at some other time in the future
(ἄλλοτε), whenever a pause of the war would come, and his fury
would have disappeared (ὁππότε ... ἐμοῖσι). I interpret the
subjunctives γένηται and ᾖσιν as purely future-referring. There
is no reason to assume that Achilles means that it should happen
repeatedly (iterative-distributive). In such cases, in my view, the
aorist ὤφελον (perfective/completed aspect) could not have been
used. According to Achilles, Agamemnon’s preparations should
not be carried out at the present moment because the present
circumstances make it undesirable – there is no pause in war and
he is furious about the death of so many Greeks by the hand of
Hector (Τ 201–2). Likewise, in example (12), which was already
given above (ex. 4), the event denoted by the infinitive is
presented as still realizable in the future. It is not precluded that
25
Zeus at some time in the future will grant due honour to
Achilles.
(12) μῆτεϱ ἐπεί μ’ ἔτεκές γε μινυνθάδιόν πεϱ ἐόντα,
τιμήν πέϱ μοι ὄφελλεν Ὀλύμπιος ἐγγυαλίξαι
Ζεὺς ὑψιβϱεμέτης· νῦν δ’ οὐδέ με τυτθὸν ἔτισεν· (Α 352–4)
Since, my mother, you bore me to be a man with a short life, therefore Zeus of the loud thunder on Olympos should grant me honour
at least. But now he has given me not even a little.
(13) [Helen to Hector:]
αὐτὰϱ ἐπεὶ τάδε γ’ ὧδε θεοὶ κακὰ τεκμήϱαντο,
ἀνδϱὸς ἔπειτ’ ὤφελλον ἀμείνονος εἶναι ἄκοιτις,
ὃς ᾔδη νέμεσίν τε καὶ αἴσχεα πόλλ’ ἀνθϱώπων.
τούτῳ δ’ οὔτ’ ἂϱ νῦν φϱένες ἔμπεδοι οὔτ’ ἄϱ’ ὀπίσσω
ἔσσονται· τὼ καί μιν ἐπαυϱήσεσθαι ὀΐω. (Ζ 349–53)
Yet since the gods had brought it about that these vile things must
be, I wish I were the wife of a better man than this is, one who knew
modesty and all things of shame that men say. But this man’s heart
is no steadfast thing, nor yet will it be so ever hereafter; for that I
think he shall pick the fruits.
Although she is the wife of Paris right now, Helen is not precluding the possibility of becoming the wife of a better man at
some time in the future. In line 353, she might already be
speculating about Paris’ death.
Unlike the imperfects above, the aorist ὤφελον always refers
to state of affairs that have become unrealizable. In most cases,
realization of the event is impossible simply because of the
natural limitations of human life. One cannot become younger,
live twice or be the son of another father.
(14) [Agamemnon to Nestor:]
ἀλλά σε γῆρας τείρει ὁμοίϊον· ὡς ὄφελέν τις
ἀνδρῶν ἄλλος ἔχειν, σὺ δὲ κουροτέροισι μετεῖναι (Δ 315–6)
But age weakens you which comes to all; if only some other of the
fighters had your age and you were one of the young men!
Agamemnon’s wish relates to the present moment. Nestor is not
able to engage in spear-fighting any more, since old age is
ipabo_66.249.78.36
26
Rutger J. Allan
pressing hard on him. Agamemnon’s wish for Nestor to be
young again is obviously unrealizable, both at the present
moment and at any time in the future.
(15)
αἴθ’ ἅμα πάντες
Ἕκτοϱος ὠφέλετ’ ἀντὶ θοῇς ἐπὶ νηυσὶ πεφάσθαι (Ω 254)
Would that all of you had been killed instead of Hector beside the
fast ships.
Hector’s death makes Priam’s wish unfulfillable.
(16) [Telemachus to Athena:]
ὡς δὴ ἔγωγ' ὄφελον μάκαϱός νύ τευ ἔμμεναι υἱὸς
ἀνέϱος, ὃν κτεάτεσσιν ἑοῖς ἔπι γῆϱας ἔτετμε. (α 217–8)
Would that I were the blessed son of some man whom old age came
upon among his possessions.
Telemachus’ wish pertains to his present situation. Since it is
obviously impossible to become the son of another man, his
wish is unrealizable. Again, the unrealizability of the wish is
expressed by the use of the aorist.
(17) [Menelaus to Telemachus:]
ἐπεὶ μάλα πολλὰ πάθον καὶ ἀπώλεσα οἶκον
εὖ μάλα ναιετάοντα, κεχανδότα πολλὰ καὶ ἐσθλά.
ὧν ὄφελον τϱιτάτην πεϱ ἔχων ἐν δώμασι μοῖϱαν
ναίειν, οἱ δ’ ἄνδϱες σόοι ἔμμεναι, οἳ τότ’ ὄλοντο (δ 95–8)
For I have suffered very much and lost a house, a very well-settled
one containing many good things. Would that I lived in the house
with only a third part of these, and that those men were safe who
perished then (...).
The loss of Menelaus’ earlier possessions is irreversable, as is
the loss of his men.
(18) [Nestor to Agamemnon: ‘I will not hide that Menelaos
sleeps.’]
σοὶ δ’ οἴῳ ἐπέτϱεψεν πονέεσθαι.
νῦν ὄφελεν κατὰ πάντας ἀϱιστῆας πονέεσθαι
λισσόμενος· χϱειὼ γὰϱ ἱκάνεται οὐκέτ’ ἀνεκτός. (Κ 116–8)
27
And he has given to you alone all the hard work. For now he ought
to be hard at work going to all the great men in supplication. This
need that has come is no more endurable.
There is no doubt about it that the aorist ὄφελεν refers to the
speech moment (cf. νῦν): ‘He should be labouring now’. The
aorist expresses that the event denoted by the infinitive
πονέεσθαι is presented by Nestor as not realizable any more.
Menelaos’ chance to do his duty is lost because he is asleep.
The next issue regarding verbal aspect concerns the dependent
infinitive. In table (2) above, the aspect form of the dependent
infinitive is marked either by an A (aorist) or a P (present).30
Upon closer examination of the data, a clear tendency can be
observed: if ὤφελλον and ὤφελον refer to the present moment
of speech, the present infinitive occurs predominantly (10 times
against only one aorist). When ὤφελλον/ὤφελον refer to a
moment in the past, the aorist is preferred (25 times against 7
presents). However, this tendency is only a side-effect of
another, more significant factor – the Aktionsart (lexical aspect)
of the verb. Present infinitives are almost always derived from
verbs which have an atelic lexical aspect, i.e. verbs of which the
state of affairs expressed by the verb stem does not reach a
natural endpoint (states and activities). Aorist infinitives, on the
other hand, show a strong tendency to be derived from telic
verbs, i.e. verbs of which the state of affairs designated by the
verb stem inherently reaches a natural endpoint (accomplishments and achievements).31 With state verbs, the present
infinitives denote ‘the continuing existence of the subject in the
__________
30
In some cases there are two infinitives dependent on ὤφελλον/ὤφελον.
The terms states, activities, accomplishments and achievements are
derived from the influential typology of state of affairs proposed by Vendler
(1957). My classification is based on Fanning’s lists of Greek verbs
categorized as to their (primary) Aktionsart. Fanning makes a subdivision
within the class of achievement verbs between climaxes and punctuals, which
is not relevant for my purpose. As a rule of thumb, verbs with telic lexical
semantics tend to occur in the aorist stem more easily, verbs with atelic
semantic in the present stem (for the latter, see Fanning 1990, 137, 145 and
Napoli 2006, 41–4). From a morphological perspective, roots with telic
meaning often have a root or thematic aorist form (Hoffmann 1970).
31
ipabo_66.249.78.36
28
Rutger J. Allan
condition indicated by the verb’ (Fanning 1990, 137); with
activity verbs the present infinitives expresses ‘a continuing
process as occurring, an action progressing without a termination being reached’ (Fanning 1990, 145). The aorist infinitive of
accomplishment verbs indicates ‘that a process occurs and runs
all the way to its termination or limit, at which it ceases’
(Fanning 1990, 153); the aorist infinitive of achievement verbs
focuses on the instantaneous termination of the event.
Table (3): distribution of present and aorist infinitives in relation to
Aktionsart
Atelic verbs
(states, activities)
Telic verbs
(accomplishments, achievements)32
ἀνάσσω → ἀνάσσεμεν [P] (Ξ 84)
εἰμί → ἔμεναι (Γ 40), ἔμμεναι [P]
(α 217, δ 97),
μετεῖναι [P] (Δ 316), εἶναι [P] (Ζ
350)
εὔχομαι → εὔχεσθαι [P] (Ψ 546)
ἔχω → ἔχειν [P] (Δ 316)
ἧμαι → ἧσθαι [P] (Α 415)
λίσσομαι → λίσσεσθαι [P] (Ι 698)
μένω → μεῖναι [A] (ν 204)
ναίω → ναίειν [P] (Σ 86, δ 97)
νικάω ‘be winner’ → νικᾶν [P]
(λ 548)
οἴχομαι → οἴχεσθαι [P] (Ζ 346)
πένομαι → πένεσθαι [P] (Τ 200)
πονέομαι → πονέεσθαι [P] (Κ 117)
σημαίνω ‘bear command’ →
σημαίνειν [P] (Ξ 84)
ἄγομαι ‘marry’ → ἀγαγέσθαι [A] (Σ 86)
αἱϱέω → ἑλέσθαι [A] (Λ 380)
ἀναβαίνω → ἀναβαινέμεν [P] (δ 472)
ἁνδάνω → ἁδεῖν [A] (Γ173)
ἀπονίναμαι → ἀπονήμενος [A] (ω 30)
γίγνομαι → γενέσθαι [A] (Ρ 686, Σ 19)
ἐγεινάμην → γείνασθαι [A] (θ 312)
ἐγγυαλίζω → ἐγγυαλίξαι [A] (Α 353)
ἐφέπω [πότμον] → ἐπισπεῖν [A] (ε 308,
ξ 274)
θνῄσκω → θανέειν [A] (Χ 426, ε 308, ξ
274)
καταφθίνω → καταφθίσθαι [A] (β 184)
κτείνω → κατακτάμεν [A] (Τ 59),
κτεῖναι (Φ 279) [A]
(ἀπ)όλλυμαι → ἀπολέσθαι [A] (Γ 40, Η
390),
ὀλέσθαι [A] (Γ 428, Ω 764, ξ 68, σ
401)
ῥάπτω → ῥάψαι [A] (Σ 367)
τίκτω → τεκέσθαι [A] (Χ 481)
As can be seen in the table, there are only two exceptions to the
rule that Aktionsart determines aspect form: aorist μεῖναι and
__________
32
The one perfect infinitive (πεφάσθαι, Ω 254) is left out of account.
29
present ἀναβαινέμεν.33 So we may safely conclude that the most
significant factor determining the aspectual form of the
infinitive is the inherent lexical semantics of the verb.
The Homeric contrast between imperfect ὤφελλον and aorist
ὤφελον disappeared in Classical Greek as the aorist form
became generalized at the cost of the imperfect. The spread of
the aorist might be explained as a type of markedness shift, that
is, the aorist form is generalized at the cost of the imperfect for
expressive reasons. The semantic feature of completedness conveyed by the aorist emphasizes that any possible future realization of the state of affairs is blocked for good, thus highlighting
the counterfactuality of the state of affairs.34 A similar development has been observed by Jespersen in colloquial English: ‘We
may sometimes, chiefly in colloquial speech, meet with a further
shifting, the ante-preterit being used not only of the past, but
also of the present time, simply to intensify the unreality irrespective of time. Thus we may say: “If I had had money enough
(at the present moment), I would have payed you,” and “I wish I
had had money enough (now) to pay you”’(Jespersen 1924,
266).
In Hellenistic Greek, the process of grammaticalization
continued with the result that the verb developed into an
indeclinable particle-like word, which was combined with a
finite verb, instead of an infinitive. Thus, in Call. Epigr. 19.1 we
find a generalized (impersonal?) 3rd person form: ὤφελε μηδ’
ἐγένοντο θοαὶ νέες ‘Would that there had been no swift ships’.
Elsewhere, the 1st person has been generalized (especially in the
New Testament), for example, ὄφελον ἐβασιλεύσατε (1Ep.Cor.
4.8) ‘Would that you had become king’ (Blass, Debrunner &
Rehkopf, 290). In NT-Greek, only augmentless forms occur
which might have been relics of the former participle (K-G, 1,
__________
33
It is, however, conceivable that the verbal root men- (originally) had a
telic Aktionsart ‘come to a halt’. This would explain the existence of the
morphologically marked present formation μίμνω (with reduplication) and
Toch. B mäsketär, A mäskatär (‹ mˆ-ské-).
34
For the phenomenon of markedness shift, see Dik 1997, 1, 44–47.
ipabo_66.249.78.36
30
Rutger J. Allan
207, Wackernagel 1926, 228–30, Schwyzer, II, 346, Ruiz
Yamuza 2008a, 28).
I come to the conclusion of my analysis of the history of
ὀφείλω. As we have seen, the emergence of new meanings of
ὀφείλω did not result in the loss of the older ones. Old and new
meanings remained to be used side by side. This diachronic
process of grammaticalization led to a synchronic polysemic
network which can be characterized as ‘layering’.35
(19) Layering: ὀφείλω as a polysemic network
owe (I)
must (IIa)
(deontic)
must (IIb) (epistemic)
should (have) (III)
(counterfactual assertion)
If only ... (IV)
(counterfactual
wish)
3. Towards a map of modality’s semantic space
Having followed the verb ὀφείλω in its journey through
modality’s space, I would like to present a sketch of how an
over-all map of modality’s semantic space in Ancient Greek
might look like. Obviously, the scope of this paper does not
allow me to flesh out the semantics of the modal verbs and
moods in a detailed and completely satisfactory way. This
section merely aims at pointing out a number of general
diachronic tendencies which can be observed in the history of
Ancient Greek. The map is intended as a starting point for future
research into grammaticalization and subjectification processes
in the domain of modality.
__________
35
Not that the path of grammaticalization is forked. This type of diachronic split is a common phenomenon which is sometimes called polygrammaticalization (see Traugott & Hopper 2003, 114–5).
31
(20) A map of modality’s semantic space in Ancient Greek
EVENT-ORIENTED
(‘PREMODAL’) LEXICAL
SPEAKER-ORIENTED
MODAL AUXILIARY
MAIN CLAUSE
VERB
ἔξεστι ‘it is allowed’
ἐκγίγνεται‘it is allowed’
ἐνδέχεται ‘it is allowed’
ὑπάϱχει ‘it is allowed’
Future-referring
SUBJ (Hom.)37
Futural SUBJ (+ἄν)
Iter.-Distr. SUBJ
(+ἄν)
Cupitive OPT
OPT in Final,
‘Fear’, Delib. Clause
OPT in Interr.
Oblique OPT
Pot., 38
Iter. OPT
ἐχϱῆν ‘must have’
Iterative past (+ἄν)
OPT (Hom.)
εἰ + Ind. II
Epistemic
OPT (Hom.)
Ind. II (+ἄν)
Counterfactuality
ὤφελ(λ)ον+Inf.
Cup.OPT (Hom.)
εἰ γάϱ/εἴθε+Ind. II
Deontic
ἔδει ‘should (have)’
ὤφελ(λ)ον ‘should have’
ἐχϱῆν ‘should (have)’
πϱοσῆκε ‘should (have)’
ἐξῆν ‘might have’
Epistemic
Pot. OPT (+ἄν)
Possible Reality
δύναμαι ‘be able, can’
ἐπίσταμαι ‘know’ > ‘can’
ἔχω ‘have’ > ‘be able’
ἱκανός εἰμι ‘be able’
οἶδα ‘know, can’
οἶός τε εἰμί ‘be able’
Final SUBJ
Deontic
(ἔξ)εστι *‘be (outside)’
ἐκγίγνεται ‘become out of’
ἐνδέχεται ‘accept’
ὑπάϱχω ‘begin’
‘Fear’ SUBJ
Epistemic
βούλομαι ‘want’
ἐθέλω ‘be prepared, want,
intend’
ἀνάγκη ‘it is necessary’
δεῖ ‘it is necessary’
ἔοικε ‘it is likely’
κινδυνεύω ‘be probable’
μέλλω ‘be probable’
ὀφείλω ‘must’
Imperative
Hort./Proh./Delib.
SUBJ
Projected Reality
ἀνάγκη ‘it is necessary’
δεῖ ‘it is necessary’
ἔοικε ‘it is fitting’36
ὀφείλω ‘must’
πϱέπει ‘be fitting’
πϱοσήκει ‘it is fitting’
χϱή ‘it is opportune’
Deontic
ἀνάγκη ‘force, constraint’
δέω ‘be needed’
ἔοικε ‘ressemble’
ὀφείλω ‘owe’
πϱέπω ’be distinguished,
ressemble’
πϱοσήκω ‘have arrived at,
belong to, concern’
*τὸ χϱή ‘need’
κινδυνεύω ‘run a risk’
SUBORDINATE
CLAUSE
The map (20) is structured according to the dimensions I have
discussed earlier. Before I will elaborate on the structure of the
map, I would like to note that the lines which separate the
__________
36
Cf. also the related construction εἰκός ἐστι.
Also combined with ἄν/κε.
38
Mostly in relative and conditional clauses. In Homer, also combined
with κε, but not with ἄν (see Ruijgh 1971, 298–9, Ruijgh 1992). The optative
in conditional clauses may at times also be seen as originally related to the
cupitive optative (Schwyzer, II, 323–4, 327). For the use of the moods in
conditional clauses, see Hettrich 1992.
37
ipabo_66.249.78.36
32
Rutger J. Allan
various categories are merely drawn for clarity’s sake. In reality,
the modal map should be viewed as a semantic continuum
without discrete boundaries. The horizontal dimension represents
the degree of grammaticalization of each modal expression
(lexical item vs. auxiliary vs. mood). The vertical dimension
relates to the degree of modal strength (projected reality vs.
possible reality vs. counterfactuality), and, secondarily, to the
distinction between deontic and epistemic meaning. Due to the
limitations of the two-dimensional written page the three deontic
(D) and the three epistemic rows (E) are depicted in alternating
fashion from the top down (D – E – D – E – D – E). In reality,
the three deontic rows should be thought of as directly
connected to one another, and so are the epistemic ones,
constituting a third dimension as can be seen in the following
schema (‘side view’ of the map):
D
D
D
E
E
E
The horizontal division of the table represents the various stages
of grammaticalization and subjectification. The least grammaticalized and subjectified are the verbs and nouns in the left-hand
column. These are ‘premodal’ full lexical predicates that assign
an objective physical or mental property to a participant or his
immediate situation. Verbs that refer to a participant’s internal
ability (δύναμαι, ἐπίσταμαι, ἔχω, ἱκανός εἰμι, οἶδα, οἷός τε εἰμί)
or willingness (βούλομαι, ἐθέλω) are sometimes called dynamic
modals (Palmer 2001, 9–10). The verbs δύναμαι ‘can’, (ἔξ)εστι
‘it is possible’, ἐνδέχεται ‘Id.’, ὑπάϱχει ‘Id.’ express root
possibility, i.e. they indicate that the ‘enabling conditions for an
agent to perform an act do not lie entirely in the agent, but also
depend on the external world’ (Bybee, Perkins & Pagliuca 1994,
192). I do not consider these verbs modal auxiliaries since they
do not indicate a subjective attitude towards the state of affairs,
but describe an objective physical or psychological situation
33
(Nuyts 2006, 17). However, the existence of this transitional
category of verbs makes it clear that the boundaries of
modality’s semantic space are not clear-cut.
The next column of the table contains the modal auxiliaries.
These verbs do not refer to objective physical or mental properties, but to the ‒ more subjective ‒ domain of human social
interaction or epistemic assessment.39 Modal auxiliaries in
Greek can be classified according to the most important modal
dimensions deontic vs. epistemic, and projected reality vs.
possible reality vs. counterfactuality. Grammaticalized modal
auxiliaries frequently acquire a deontic meaning first, and then
grammaticalize (and subjectify) to epistemic meanings. Examples of this diachronic pathway (see arrows) are ἀνάγκη
‘constraint, force’ (premodal) > ἀνάγκη (ἐστί) ‘it is necessary’
(deontic) > ‘it is necessary’ (epistemic), δέω ‘be needed’ (premodal) > δεῖ ‘it is necessary’ (deontic) > ‘it is necessary’
(epistemic),40 ὀφείλω ‘owe’ (premodal) > ‘must’ (deontic) >
‘must’ (epistemic). There is evidence that the development of
the verb ἔοικα has been slightly different. It appears that its
epistemic meaning was not based on the deontic meaning, but
evolved directly from the original meaning ‘resemble’ (Ruijgh
1986, 608–9).41 The impersonal deontic modal verbs ἔξεστι,
ἐκγίγνεται, ἐνδέχεται, πϱέπει, πϱοσήκει, ὑπάϱχει and χϱή (all
expressing appropriateness or permission) did not develop
epistemic uses.42 Two premodal verbs have evolved directly into
__________
39
I distinguish modal auxiliaries from their lexical predecessors primarily
by semantic criteria, i.e. in terms of degree of subjectification. It is, however,
important to stress once again that this criterion is a gradient one, and it
therefore does not allow us to draw clear-cut boundaries. A semantic definition is, of course, by no means the only possible definition of auxiliaries. It is
suited, however, for my purpose, that is, an analysis of modality’s semantic
space.
40
The epistemic use of δεῖ occurs only rarely, e.g. Hdt. 7.51.2, Pl. Prm.
163e4, X. Hell. 7.4.36. For the use of δεῖ, see Goodell 1914, Bernadete 1965,
Ruiz Yamuza 2008a.
41
For a semantic analysis of the ἔοικα and its cognates, see Ruijgh 1986.
42
The epistemic use of counterfactual ἐχϱῆν is rare, e.g. Hdt 2.20.3, Th.
2.51.2, Lys. 19.29. For the use of χϱή, see Goodell 1914, Redard 1953,
Bernadete 1965, Ruiz Yamuza 2008a.
ipabo_66.249.78.36
34
Rutger J. Allan
epistemic modals without going through a deontic stage:
κινδυνεύω43 and μέλλω.44 I have no evidence that auxiliary
verbs with epistemic meaning developed into deontic modals.
This unidirectional relation deontic meaning > epistemic
meaning is in keeping with the grammaticalization data of
Bybee, Perkins & Pagliuca 1994, 197–9 en Van der Auwera &
Plungian 1998, 111.
It is interesting to note, incidentally, that many modal auxiliaries in Greek are impersonal. Following Langacker (1991, Ch.
8), I would like to propose that impersonal verbs predicate
something about an abstract setting. In cases like English It is
raining, the abstract setting (referred to by it) can be identified
with the spatiotemporal environment. With impersonal modal
auxiliaries, however, the abstract setting is of an epistemic kind.
It can be identified as the range of knowledge which is potentially invoked as the basis for the propositional judgment in
question (Langacker 2004, 270–1). According to Langacker,
epistemic impersonal constructions such as ‘it is certain that’
defocus from the mental activity of any particular conceptualizer
(such as the speaker). Instead, they evoke a generalized conceptualizer. In other words, the state of affairs designated by an
impersonal construction can be seen as an objective entity which
is accessible to any conceptualizer who considers the matter. In
my view, this general accessibility ‒ or intersubjectivity (Nuyts
2006, 14) ‒ of impersonal states of affairs is due to the focus on
the role of the general knowledge of the overall (participantexternal) circumstances of the state of affairs. Specifically, these
circumstances can be either of a deontic character (a general
social norm or authority) or of an epistemic character (a general
regularity in the structure of the world and the course of events).
For example, a personal construction like χλιδᾶν ἔοικας τοῖς
__________
43
For κινδυνεύω as a modal verb, see Ruiz Yamuza 2008b.
The etymology of μέλλω is obscure. In Homer, the verb is used as an
epistemic modal auxiliary. In classical Greek, the verb has developed into a
future tense auxiliary. For studies on the meaning of μέλλω I refer to Basset
1979; Ruijgh 1985; Rijksbaron 20064; Wakker 2006, 2007.
44
35
παϱοῦσι πϱάγμασι ‘You seem to revel in you present circumstances’(A.Pr. 971) signals that the speaker himself (Hermes) is
led to the conclusion on the basis of Prometheus’ personal
(verbal) behaviour. By contrast, impersonal constructions such
as ὡς ἔοικεν ‘as it seems’ invoke the idea of a general
knowledge of the structure of the world on the basis of which
any person can judge the proposition likely to be true.
The impersonal auxiliaries designating counterfactual states
of affairs are all past tense forms. With the exception of aorist
ὤφελον, they tend to be imperfect forms (K-G, I, 204–6). As I
have argued in the case of ὤφελλον/ὤφελον, I would hypothesize that these past forms were initially restricted to past
counterfactuality. The feature of counterfactuality then extended
to the present through a process of generalized pragmatic implicature. In other words, the fossilized past tense form of the
counterfactual modals can historically be explained by their
origin as past-referring forms. From a synchronical point of
view, the interpretation of the past tense form of these verbs is
less straight-forward. An attractive explanation is to view the
Greek augmented verb forms as markers of epistemic distance.
In the prototypical case, epistemic distance is interpreted with
reference to the time-line (i.e. as temporal distance) and thus
specified as referring to past time. In special cases, however, the
augmented forms are interpreted as marking counterfactuality
(in which the abstract conception of distance is viewed as a
distance from reality).45
__________
45
In Langacker’s terminology, the past tense form constitutes a polysemous network, of which epistemic distance is the abstract schema, of
which past time-reference (prototype) and counterfactuality (extension) are
elaborations (see e.g. Langacker 1987, Ch. 10). The development from past
tense meanings to counterfactual meanings is a gradual process. Typically, a
form originally had a past tense meaning. Next, an invited inference arises
implying that the event is not only past tense but also counterfactual. Finally,
the form is also used in present contexts with a purely counterfactual
meaning. Schematically: PAST > PAST COUNTERFACTUAL > COUNTERFACTUAL. For the view of past tense as a marker of conceptual
distance, see also James (1982), Fleischman (1989).
ipabo_66.249.78.36
36
Rutger J. Allan
The next category to address is the most grammaticalized and
subjectified modal expressions, that is, the grammatical moods.
The moods are maximally subjectified: the speaker, as the locus
of potency, is maximally involved in the conceptualization of
the state of affairs. In other words, the speaker is construed as
responsible for the judgment regarding the desirability (deontic)
or likelihood (epistemic) of the occurrence of a certain state of
affairs. In the map, a distinction is made between main clause
and subordinate clause uses. This distinction can be viewed as a
difference in degree of grammaticalization. The obligatory use
of the moods in certain types of subordinate clauses constitutes
the last stage of the grammaticalization chain (see Bybee, Perkins & Pagliuca 1994). It is interesting to note that the boundary
between auxiliary and grammatical mood is not crossed very
often (the exception being ὤφελλον/ὤφελον).
In the region of necessity-probability (projected reality), the
hortatory, prohibitive and the deliberative subjunctive (negation
μή) can be identified as expressions of deontic modality.46 These
expressions imply the will of the speaker that the event be
realized.47 Linked to these deontic uses of the subjunctive are
two uses found in subordinate clauses: the subjunctive in clauses
dependent on verbs of fearing (K-G, 2, 391–3) and the subjunctive in final clauses (K-G, 2, 378–82). Both types of
subjunctive express the desirability of the state of affairs from
the point of view of the speaker.48 It seems justified to assume
that these uses in subordinate clauses are the product of grammaticalization of main clause uses. In Homer, we still find uses
__________
46
See K-G, 1, 219–21. The deontic function of the subjunctive is
equivalent to Chantraine’s ‘valeur de volonté’, whereas the epistemic function of the subjunctive can be compared to his ‘valeur d’eventualité’ (Chantraine 1953, 206 and elsewhere).
47
By the deliberative subjunctive, of course, the speaker’s will is
questioned (cf. Ruijgh 1971, 275).
48
See Ruijgh 1971, 275. In the case of subordinate clauses with verbs of
fearing the desirability is of course negated (μή).
37
in main clauses which are semantically closely related to
subordinate uses.49
The subjunctive in final clauses is also related to the
epistemic uses of the subjunctive. The way in which futurereferring subjunctive can be introduced in subordinate final
clauses can be observed in subordinate clauses with conjunctions that originally had non-final meaning, such as ἵνα ‘where’,
ὄφϱα ‘until’, ὡς ‘how’. At an earlier stage, the subjunctive was
used in these clauses in its purely epistemic future-referring
meaning. In example of this earlier usage may be Γ 130: δεῦϱ’
ἴθι ... ἵνα θέσκελα ἔϱγα ἴδηαι (‘come here [...] where you will
see wondrous deeds’). This type of relative clause subsequently
came to be reinterpreted, through a process of invited inference,
as a purpose clause while the locative aspect of ἵνα gradually
disappeared. As a result, the subjunctive became grammaticalized as an obligatory marker of final clauses.50 Another phenomenon which shows the relationship between the final subjunctive and the epistemic uses of the subjunctive is the incidental
occurrence of the particles ἄν and κε in final clauses in Homeric
Greek (Chantraine 1953, 270–1, Ruijgh 1971, 276).51
The subjunctive is also employed in a purely epistemic way.
In Homeric Greek, the subjunctive is employed in main clauses
as a marker of futurity (negation: οὐ).52 The subjunctive expresses that the speaker is committed to the future realization of the
state of affairs (projected reality). An example is ζ 201: οὐκ ἔσθ’
οὗτος ἀνὴϱ διεϱὸς βϱοτὸς οὐδὲ γένηται ‘There is no mortal man
__________
49
Examples are: ἀλλ’ ἄγε νῦν ἐπίμεινον, ἀϱήια τεῦχεα δύω ‘But come on,
wait, [in order that] I will put on my armor’ (Ζ 340); μή νυ τοι οὐ χϱαίσμῃ
σκῆπτϱον καὶ στέμμα θεοῖο (A 28) ‘[I fear that] the sceptre and the god’s ribbon will not help you’. For more examples, see Chantraine 1953, 207–8, 266,
269, Tichy 2006, 293–9, Willmott 2007, 256–6.
50
This process of reinterpretation is discussed more extensively by Chantraine 1953, 266–73 and Willmott 2007, 155–62.
51
The presence of these particles possibly emphasizes the consecutive
aspect of final clauses (cf. Ruijgh 1971, 276). For the consecutive meaning of
the particle, see Monro 1891, 285–6, Ruijgh 1992, 75.
52
See Goodwin 1889, 2–3, K-G, 1, 217–8, Ruijgh 1971, 275–6, 287,
Hettrich 1987, Willmott, 2007, 54–81.
ipabo_66.249.78.36
38
Rutger J. Allan
so slippery, nor will there be one’ (note the negation οὐδέ,
precluding a deontic reading). This future-referring use of the
subjunctive goes back to Proto-Indo-European (see e.g. Ruijgh
1992, 76, Sihler 1995, 592). Whether the future-referring
(epistemic) subjunctive or the hortatory (deontic) subjunctive is
the source meaning of the subjunctive is not clear.53 The distinction between the deontic and the epistemic uses of the subjunctive is a gradual one.54 The Homeric use of the 1st person subjunctive expressing the intention of the speaker can be viewed as
an intermediary type.55 With the hortative use, the intention-type
has in common that the speaker views the event as desirable;
with the purely epistemic use, it shares the element that the
realization of the event is presented as probable/expected. The
diachronical path obligation (cf. the hortatory use) > intention >
future (cf. the epistemic use) is well attested cross-linguistically
(Bybee, Perkins & Pagliuca 1994, 263).56 The subjunctive as a
marker of futurity can also be found in subordinate clauses
(futural subjunctive). In Homer, it is mostly combined with the
particles ἄν or κε; in Classical Greek always.57 Here, the subjunctive expresses that the speaker views the future realization
of the state of affairs as probable or, at least, very well possible
(prospective use: K-G, 2, 473, Schwyzer 1950, II, 310–3, Ruijgh
1971, 274–8, 283–8, Ruijgh 1991, 202, Wakker 1994, 174).
__________
53
Stahl 1907, 234–6 makes an attractive case for the ‘voluntative’ (i.e.
deontic in my terminology) subjunctive as the original meaning (‘Grundbedeutung’). Recently, Tichy 2006 has argued for the expectative function of
the subjunctive. I will not go into the genesis of the subjunctive in PIE. I refer
the reader to Hahn 1953, Gonda 1956, Rix 1986, Strunk 1988, Tichy 2002
and 2006.
54
In Homer, the distinction between deontic and epistemic uses of the
subjunctive is less clear than in Classical Greek due to the non-obligatory
status of the particles ἄν and κε in the epistemic uses.
55
For this usage type of the subjunctive, see Chantraine 1953, 209,
Willmott 2007, 86–9.
56
For more examples, see Chantraine 1953, 209, Ruijgh 1971, 288, 1992,
76, Willmott 2007, 63–5. The relation between subjunctive and futurity
shows that the semantic space of modality is continuous with the domain of
tense.
57
For the Homeric distribution of the particles ἄν and κε, see Monro 1891,
327–34, Chantraine 1953, 345–50, Ruijgh 1992, Wakker 1994, 205–14.
39
Closely related to the futural subjunctive is the iterativedistributive (or generic) subjunctive (Ruijgh 1971, 278–80, 1991,
203). In Homeric Greek, this use is sometimes accompanied by
the particles ἄν or κε, in Classical Greek always (Chantraine
1953, 349–50, Ruijgh 1971, 286–88, 1992, 80), which is a sign
of grammaticalization. As with the futural subjunctive, the subjunctive can be said to express the speaker’s expectation that the
state of affairs will be realized (repeatedly). In the case of the
distributive-iterative subjunctive, the expectation is based on the
assessment that the (repeated) occurrence of the state of affairs
is in accordance with the way in which the world is generally
structured.
In conclusion, the subjunctive mood can be analysed as a
polysemy, i.e. a network of semantically related uses. It is,
furthermore, also possible to discern a common semantic element
(abstract schema) which is shared by all concrete uses of the
subjunctive. In all cases, the subjunctive expresses that the
realization of the event is viewed by the speaker as probable or,
at least, very well possible (cf. Kühner-Gerth, 1, 201: ‘Modus
der Erwartung’). The momentum of the course of events is
conceived as strong enough such that future reality can, to a
certain degree, be projected.
The optative in main clauses can have two functions: the
cupitive optative (deontic, negation μή) and the potential optative, which is often accompanied by ἄν or κε (epistemic, negation οὐ).58 In Homer, the optative is also used in main clauses
with a (past or present) counterfactual meaning. The past tense
(+ ἄν) begins to be used in counterfactual propositions referring
__________
58
Cupitive optative: K-G, 1, 226–8, Schwyzer, II, 320–4; potential optative: K-G, 1, 231–6, Schwyzer, II, 324–31. Both meanings go back to PIE. It
is not clear which of the two is the source meaning of the optative. Given the
fact that epistemic meanings in most cases develop from deontic meanings,
we may assume that the cupitive meaning is the oldest (see also Stahl 1907,
238–9 and Schwyzer, II, 320, with literature). For the semantic relatedness
between the cupitive and potential optative, see Ruijgh 1971, 293–4. For
ideas on the genesis of the optative in PIE. see e.g. Hahn 1953, Gonda 1956
and Tichy 2006.
ipabo_66.249.78.36
40
Rutger J. Allan
to the past.59 In Classical Greek, the counterfactual meaning of
the optative has yielded completely to the counterfactual use of
the past tenses (+ ἄν) both in assertions and in wishes.
The potential optative is also used in conditional and relative
subordinate clauses. The potential optative probably also gave
rise to a number of uses of the optative in other types of subordinate clauses (typically with a historical tense in the main
clause), such as (1) the iterative optative,60 (2) the optative in
indirect interrogative questions,61 (3) the optative occurring in
final clauses, (4) in clauses dependent on verba timendi and (5)
in indirect deliberative questions (see K-G, 1, 254; Chantraine
1953, 223–5). In these contexts, the optative indicates that the
state of affairs is merely imagined (K-G, 1, 253: ‘bloss vorgestellt (potential)’), that is, not expected such as a subjunctive
would imply (similarly, Monro 1891, 275). The optative in final
clauses, in complement clauses with verbs of fearing and in
indirect deliberative questions involves a representation of the
thought and attitudes of the subject of the main clause. These
uses of the optative appear to be related to the cupitive optative
since there is also an implication that the state of affairs is
desirable (Schwyzer, II, 323, Ruijgh 1971, 297). The increased
appearance of the optative in various types of subordinate
clauses, and especially the occurrence of the optative in indirect
questions, may have triggered the post-Homeric development of
the optative used in indirect speech (oblique optative), of which
__________
59
See Chantraine 1952, 226–7, Ruijgh 1971, 301–2, 1992, 81–2 (who
also gives an explanation for the emergence of the counterfactual past tense),
Wakker 1994, 210–4, Horrocks 1995, 162.
60
According to Schwyzer (II, 335–6), there is in essence no distinction
between the potential and iterative optative. The iterative meaning of the
optative shows that the semantic domain of modality borders on the domain
of aspectuality. Cf. also the habitual meaning of the secondary indicative
with ἄν. The relation between iterativity and non-reality may be that they
both do not pertain to a specific event occurring at a specific moment (for
which there is specific evidence).
61
For the potential optative in direct and indirect questions, see
Schwyzer, II, 327–8.
41
the original potential meaning has bleached away (K-G, 1, 255,
Ruijgh 1971, 298, Méndez Dosuna 1999).
In the region of counterfactual mood, we find a number of
different formations. In Homer, the optative can be used with
counterfactual meaning (mostly referring to the present). This
use disappears in Classical times. Already in Homer, the past
tense (+ ἄν/κε) is used as a marker of past counterfactuality. In
Classical Greek, the counterfactual use of the past tense is also
used in reference to the moment of speaking (K-G, 1, 212–3).62
Furthermore, the past tense combined with ἄν was used to refer
to past iterative events (K-G, 1, 211–2).
The domain of possible reality and the domain of counterfactuality constitute a semantic continuum. There are a number
of formal indications which confirm their semantic relationship:
(1) In Homer, both semantic notions can be expressed by means
of the optative. (2) In both domains, secondary endings are used.
In my view, secondary endings imply epistemic distance, as they
either refer to past, counterfactual or merely possible, state of
affairs. By contrast, primary endings are markers of epistemic
proximity (or, in more traditional terms, actuality) since they are
found in the present tenses (i.e. immediate reality) and in the
future and subjunctive (projected reality, i.e. events which are
expected to become realized provided that nothing unforeseen
intervenes). (3) The expansion of the particles ἄν/κε seems to
have roughly followed the scale of modal strength (vertical
dimension on map). Originally it was used with the future and
the subjunctive (projected reality). Next, it spread to the optative
(potential reality) and, finally, to the past tense with counterfactual meaning (non-reality).
The semantic distinction between the moods is ultimately
based on their epistemic implications, even when they are used
in a deontic sense. This means that the distinction between the
various types of deontic expressions is, in the end, dependent on
the degree of epistemic modal force. For example, the contrast
__________
62
The link between counterfactuality and past tense clearly shows that the
semantic space of modality constitutes a continuum with the domain of tense.
ipabo_66.249.78.36
42
Rutger J. Allan
between the hortative subjunctive and the cupitive optative is
not concerned with a difference in strength of the speaker’s will,
but with a difference in degree to which the speaker deems it
likely that the event will be realized.63
4. Summary and conclusion
Modality is a murky matter. A promising method to deal with
the complexities of modality in Ancient Greek is to take a
diachronical perspective. By analysing diachronical changes as
forms of grammaticalization and subjectification, it is possible
to understand and describe the syntactic and semantic changes
and their causes in a more insightful way. On the basis of an
inventory of the wide range of expressions of verbal modality in
Ancient Greek a map of modality’s semantic space can be
drawn, a map which represents both synchronical and diachronical semantic relationships between the various types of modal
meanings. The map demonstrates that there are forms of regularity in semantic change. Diachronical semantic changes turn out
to follow connected paths through semantic space and they tend
to be unidirectional. Major tendencies which can be observed
are: the regular change from event-based to more speaker-based
forms of modality (especially the development from premodal
verbs to auxiliaries), and the development of epistemic meanings
out of deontic meanings. Both types of semantic change are
forms of subjectification. What the semantic map approach
shows is that the complex network of family resemblances
within the domain of modality can be insightfully analyzed as a
__________
63
The fact that deontic meanings necessarily involve an epistemic element
implies that the diachronic development of purely epistemic meanings out of
deontic ones can be seen as a process of semantic bleaching: the deontic
aspect disappears, the remaining epistemic aspect comes to the fore. This
process is typical for subjectification: the subjective element does not replace
the object element, but the subjective element was there all along (see
Langacker 2000, 298).
43
coherent network of synchronically and historically related
meanings.
Bibliography
Achard, M. (1998): Representation of cognitive structures: Syntax and
semantics of French sentential complements, Berlin.
– (2002): ‘The meaning and distribution of French mood inflections’, in: F.
Brisard (ed.), Grounding: The epistemic footing of deixis and reference,
197–249, Berlin: Mouton de Gruyter.
Allan, R. J. (2003): The Middle Voice in Ancient Greek: A Study of Polysemy,
Amsterdam: Gieben.
Anderson, G. D. S. (2006): Auxiliary Verb Constructions, Oxford.
Athanasiadou, A., C. Canakis & B. Cornillie (eds.) (2006): Subjectification:
Various Paths of Subjectivity, Berlin-New York: Mouton de Gruyter.
Auwera, J. van der & V. A. Plungian (1998): ‘Modality’s semantic map’,
Linguistic Typology, 79–124.
Basset, L. (1979): Les emplois périphrastiques du verbe grec μέλλειν, Lyon.
– (1989): La syntaxe de l’imaginaire. Etude des modes et des négations
dans l’Iliade et l’Odyssée, Lyon.
Bernadete, S. (1965): ‘ΧΡΗ and ΔΕΙ in Plato and Others’, Glotta 43, 285–98.
Blass, F., A. Debrunner & F. Rehkopf (200118): Grammatik des neutestamentlichen Griechisch, Göttingen.
Bybee, J. L., R. Perkins, & W. Pagliuca (1994): The Evolution of Grammar:
Tense, Aspect, and Modality in the languages of the world, ChicagoLondon: University of Chicago Press.
– (1995): ‘The semantic development of past tense modals in English, in: J.
L. Bybee & S. Fleischman, Modality in Grammar and Discourse,
Amsterdam-Philadelphia, 503–518.
Chantraine, P. (1953): Grammaire Homérique, II: Syntaxe, Paris: Klincksieck.
Croft, W. (2001): Radical Construction Grammar, Oxford.
Dahl, Ö. (1997): ‘The relation between past time reference and counterfactuality: a new look’, in: Athanasiadou, A. et al. (eds.), On Conditionals
Again, Amsterdam, 97–114.
Dik, S. C. (19972): The Theory of Functional Grammar, Part 1, The
Structure of the Clause, Berlin-New York.
Fleischman, S. (1989): ‘Temporal distance: A basic linguistic metaphor’,
Studies in Language 13, 1–50.
ipabo_66.249.78.36
44
Rutger J. Allan
– (1995): ‘Imperfective and Irrealis’, in: J. L. Bybee & S. Fleischman,
Modality in Grammar and Discourse, Amsterdam-Philadelphia, 519–551.
Fortson, B. W. (2004): Indo-European Language and Culture, Oxford.
Frawley, W. (ed.) (2006): The Expression of Modality, Berlin-New York.
Gonda, J. (1956): The Character of the Indo-European Moods, Wiesbaden.
Goodell, T. D. (1914): ‘χϱή and δεῖ’, CQ 61–102.
Goodwin, W. W. (1889): Syntax of the Moods & Tenses of the Greek Verb,
London.
Haan, F. de (2006): ‘Typological approaches to modality’, in: Frawley, W.
(2006), 27–69.
Hahn, E. A. (1953): Subjunctive and optative: their origin as futures, New
York.
Heine, B., U. Claudi, & F. Hünnemeyer (1991): Grammaticalization. A
Conceptual Framework, Chicago.
Heine, B. (1995): Agent-Oriented vs. Epistemic Modality: Some
Observations on German Modals’, in: J. L. Bybee & S. Fleischman,
Modality in Grammar and Discourse, Amsterdam-Philadelphia, 17–54.
Hettrich, H. (1987): ‘Zur Entwicklung der Finalsätze altindogermanischer
Sprachen’, ZVS 100, 219–237.
– (1998): ‘Die Entstehung des homerischen Irrealis der Vergangenheit’, in:
J. Jasanoff e.a (eds.), Mír Curad: Studies in honor of Calvert Watkins,
Innsbruck, 261–270.
– (1992): ‘Lateinische Konditionalsätze in sprachvergleichender Sicht’, in:
O. Panagl & Th. Krisch (eds.), Latein und Indogermanisch. Akten des
Kolloquims der Indogermanischen Gesellschaft, Salzburg, 23.–26.
September 1986, 263–284.
Hoffmann, K. (1970): ‘Das Kategoriesystem des indogermanischen Verbums’, MSS 28, 19–24.
Hopper, P. J., & E. Closs Traugott (2003): Grammaticalization, Second
Edition. Cambridge: UP.
Horrocks, G. (1995): ‘On Condition...: aspect and modality in the history of
Greek’, Proceedings of the Cambridge Philological Society, 41, 153–73.
James, D. (1982): ‘Past Tense and the Hypothetical: A Cross-linguistic
Study’, Studies in Language 6, 375–403.
Jespersen, O. (1924): The Philosophy of Grammar, London.
Kuteva, T. (2001): Auxiliation. An Enquiry into the Nature of Grammaticalization, Oxford.
Lamberterie, C. de (1992): ‘Le problème de l’homonymie: les trois verbes
ὀφέλλω en grec ancien’, in: F. Létoublon (ed.), La langue et les textes,
Amsterdam, 201–17.
Langacker, Ronald W. (1987–1991): The Foundations of Cognitive Grammar, 2 Vols., Stanford: University Press.
45
– (1990): ‘Subjectification’, Cognitive Linguistics 1, 5–38.
– (2000): Grammar and Conceptualization, Berlin-New York.
– (2004): ‘Aspects of the Grammar of Finite Clauses’, in: M. Achard & S.
Kemmer (eds.), Language, Culture, and Mind, Berlin-New York, 535–
576.
Lehmann, C. (2002): Thoughts on Grammaticalisation, Second, revised
edition, (Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt 9), Erfurt.
Méndez Dosuna, J. V. (1999): ‘La valeur de l’optatif oblique en grec: un
regard fonctionel-typologique’, in: B. Jacquinod (ed.), Les complétives en
grec ancien, Saint-Étienne: Publications de l’université de Saint-Étienne.
Monro, D. B. (1891): Homeric Grammar, Oxford.
Mortelmans, T. (2006): ‘Langacker’s ‘subjectification’ and ‘grounding’: A
more gradual view’, in: Athanasiadou, Canakis & Cornillie (2006), 151–
176.
Napoli, M. (2006): Aspect and Actionality in Homeric Greek. A Contrastive
Analysis. Milan.
Narrog, H. (2005): ‘Modality, mood, and change of modal meanings’,
Cognitive Linguistics 16, 677–731.
Nuyts, J. (2006): ‘Modality: Overview and linguistic issues’, in: Frawley, W.
(2006), 1–26.
Redard, G. (1953): Recherches sur χρή, χρῆσθαι. Étude semantique, Paris.
Rijksbaron, A. (1988): ‘The discourse function of the imperfect’, in: A.
Rijksbaron, H. A. Mulder, G. C. Wakker (eds.), 1988, In the Footsteps of
Raphael Kühner, 237–54.
– (2006): The Syntax and Semantics of the Verb in Classical Greek. An
Introduction, Chicago.
Rix, H. (1986): Zur Entstehung des urindogermanischen Modussystems, IBS,
Vorträge und kleine Schriften 36. Innsbruck.
Ruijgh, C. J. (1971): Autour de “TE épique”, Amsterdam.
– (1985): Review Basset 1979, Lingua 65, 323–33.
– (1986): Review Συνοδίνου, ‘Ἔοικα-εἰκὸς καὶ συγγενικὰ ἀπὸ τὸν Ὅμηϱο
ὡς τὸν Ἀϱιστοφάνη’, Mnemosyne 39, 153–8 [= Scripta Minora, II, 607–
612].
– (1991): ‘Les valeurs temporelles des formes verbales en grec ancien’, The
Function of Tenses in Texts, Amsterdam, 197–217 [= Scripta Minora, II,
656–76].
– (1992): ‘L’emploi le plus ancien et les emplois plus récents de la particule κε/ἄν’, in: La langue et les textes en grec ancien, Amsterdam, 75–84
[= Scripta Minora, II, 677–686].
Ruiz Yamuza, E. (2008a): Tres verbos que significan “deber” en Griego
Antiguo, Zaragoza.
ipabo_66.249.78.36
46
Rutger J. Allan
– (2008b): ‘Análisis del comportamiento del verbο ΚΙΝΔϒΝΕϒΩ’, HABIS
39, 377–395.
Schwyzer, Ed. (1950): Griechische Grammatik, 2. Band, München.
Sihler, A. L. (1995): New Comparative Grammar of Greek and Latin,
Oxford.
Slings, S. R. (1975): ‘The etymology of ΒΟϒΛΟΜΑΙ and ΟΦΕΙΛΩ’,
Mnemosyne 28, 1–16.
Stahl, J. M. (1907): Kritisch-historische Syntax des griechischen Verbums
der klassischen Zeit, Heidelberg.
Strunk, K. (1988): ‘Zur diachronischen Morphosyntax des Konjunktivs’, in:
A. Rijksbaron, H. A. Mulder, G. C. Wakker (eds.), In the footsteps of
Raphael Kühner, Amsterdam, 291–312.
Tichy, E. (2002): ‘Zur Funktion und Vorgeschichte der indogermanischen
Modi’, in: H. Hettrich (ed.): Indogermanische Syntax-Fragen und
Perspektiven. Wiesbaden, 189–206.
– (2006): Der Konjunktiv und seine Nachbarkategorien. Studien zum
indogermanischen Verbum, ausgehend von der älteren vedischen Prosa,
Bremen: Hempen.
Traugott, E. Closs & R. B. Dasher (2002): Regularity in Semantic Change:
Cambridge: UP.
Traugott, E. Closs (1989): ‘On the rise of epistemic meaning in English: an
example of subjectification in semantic change’, Language 57, 33–65.
– (2003): ‘Constructions in grammaticalization’, in: R. D. Janda & B. D.
Joseph (eds.), The Handbook of Historical Linguistics, 624–47, Oxford:
Blackwell.
Vendler, Z. (1957): ‘Verbs and Times’, Philosophical Review 66, 143–60.
Wackernagel, J. (1926): Vorlesungen über Syntax, 1. Reihe, Basel.
Wakker, G. C. (2006): ‘Future Auxiliaries or Not’, in: E. Crespo, J. de la
Villa & A. R. Revuelta (eds.), Word Classes and Related Topics in
Ancient Greek. Proceedings of the Conference on ‘Greek Syntax and
Word Classes’ held in Madrid on 18–21, June 2003, Louvain-la-Neuve,
237–55.
– (2007): ‘Intentions and Future Realisations’, in: R. J. Allan & M. Buijs
(eds.), The Language of Literature: Linguistic Approaches to Classical
Texts, Leiden-Boston, 168–87.
Willmott, J. (2007): The Moods in Homeric Greek, Cambridge.
In search of the lexical meaning of the Latin
terms cetarius and cetaria
By PIOTR BERDOWSKI, Rzeszow
Abstract: The Latin term cetaria (or cetarium, neuter) and its adjectival form cetarius are very rare in ancient sources. In the majority of
Latin dictionaries cetaria is translated as “a fishpond” and often collated
with piscina: another Latin term commonly used for fishpond. Obviously this is reflected in modern translations of the ancient authors’ works
in which cetaria and cetarius appeared. However, nothing suggests
that cetariae served as ponds for raising fish. On the contrary: evidence suggests we are dealing with the relatively small basins which were
used in the production of fish sauces or other salted fish products.
Leafing through ancient Latin sources, one hardly finds the
term cetaria (or cetarium, neuter). Including the adjectival form
cetarius and relational adjective, there are only a handful of
occurrences of the words mentioned by just a few authors.1 In
the majority of Latin dictionaries cetaria is translated as “a fishpond” and often collated with piscina: another Latin term commonly used for fishpond. One finds such an definition in both A
Latin Dictionary by C.T. Lewis and C. Short and The Oxford
Latin Dictionary, referring only two essential Latin-English
dictionaries.2 As a result these meanings are reflected in modern
translations of the ancient authors’ works in which cetaria
(cetarium) appeared. However, nothing suggests that cetariae
served as fish breeding ponds. In fact evidence suggests we are
__________
1
All the occurrences are quoted in further paragraphs. Preserved scholia
to Terence and Horace do not change this proportion at all.
2
A Latin Dictionary. Founded on Andrews' edition of Freund’s Latin dictionary, revised, enlarged, and in great part rewritten by Charlton T. Lewis,
Ph.D. and. Charles Short, LL.D. Oxford. Clarendon Press. 1958 [1879];
Oxford Latin Dictionary. Oxford. Clarendon Press 1968. Cf. dictionaries in
other languages, for example: Karl E. Georges, Lateinisch-deutsches und
deutsch-lateinisches Handwörterbuch, Leipzig 1843, 1103; F. Gaffiot, Dictionnaire latin français, Paris 1934, 296.
Glotta 89, 47–61, ISSN 0017–1298
ipabo_66.249.78.36
Piotr Berdowski 48
dealing with the relatively small basins which were used in the
production of fish sauces (such as garum) or other salted fish
products (salsamenta). Hence, the following paper is intended to
analyze all the available sources to support my thesis.
The earliest mention of the basins comes from Horace’s
Serm. 2.5.44, but it is worth moving possibly farther back to the
text of Eunuchus by Terence, where the adjectival form of cetarius is found. Obviously it has the same stem as cetarium, but it
is hard to say which form is earlier (it might be cetarius but this
is only a speculation). The term cetarius appears in the monologue of Gnatho, one of the main characters of the play (l. 257).
He is a typical comedic parasite, who mastered cheating naive
people to perfection. Blowing his own trumpet he speaks to
himself:
dum haec loquimur, interea loci ad macellum ubi advenimus, ||
concurrunt laeti mi obviam cuppedenarii omnes, || cetarii lanii
coqui fartores piscatores, || quibus et re salva et perdita profueram
et prosum saepe: || salutant, ad cenam vocant, adventum gratulantur.
While we were thus talking, in the mean time we arrived at the
market-place; overjoyed, all the confectioners ran at once to meet
me; fishmongers, butchers, cooks, sausage-makers, and fishermen, whom, both when my fortunes were flourishing and when
they were ruined, I had served, and often serve still: they complimented me, asked me to dinner, and gave me a hearty welcome.3
While we were talking like this, in the meantime we arrived at the
market, and up there rushed, glad to meet me, all the sellers of
fancy foods, the tunny-sellers, butchers, cooks, poulterers, and
fishmongers, who’ve profited from me both before and after I’d
lost my property, and who often still do. They greeted me, invited
me to dinner, welcomed me home.4
__________
3
The Comedies of Terence. Henry Thomas Riley. New York. Harper and
Brothers. 1874. See also Tromaras (1994, 155–156); Karakasis (2005, 30,
122–123).
4
Terence. The Comedies, translated by Peter Brown. Oxford University
Press. 2006.
In search of the lexical meaning of the Latin terms cetarius and cetaria
49
These two modern translations of Eunuchus are separated by
more than a century. An interesting shift is observed here. In the
first version cetarii are translated as “fishmongers” whereas in
the second paragraph they are “tunny-sellers”. Respectively the
term piscatores is translated as “fishermen” and “fishmongers”.
It is worth stressing that the usage of cetarii was isolated until
the end of the Republic, when the term appears in Cicero (it is
no more than the citation from Terence’s Eunuchus) and M.
Terentius Varro. Valuable comparative material comes along only
in the first century C.E. Unfortunately this makes all the reconstructions of the meaning of cetarii in Terence hypothetical.
It would be very useful to know the Greek equivalent of
cetarii. Unfortunately, the Greek prototype of the Eunuchus,
which was Menander’s play, has not survived, except for a few
fragments (cf. Lefèvre 2003). The ancient commentators agreed
that the term cetarius is derived from the Greek noun κῆτος
which designated a marine monster or a big marine fish, for
example a dolphin, whale, or shark.5 The tuna fish, so popular
among the Greeks and Romans, was also possibly indicated by
the term. This does not necessarily mean that Menander used the
term which was compounded with κῆτος. There were more
possibilities. In the Greek text there might have been for example
ταριχοπῶλαι or another term.
Another option for reconstructing Terence’s meaning is studying the scholia to the Eunuchus. Aelius Donatus (fl. 4th cent.
C.E.) in his commentary to the Eunuchus writes: cetarii qui cete,
id est magnos pisces, venditant et bolonas exercent (In Eunuch.
257 [II 2.26]). Admittedly he speaks about fishermen who specialize in fishing for large fish (exercent bolonas) but he stresses
at the same time that they also sell the fish. Quite a different
explanation of the cetarii was preserved in the scholia, edited by
Fridericus Schlee: cetarii] cetus est genus marinae beluae. ab
hoc genere abusive piscatores cetarii dicuntur et qui tractant ea,
quae ex piscibus fiunt, ut sunt liquaminarii, qui ex corporibus
__________
5
Short & Lewis, s.v. cetus; LSJ, 949–950, s.v. κῆτος.
ipabo_66.249.78.36
50
Piotr Berdowski piscium humorem liquant.6 The anonymous scholiast presumes
that the term cetarii is loosely (abusive) associated with fishermen (piscatores) and those who operate in fish-salting workshops. It looks as if the author of the commentary (or more precisely the source he used) knew that the fish were processed in
the basins called cetariae, although he did not mention of them.
As can be seen, scholiasts give three explanations: cetarii were
fishermen, producers of fish sauces and fishmongers. Which explanation was right in the time of Terence? If the scholiasts
formulated their opinion during the imperial period when the
production of fish sauces and salted fish was very common it
might have been that their definition was based on their personal
observations and not necessarily on the reading of the earlier
authors.7 This is quite significant, because it means that in Terence’s time the meaning of cetarius might have been different.
Modern scholars prefer (after Short and Lewis Latin Dictionary)
the translation of cetarii as “fishmongers”. Obviously it is not
easy to decern if this is correct. One can only point out that the
other professions enumerated by Terence are all productive ones
(not selling): confectioners (cuppedenarii), butchers (lanii), cooks
(coqui), sausage-makers (fartores). It may weaken the fishmongers explanation. On the other hand, the demarcation line between the production and the trade might have been fluid. We
know almost nothing about the production of fish sauces in Italy
in Terence’s time. Archeologists have discovered no salting
basins similar to those found in the Western provinces (e.g. in
__________
6
Schol. Terent. ad Eun. 257 (Commentarius antiquior). The edition by F.
Schlee, Bibliotheca Teubneriana, Lipsiae 1894. The determining of the
authors and the time of the commentaries in Schlee’s edition is a debatable
issue. For the most part he ascribed them principally to Servius and Priscian.
It is safe to treat Commentarius antiquior as a medieval text belonging to one
author, edited not earlier than the 8th century. Cf. Rand (1909, 359–389).
7
It is to be regretted that the works of Verrius Flaccus, lexicographer and
grammarian from the time of Augustus mostly have not survived.
51
Spain, Gaul and Mauretania).8 Nevertheless, it seems that the
term cetarius was intelligible for Terence’s public.9
The fragment from Terence’s Eunuchus was quoted by Cicero
in his treatise De officiis (1.150) in the famous evaluative passage on work and different professions. Cicero does not comment
on the term cetarii and it seems that it was intelligible to his
contemporaries. The term Cetarini (different spelling Citarini)
also appears in another text of Arpinata (In Verr. act. II, 3.103).
In this case it is used to designate dwellers of the town in the
western part of Sicily. Possibly the majority of whom were fishermen and fish workers who specialized in catching large fish like
tuna.10 We know of at least several fish salting workshops which
were discovered in Sicily. They were equipped with salting basins, which were in use in Cicero’s time (Curtis, 1991, 99–101).
However, we have no evidence to link them with Cetarini.
More or less at the same time the form cetarius known from
Terence is found in one of Varro’s Menippean satires (fr. 209,
Riese): Γνῶθι σεαυτόν: Non animadvertis cetarios, cum videre
volunt in mari thunnos, escendere in malum alte, ut penitus per
aquam perspiciant pisces? Cetarii, who get up on a high pillar
(malus) to spot schools of tuna in the sea water, are workers of
fish salting workshops (officinae) or fishermen (who were in
fishing and fish processing as well). It is hard to think that they
were fishmongers. It is worth adding that thynoskopeia (“tuna
fish watchtowers”), similar to those mentioned by Varro, are
known from other ancient sources and they were still in use at
the beginning of the twentieth century (cf. Mylona 2008, 49).
__________
8
However, the epigraphic material from Pompeii and Herculanum leaves
no doubt that from the first century C.E. on the production of fish sauces in
Italy was an objective fact. See Curtis (1991, 85–96).
9
However, it is difficult to understand why Robert Maltby (1985, 113–
114) counted the term cetarius among those “well integrated in the language
by Terence’s time”, since it was the first occurrence of the word, isolated
until the Late Republic.
10
Cetarinii on Sicily are mentioned by Pliny the Elder (HN 3.8.14). Cf.
Ptolem. Geogr. 3.4: Κηταρία. See also Smith (1854), s.v. cetaria; Holm
(1866, 323). In modern Italy in Campania there is a small town Cetara, whose
name comes evidently from the ancient cetaria.
ipabo_66.249.78.36
52
Piotr Berdowski Only fragments of Varro’s Manipean satires have survived. We
owe that mention of cetari to Nonnius Marcellus, a grammarian
living in the time of the Severian dynasty (Keyser 1994, 369–
389; Albrecht 1997, 1469–1470). He cited Varro to illustrate his
own explanation of the term cetarii: genus est piscatorum quod
maiores pisces capit, dictum ab eo quod cete in mari maiora
sunt piscium genera: a quo Vergilius: inmania cete. It is clear
that for the grammarian cetarii were fishermen who specialized
in fishing for large fish. This explanation coincides with the
Scholia Terentiana cited above. Nonnius Marcellus suggests that
fishermen loaned their name to fish, not the other way round.
The first substantive form of cetarium is present in Horace’s
Serm. 2.5.44, which treats the “art” of easy profit in Rome by
coming into an inheritance. Horace compares the “catching” of
naive wealthy men by a highly specialized fraudster with the
fishing of tuna fish. The fish arrive (adnabunt) and fill salting basins (cetaria crescent): “nonne vides” aliquis cubito stantem
prope tangens / inquiet, “ut patiens, ut amicis aptus, ut acer?”/
plures adnabunt thynni et cetaria crescent. The phrase cetaria
crescent means that the basins called cetaria will be filled up by
the tuna fish. Actually, this kind of basin was discovered by archaeologists in the western provinces of the Empire (for more
about the archaeological sources see below). After Lewis & Short
Latin-English Dictionary the noun cetarium was usually understood as “fishpond”. However, there is no reason to translate
cetaria in this way. They differed substantially from the piscinae:
the fish were never bred in cetaria, even for a short period of time
(judging by the cetaria uncovered by archaeologists they were
simply too small for this purpose). The translation of Horace’s
passage should respect this meaning or alternatively the original
Latin term should be appended with a gloss in the footnotes.
This interpretation of cetarium is confirmed by the scholia of
Horace. Two interesting commentaries can be found in the Pseudacronis Scholia in Horatium Vetustoria, which is an early me-
53
dieval compilation from Roman grammarians.11 In one of them
(Γ’ b E) it is stated clearly that cetaria are places where the fish
are salted: cetaria dicuntur proprie loca, in quibus salsamenta
fiunt. Were the above mentioned loca basins used for fish processing? This is not certain. In the second commentary (Γ’ b f e
ζ) the emphasis was put on a workshop (officina), where the fish
sauce liquamen is made: officina, in qua liquamen conficitur. It is
likely that the loca mentioned in the first commentary refer to the
officina. Of course, this broader understanding of the cetaria by
the scholiasts was not necessarily known to the poet himself. The
scholiasts might or might not have been aware of the semantic
changes which had taken place with cetarium from Horace’s
time. In any case, there is no reason to think that cetaria served
as fishponds. It looks so if one compares the above referenced
sources with the text of Historia Naturalis by Pliny the Elder.
The first passage (9.49) discusses the fish species such as the
pelamys and mackerel, which grow quickly in the favorable
environment of the Black Sea basin. Along with tuna schools
they enter the Black Sea so as to come back to the waters of
Spain and “to fill Spanish fish salting basins”: Hispaniae cetarias
hi replent.12 Although one reads in the Loeb Library translation
of the Historia Naturalis by H. Rackham about “fishponds”,
there is no reason to think that Pliny meant fishponds in fact.
Possibly Pliny’s understanding of the cetariae was similar to
that of Horace’s scholiasts, which was presented above. It seems
that this broader meaning can also be found in HN 9.92, where a
gigantic polyp is the central figure of the story told by Pliny. It
used to wreak havoc on fish salting basins which were located
along the seashore of Carteia (Spain):
Carteiae in cet<a>r<i>is adsuetus exire e mari in lacus eorum apertos atque ibi salsamenta populari – mire omnibus marinis expetentibus odorem quoque eorum, qua de causa et nassis inlinuntur –,
__________
11
12
Ed. O. Keller, vol. 1‒2, Lipsiae 1902–04.
In Pliny the noun cetaria is feminine.
ipabo_66.249.78.36
Piotr Berdowski 54
convertit in se custodum indignationem adsiduitate furti inmodica<m>.
In the fishponds at Carteia a polyp was in the habit of getting into
their uncovered tanks from the open sea and there foraging for
salted fish – even the smell of which attracts all sea creatures in a
surprising way, owing to which even fish-traps are smeared with
them – and so it brought on itself the wrath of the keepers.
Afterwards Pliny tells the story about the insidious capture of
the recalcitrant polyp. The tale itself (indebted – as Pliny states –
to Trebius Niger, proconsul in Spain in the second century
B.C.E.) was obviously a fabrication. It is hard to imagine a huge
polyp which would walk along the seashore in search of an
evening meal.13 Whether the tale is true is not crucial at the
moment – what matters is that Pliny’s lacus aperti are the basins
with the salsamenta, which were gobbled by the incriminated
polyp. The pronoun eorum also makes an important point in the
cited text. Its linguistic usage means that the cetariae were
something more than fish salting basins; they referred simply to
the fully equipped workshops. In this particular instance it is
possible to equate the cetariae with the officinae. It is interesting
to notice in this context that archaeologists discovered the
remains of two fish salting installations in Carteia. One of them
operated in the second and first century B.C.E., and the second
from the first to the third century C.E. (López Castro 1993, 117;
Ponsich, Tarradel, 1965, 85, 118).
L. Iunius Moderatus Columella was the last writer (aside
from the subsequent works of lexicographers) who referred to
the cetariae. The author of De re rustica compiled his treatise
probably in the 70s of the first century C.E. The passage of
interest to us comes from the last chapter of Book Eight, and it
is devoted wholly to the subject of the ideal fishpond. According
to Columella this is a seaside fishpond. The author presents
different aspects of fish breeding and subsequently discusses
__________
13
See modern analogies in Asplund Ingemark (2008, 145–170). I would
like to thank the author for making her paper available to me.
55
how to nourish fish and which species may be used for food. He
concludes that the salted fish (salsamenta) may also serve as
nourishment (8.17.12): (…) et – ne per singula enumerem –
salsamentorum omnium purgamenta, quae cetariorum officinis
euerruntur. Obviously, Collumella spoke of the remnants of
salted fish (purgamenta), which were removed from the basins
during the cleaning of the officinae after a particular production
cycle was finished. Columella labeled the above mentioned
workshops as the officinae cetariorum. I think that in this case
the form cetarii refers to workers of the officinae (it is less likely
that Columella had in mind the owners of the officinae).
The sources which were analyzed above are crucial for demarcating the semantic field for cetarius and cetaria (cetarium).
Both terms were mentioned sporadically in later works, including those by lexicographers. In addition to the above quoted
scholia to Terence and Horace, one can find an interesting passage in the treatise De verbis dubiis, preserved under the name
of Flavius Caper, a grammarian from the second century C.E. It
is probably an excerpt from the lost treatise De dubiis generibus
by Flavius (Sallmann 2002, 232–5). The author explains the
term cetariae in the following way: cetariae tabernae, quae
nunc cretariae non recte dicuntur. The link between cetariae
and tabernae is unexpected. Cretariae means chalk-shops or
maybe even workshops where clay products were made (OLD,
s.v. creta), but this is probably not crucial in Flavius’ statement:
he underlines that cretariae are non recte associated with cetariae. The similarity between them ends with the sound of the
words. But why did Flavius name them cetariae tabernae? The
latter were generally places where various items (among them
food) were sold. Therefore were cetariae not only the workshops where fish products were made but at the same time the
shops where they were sold? This is rather unlikely. No other
evidence confirms the selling function of cetariae. Hence
Flavius’ explanation should be rejected, which does not obliga-
ipabo_66.249.78.36
56
Piotr Berdowski torily exclude the fact that individual cetariae may have sold
their own products.14
One of the examined terms appears again on the margin of the
statement of Pomponius Secundus, who used the plural form in
the third declension: cetaria, -ium (Plin. Sec. Dub. sermo, fr 77).
The usage of cetarius is also found in Augustine (Contra Iulianum 2.10.37), but the context does not allow the identification of
the meaning. The juxtaposition of cetarii and other professions
(nautae, tabernarii, coqui, lani), indicates that Augustine knew
Terence’s play Eunuchus, maybe even Flavius Caper’s treatise
(the only ancient author who had seen the link between cetariae
and tabernae – Augustine enumerates cetarii next to tabernarii).
Concluding with the written sources where cetarii and cetariae
appear it is necessary to mention early medieval glossaries. Some
of them registered the term cetarii. The Codex Vaticanus 3321
(= CGL IV 32.23) for example noted down: cetaris liquaminari.
It is clear that in this case liquaminari were the workers who
produced fish sauces (for example liquamen).15 This Italian codex
cannot be precisely dated (the proposals vary from 6th to 8th
century), and connected with any particular author (Nettleship
1890, 255; Lowe 1921, 189–191). Next the Codex Leidensis 67F
(glossae “abavus” = CGL IV 318.7) gives another explanation:
cetari qui salsamenta vendunt. Here cetari are fishmongers. The
codex is dated to 8th or 9th century (Nettleship 1890, 255).
Almost three centuries later Papias the Lombard tried to collect
all the earlier explanations found in the classical and later literature.16 At first glance medieval glossaries seem to be useful to
define the meaning of cetari in classical and late Latin, but the
difficulties arises when we realize that they were compiled centuries after Terence, Horace and other authors. Additionally we
do not know what source of material was used by their editors.
__________
14
It seems to me that Flavius Caper’s juxtaposition of cetariae with
taberne was taken again by Papias, lexicographer from Lombardia living in
the 11th century. See Du Cange, Glossarium mediae et infimae latinitatis, vol.
2, Niort 1883, s.v. cetarius.
15
This could be taken from Scholia in Eunnuchus by Terence (cf. above).
16
Du Cange, vol. II, s.v. cetarius.
57
One can choose from the glossaries which definition fits better
any one particular usage but the choice will remain arbitrary.
Apart from the narrative texts and philological sources, it is
worth asking if archaeological objects and epigraphic materials
could verify the findings presented above. Fortunately, the last
decades were marked by intensive studies on ancient fish processing. Numerous reports from the archaeological discoveries
(mainly in Spain, Africa and Gaul) have been published since
then. The number of sites with basins covered with a waterproof
layer of the opus signinum, in which the fish mass mixed with
salt was exposed to the sun (the process of autolysis melts fish),
had already exceeded two hundred installations. Frequently the
basins were accompanied by utility buildings.17 Systematic studies on amphorae, their shapes and painted inscriptions helped to
distinguish different fish products and understand the mechanisms of the distribution of them in the Mediterranean basin.
Today one cannot doubt that the production of salted fish products constituted a significant part of the ancient economy, and
especially the Roman economy. Most of the ancient workshops
produced fish sauces and salsamenta in quantities which exceeded local needs. Consequently they exported them to other
provinces of the Empire. Some workshops were huge industrial
complexes like Lusitanian Tróia or Mauretanian Lixus, where
the production took place for the larger part of the year (Étienne,
Makaroun, Mayet, 1994; Ponsich, Tarradell, 1965, 9–37). Of
course small workshops are known as well. Sometimes they
functioned in larger economic structures as for example the villa
farms.18 The production of fish sauces and salsamenta required
large amounts of salt which was acquired in salt mines (most of
the ancient salinae have been exploited until the present time)
and by evaporation of sea water as well (cf. Villalobos, Prieto,
Ménanteau 2003, 317–332).
__________
17
For more about the technology and the modern equivalent of ancient
fish sauces popular today especially in South-East Asia see Curtis, (1991, 6–
26).
18
Cf. Berdowski 2000, 259–293.
ipabo_66.249.78.36
Piotr Berdowski 58
There is no doubt that in the majority of the workshops the
fish sauces were made of marine fish delivered successively by
fishermen. The participation of the pisciculture (piscicultura) in
fish supply to the workshops is a controversial issue, but it
seems that it had marginal significance. Whether the fishponds
(piscinae) bordered on the cetariae is another debatable
question. Our knowledge about fish breeding in Roman Spain is
still modest (cf. Lagóstena 2007, 109–115). We know much
more about the fishponds in Italy, although ironically enough,
the archaeologists did not discover in Italy a single basin
(cetaria) for fish processing similar to those known from the
Western provinces of the Roman Empire.19
The accessible archaeological material combined with the
narrative sources does not argue for the thesis that the terms
cetariae (cetaria) might have referred to fishponds. The latter
were labeled piscinae, vivaria, sometimes stagna.20 Even if fish
were kept in storage basins for a short period of time (to provide
a steady delivery of fish resources for example – the fluctuation
of fishing must have been natural for the ancients), they were
called the vivaria. The phrase navis vivarius was used to
describe a ship equipped with live tanks for transporting fish
(Higginbotham 1997, 7). Also the term cetarii had nothing to do
with fish-farmers or owners of fishponds.
The terms cetaria and cetarius are not confirmed with
certainty in the epigraphic material, though some scholars believe they were. The problem concerns a series of stamps on the
amphorae from Rinconcilli (Algeciras) and Belo in Spanish
Betica: S.CET. and SCG. Robert Étienne and François Mayet
(1994, 134) suggested that S(ocii) CET(arii) in the first stamp,
should be understood to be referring to the company of fish producers (cf. Lagóstena Barrios 2001, 391; Haley, 62). The second
stamp was read in the following way: S(ocii) C(etarii) G(aditani)
or S(ocietas) C(etariorum) G(aditanorum). Other sources con__________
19
The proximity of piscinae/vivaria with cetariae might be a Spanish
feature.
20
Higginbotham (1997, 7).
59
firm that the societates of fish producers actually existed
(Étienne 1970, 297–313; Curtis, 1991, 64); nevertheless in this
context we may also be dealing with the tria nomina. The
quoted inscriptions appear on the amphorae which were typical
for the fish products: Dressel 7/11 and Beltrán IIA 1 (Dressel 1
type, normally used to transport olive oil, is confirmed as
well21). Thus the problem of interpretation of the stamps S.CET
and SCG has not been satisfactorily solved. If the interpretation
of Étienne and Mayet is correct, the cetarii should be understood
as producers or perhaps as even the owners of workshops (officinae). They were not necessarily the same persons.
The archaeological and epigraphic sources are not crucial to
the problem discussed in this paper, but they do not contradict
the findings based on the literary texts. Nothing is suggestive of
the connection between the cetariae and cetarii and fish
breeding, while it seems firmly proved that the terms were
referring to the fish salting industry.
In conclusion, the present state of our knowledge should incline us to modify the entries of cetarius and cetaria / cetarium
in many Latin dictionaries (not only Latin-English ones), which
will lead to more accurate and precise translations of the Latin
terms for modern languages. Thus I would like to propose a
model entry of cetaria and cetarius in a future Latin-English
dictionary.
cetaria, ~ae, f. and cetarium, ~i (ii), n. in plural basins where fish
mixed with salt and exposed to the sun were processed. –
Hispaniae ~as hi [scombri] replent (Plin. HN 9.49); – Carthaginis
spartariae ~is (31.94); – [garum conficiebatur] nunc a scombro
pisce laudatissimum in Carthaginis spartariae ~is (31.94); – plures
adnabunt thynni et ~a crescent (HOR. Sat. 2.5.44); 2. in plural
workshops (officinae) including not only basins but also additional
equipment. – Carteiae in ~is adsuetus [polypus] exire e mari in
lacus eorum apertos atque ibi salsamenta populari (9.92); – ~ia
__________
21
It is not surprising because occasionally fish products were transported
in the containers usually used for other products. See Zevi (1966, 225–226);
cf. Peña (2007, 39–118).
ipabo_66.249.78.36
60
Piotr Berdowski dicuntur proprie loca, in quibus salsamenta fiunt (Pseudoacron.
schol. in Horat. Vetust.); cf. ibid. [cetaria] officina, in qua liquamen
conficitur.
cetarius, ~ii, m. A fisherman or fishmonger. – ~ii, lanii, coqui,
fartores, piscatores (TER. Eun. 257); – non animadvertis ~os cum
videre volunt in mari thunnos, escendere in malum alte (VARRO
Sat. Men. fr. 209). 2. worker in the workshop (officina) where fish
are processed; owner of the cetariae. – salsamentorum omnium
purgamenta, quae ~orum officinis euerruntur (COLUM. 8.17.12); –
~ii. cetus est genus marinae beluae. ab hoc genere abusive
piscatores cetarii dicuntur et qui tractant ea, quae ex piscibus fiunt,
ut sunt liquaminarii, qui ex corporibus piscium humorem liquant
(Schol. Terent. ad Eun. 257, comment. antiq.).
Bibliography
Albrecht, M. von (1997): A History of Roman Literature from Livius Andronicus to Boethius with Special Regard to Its Influence on World Literature, Leiden: E. J. Brill (Mnemosyne Supplement 169).
Asplund, I. C. (2008): “The Octopus in the Sewers: An Ancient Legend
Analogue”, Journal of Folklore Research 45, 145–170.
Berdowski, P. (2000): “An attempt at a new way of looking at villa-farms
and their owners in Spain (I–II century A.D.)”, in: E. Dąbrowa (ed.),
Electrum. Studies in ancient history, vol. 4, Kraków, 11–30 [in Polish].
Curtis, R. I. (1991): Garum and salsamenta. Production and Commerce in
materia medica, Leiden: E. J. Brill (Studies in Ancient Medicine, 3).
Étienne, R. (1970): “A propos du garum sociorum”, Latomus 29, 297–313.
Étienne, R., Makaroun, Y. and Mayet, F. (1994): Un grand complexe
industriel a Tróia (Portugal), Paris: De Boccard.
Étienne, R., Mayet, F. (1994): “A propos de l’amphore Dressel 1C de Belo
(Cadix)”, MCV 30, 131–138.
Haley, E. W. (2003): Baetica Felix. People and Prosperity in Southern Spain
from Caesar to Septimius Severus, Austin: University of Texas Press.
Higginbotham, J. (1997): Piscinae: Artificial Fishponds in Roman Italy,
Chapel Hill: The University of North Carolina Press.
Holm, A. (1866): Beiträge zur Berichtigung der Karte des alten Siciliens,
Lübeck.
Karakasis, E. (2005): Terence and the Language of Roman Comedy,
Cambridge: Cambridge University Press.
61
Keyser, P. T. (1996): “Late Authors” in “Nonnius Marcellus and Other
Evidence of His Date”, HSCP 96, 369–389.
Lagóstena Barrios, L. (2001): La producción de salsas y conservas de
pescado en la Hispania Romana (II a.C. – VI d.C.), Barcelona:
Universidad de Barcelona (Collecció Instrumenta, 11).
– (2007): “Columela, De Re Rustica VIII, 16–17: una fuente para el
conocimiento de la piscicultura en Baetica”, in Actas del Congreso
Internacional CETARIAE. Salsas y salazones de pescado en Occident e
durante la Antigüedad, Universidad de Cádiz, Noviembre de 2005, 109–
115, Oxford: Archaeopress. Publishers of British Archaeological Reports
(BAR IS 1686).
Lefèvre, E. (2003): Terenz’ und Menanders Eunuchus. München: C. H. Beck
(Zetemata, 117).
López Castro, J. L. (1993): “La producción fenicia occidental de salazón de
pescado”, in II Congresso peninsular de Historia Antiga. Actas Coimbra,
1990, Coimbra.
Lowe, E. A. (1921): “On the Oldest Extant MS. of the Combined Abstrvsa
and Abolita Glossaries”, CQ 15, 189–191.
Maltby, R. (1985): “The distribution of Greek loan-words in Terence.” CQ
35, 110–123.
Mylona, D. (2008): Fish-eating in Greece from the fifth century B.C. to the
seventh century A.D. A story of impoverished fishermen or luxurious fish
banquets, Oxford (BAR IS 1754).
Nettleship, H. (1890): “The fourth volume of Goetz’s Corpus Glossariorum
Latinorum”, CR 4, 255.
Peña, T. J. (2005): Roman pottery in the archaeological record, Cambridge:
Cambridge University Press.
Ponsich, M., and Tarradel, M. (1965): Garum et industries antiques de salaison dans la Méditerranée occidentale, Paris: Presses universitaires de France.
Rand, E. K. (1909): “Early Mediaeval Commentaries on Terence”, CP 4,
359–389.
Sallmann, K., ed. (2002): Die Literatur des Umbruchs. Von der römischen
zur christlichen Literatur 117 bis 284 n. Chr., München.
Smith, W., ed. (1854): Dictionary of Greek and Roman Geography, London.
Tromaras, L., ed. (1994): P. Terentius Afer Eunuchus. Einführung, kritischer
Text und Kommentar, Hildesheim: Georg Olms Verlag.
Villalobos, C. A., Prieto, F. J. G. and Ménanteau, L. (2003): “Las salinas de
la Bahía de Cádiz durante la antigüedad: visión geoarqueológica de un
problema histórico”, SPAL 12, 317–332.
Zevi, F. (1966), “Appunti sulle anfore romane: I, La tavola tipologica del
Dressel”, Archeologia Classica 17, 208–247.
ipabo_66.249.78.36
Eine Anmerkung zu μελάνδετος
Von SARA CHIARINI, Berlin
Das griechische Adjektiv μελάνδετος taucht bereits im archaischen Epos auf: einmal in der Iliad (O 713: πολλὰ δὲ φάσγανα
καλὰ μελάνδετα κωπήεντα) und ein weiteres Mal im Scutum
Herculis (V. 221: ὤμοισιν δέ μιν ἀμφὶ μελάνδετον ἆοϱ ἔκειτο).
In beiden Zusammenhängen bezieht es sich auf ein Schwert,
wobei die wortwörtliche Übersetzung etwa dem „schwarz gehüllten Schwert“ entspricht.
Die Doppeldeutigkeit des Ausdrucks ist offensichtlich, weshalb schon die antiken Kommentatoren eine Erklärung zu
formulieren suchten. Sowohl die Homerischen Scholia, als auch
die Scholia zum Scutum Herculis, gingen die quaestio im
Zeichen tiefen Rationalismus an - welcher in der modernen
Kritik noch fortbestand:
Schol. D O 713:
μελάνδετα· σιδεϱόδετα, ἢ μέλαινας λαβὰς ἔχοντα·
ἔστι γὰϱ ἡ λαβὴ δεσμός τις τοῦ ξίφους.
Schol. Sc. 221:
[…] εἴη δ’ἂν μελάνδετον, ἢ τὸ μέλαιναν θήκην
ἔχον, ὡς ἐκ δέϱματος κατεσκευασμένην, ἢ τὸ ἐκ
μέλανος σιδήϱου δεδεμένον.
Beide Scholia erklären, dass das Epitheton des Schwertes entweder auf die Eisennieten, mit denen die Metallplatten auf dem
Handgriff des Schwertes befestigt werden, oder auf die ledergewickelte Degenglocke, die das Greifen erleichtern sollte,
anspielen könnte. Im ersten Fall wäre μελάνδετος ein einfaches
Synonym für σιδεϱόδετος. Im zweiten Fall würde sich die
Bedeutung vom durch den Gebrauch geschwärzten Leder herleiten lassen.
Glotta 89, 62–67, ISSN 0017–1298
63
Unter den Wissenschaftlern der Moderne fand letztere Erklärung den größten Anklang,1 obgleich auch andere Hypothesen
aufgestellt wurden. Allerdings bezog man sich auch bei der
Entwicklung alternativer Erklärungsmuster auf eine materielle
Ursache für die Assoziation des Schwertes mit Schwärze.2
Bei Betrachtung der vorliegenden Argumente, die zwischen
ausufernder Komplexität einerseits und rationalistischer Analyse
andererseits changieren, stellt sich die Frage, inwieweit die
Möglichkeit besteht, das Epitheton einfacher zu erklären.
Betrachtet man die späteren Okkurrenzen des Wortes, scheint
diese Überlegung, wie im Folgenden bewiesen wird, umso
berechtigter.
In der Tragödie verliert das Adjektiv seine materielle Bedeutung vollständig, und nimmt einen evokativen und übertragenen
Sinn ein. Einige Ausschnitte sind in dieser Hinsicht sehr
deutlich:
Eur. Or. 819–822: τὸ καλὸν οὐ καλόν, τοκέων
πυϱιγενεῖ τεμεῖν παλάμᾳ
χϱόα, μελάνδετον δὲ φόνῳ
ξίφος ἐς αὐγὰς ἀελίοιο δεῖξαι·
Das Schwert, mit welchem Orestes seine Mutter ermordet hat,
wird als μελάνδετον δὲ φόνῳ bezeichnet, und zwar als „schwarz
__________
1
Helbig, 338–339; Leaf, I 584; Russo, ad v. 221 und Janko, 305 (ad O
713–715). Beck, LfGrE III 93–94 (s.v. μελάνδετον) betrachtet diese Erklärung als die plausibelste, obwohl er sie nicht ohne Zögern genehmigt. Er
gibt zu, dass der δεσμός auch auf die Niete anspielen könnte, durch welche
die Metallplatten auf dem Schwertgriff befestigt werden. Diese letzte Bedeutung wird aber von Homer mit dem Epitheton ἀϱγυϱόηλος normalerweise
angedeutet und, obwohl Silber tatsächlich schwarz werden kann, wird dieses
Metall im Epos immer als glänzend beschrieben.
2
Trümpy, 62–63 meinte eher die Farbe des Scheideleders, in der das
Schwert aufbewahrt wird. Buchholz, ArchHom E 238; 253–254 beruft sich
sogar auf die Knöpfe der mykenischen Degenglocken, die manchmal aus
dunklem Alabaster gefertigt wurden, sonst aus Elfenbein (es gibt einige
Exemplare aus den Gräbern der Kreise A und B von Mykene).
ipabo_66.249.78.36
64
Sara Chiarini
vom Mordfall“, oder - um innerhalb der Sphäre visuellen Ausdruckes zu bleiben - „schwarz vom Blut des Mordopfers“.3
Eur. TrGF V 1 F 373: πᾶς δ’ἐξεθέϱισεν ὥστε πύϱινον ‹στάχυν›
σπάθῃ κολούων φασγάνου μελανδέτου.
In diesem Euripides Fragment aus dem Satyrspiel Eurystheus
wird das Schwert, mit welchem Herakles die Hydra gerade
enthauptet hat, mit dem gleichen Epitheton und der gleichen
metaphorischen Bedeutung gekennzeichnet.
Bezüglich beider Ausschnitte äußern sich alle Kommentatoren und Übersetzer einheitlich, man deutet die Schwärzung als
durch Blut, und in übertragenem Sinne, durch Tod hervorgerufen, wofür die Waffe das Werkzeug ist.
Von den zwei anderen Okkurrenzen μελάνδετος in der
griechischen Literatur, bei Aesch. Th. 44 und Eur. Ph. 1091, ist
die erste sehr interessant, weil sie die einzige ist, bei der das
Epitheton nicht im Zusammenhang mit Schwertern Verwendung
findet, sondern mit einem Schild:
Aesch. Th. 42–44: ἄνδϱες γὰϱ ἑπτά, θούϱιοι λοχαγέται,
ταυϱοσφαγοῦντες ἐς μελάνδετον σάκος,
καὶ θιγγάνοντες χεϱσὶ ταυϱείου φόνου
[...]
Diese Wörter werden von Eteokles Boten verkündet, der seinem
Herrscher das gegen Thebes marschierende Heer beschreibt. In
diesem Fall erscheint es noch schwieriger, den Ausdruck archäologisch zu begründen.4 Diesmal erklären die Scholia schon alles
__________
3
Vgl. Willink, 219 (ad vv. 821–822): «to display the sword black-adorned
with their [Klytaimestra und Aigistos] blood». Es ist vielleicht überflüssig, an
die Häufigkeit zu erinnern, mit welcher Blut im Epos mit Schwarz verbunden
wird. Es reicht hier, ein paar Stellen aus der Iliad zu zitieren: Δ 149, E 798, H
262, K 469, Π 667, ϒ 470...
4
Groeneboom, 90–91 (ad vv. 42–44) leugnet das schon. Der einzige Versuch, den Ausdruck von Aeschilus antiquarisch zu übersetzen, ist das unwahrscheinliche „iron-rimmed“ von LSJ, s.v. μελάνδετον. Ein solches Merkmal findet aber keine archäologische Bestätigung und dieser Interpretation
65
und bestätigen die gestimmte Metonymie des Adjektivs:
Schol. vet. Aesch. Th. 43a: τὸ μελανισθὲν τῷ αἵματι, ἢ τὸ ἐκ μελαινῶν
βυϱσῶν πεϱιβεβλημένον. Οὕτως δὲ
θύοντες ἐπάνω τῶν ἀσπίδων ἐμαντεύοντο
[…].
Schol. vet. Aesch. Th. 43b: τὸ μελαινόμενον τῷ αἵματι ἐν τῷ πολέμῳ
[…].
Was spräche dagegen, den früheren Okkurrenzen denselben Sinn
zuzuschreiben? Wendet man sich erneut dem Homerischen
Ausschnitt und jenem aus dem Scutum zu, so stellt man fest,
dass eine metaphorische Funktion des Epithetons völlig passend
wäre. Im Vers 713 des XV. Ilias Liedes findet man den Höhepunkt der vor den archaischen Schiffen wütenden Schlacht
beschrieben. Um das Bild des Wirrwarrs und des Dramas
besonders hervorzuheben, verweilt Homer in der detaillierten
Beschreibung der von den Kriegern verlorenen Schwerter. Deshalb könnte man den Ilias Vers etwa wie folgt übersetzten:
„viele schöne, griffausgestattete und mit Blut befleckte (oder:
vom schwarzen Tod umschlossene) Schwerter“.5
Genauso könnte man die Stelle des Scutum angehen, an der
sich der metaphorische Sinn des Adjektivs in zweifacher
Hinsicht geradezu aufzwingt. Perseus hat soeben Meduse genau
mit jenem ἆοϱ enthauptet, das also nicht nur tödlich per se,
sondern nun auch wortwörtlich von Medusas schwarzem Blut
befleckt ist. Darüber hinaus würde diese Deutung wohl dem
Geschmack des Dichters des Scutum entsprechen, welchen
Russo „expressionistisch“ nannte. 6 Das Epitheton würde zu
ähnlichen Ausdrücken im Gedicht passen, wie etwa jenem,
__________
folgte niemand.
5
Janko, 305 (ad vv. 713–715) erfasste die evokative Kraft des Homerischen Ausdrucks und näherte sich unserem Verständnis des Ausschnitts –
obwohl er das Konzept nicht ausführlich behandelt hat –, wenn er schreibt,
dass “the ground traditionally flows with blood in the mêlée” und dass “the
‘dark’ hilts and earth evoke the colour of the blood”.
6
Russo, 7–22.
ipabo_66.249.78.36
66
Sara Chiarini
welcher für die Pfeile des Herakles in der Bewaffnungsszene
benutzt wird: ὀιστοὶ || ῥιγηλοί, θανάτοιο λαθιφθόγγοιο δοτῆϱες
(Vv. 130–131).
Aufgrund der relativen Einfachheit, mit welcher der Sinn des
Wortes μελάνδετος in einer homogenen Perspektive erklärt
werden kann, stellt sich sogleich die Frage, wodurch die Suche
nach dem hinter dem Epitheton versteckten archäologischen
Datum ausgelöst wurde.
Die Antwort setzt sich aus zwei Überlegungen zusammen:
erstens, hat die Tatsache, dass μελάνδετα am O 713 neben
κωπήεντα erscheint, ein anderes Epitheton des Schwertes,
welches „mit Degenglocke ausgestattet“ bedeutet, in die Irre
geführt. Das hat manche veranlasst, μελάνδετα κωπήεντα unterzuordnen, und sie nicht weiterhin als zwei gleichwertige
Epitheta von φάσγανα zu betrachten. Dass diese letzte Interpretation die richtige ist, bestätigt der Vers in seiner Gesamtheit
selbst schon, da er aus vier selbständigen und gleichwertigen
Attributen der Schwerter besteht.7
Zweitens, hat die spätere Bildung von Adjektiven wie
σιδηϱόδετος, 8 χϱυσόδετος, 9 χαλκόδετος, 10 λινόδετος 11 und ἐλεφαντόδετος12 Anlass geboten, auch für μελάνδετος nach einem
stofflichen Zusammenhang zu suchen. Aber μέλας darf nicht auf
die gleiche Art wie klare Materiesubstantive (σίδηϱος, χϱυσός
usw.) bewertet werden.
Abschließend soll hier nochmals der Vorschlag, das Epitheton
μελάνδετος im Sinne von „von schwarzem Blut umschlossen“
__________
7
Das Epitheton κωπήεις ist bei Homer stets semantisch unabhängig, wie
seine zwei anderen Okkurrenzen beweisen (Π 332, Y 475).
8
Erst nachweisbar in Bacch. Fr. 4, 69 Maehler und Hdt. IX 37, 2.
9
Alc. Fr. 350, 2 Lobel-Page; Soph. El. 837. Im Alkaios Fragment liest
man von der chrysoelephantinischen λαβά eines ξίφος (ἐλεφαντίναν […]
χϱυσόδεταν). Dieser Ausschnitt könnte das Missverständnis geschürt haben.
10
Aesch. TrGF I F 57, 6; Soph. Ant. 945; Strab. X 3, 16, 9.
11
Tim. PMG Fr. 791, 4; Eur. IT 1043; Aristoph. Nu. 763.
12
Eur. IA 587; Aristoph. Av. 219.
67
oder „vom schwarzen Tod umhüllt“ zu übersetzen, bekräftigt
werden.13
Literatur
Groeneboom, P. (hrsg.) (1938): Aeschylus’ Zeven tegen Thebe. Met Inleiding,
critische Noten en Commentaar, Groningen, J. B. Wolters’ UitgeversMaatschappij.
Helbig, W. (1884): Das Homerische Epos aus den Denkmälern erläutert,
Leipzig, B. G. Teubner.
Hutchinson, G. O. (ed.) (1985): Aeschylus Septem contra Thebas, Oxford,
Clarendon Press.
Janko, R. (1992): The Iliad: a Commentary. Volume IV: Books 13–16,
Cambridge, Cambridge University Press.
Leaf, W. (hrsg.) (2002): The Iliad, Milano, Vita e pensiero, (Nachdruck der
Herausgabe1900‒2²).
Liddel, H. G. - Scott, R. - Jones, H. S. (1996): A Greek English Lexicon,
Oxford, Clarendon Press.
Mastronarde, D. J. (hrsg.) (1994): Euripides Phoenissae, Cambridge,
Matz, F. - Buchholz, H.-G. (hrsg.) (1967–90): Archaeologia homerica: die
Denkmäler und das frühgriechische Epos, Göttingen, Vandenhoeck &
Ruprecht.
Russo, C. F. (hrsg.) (1968²): Hesiodi Scutum, Firenze, La Nuova Italia.
Snell, B. et alii (edd.) (1979‒2010): Lexikon des frühgriechischen Epos,
Göttingen, Vandenhoeck & Ruprecht.
Trümpy, H. (1950): Kriegerische Fachausdrücke im Griechischen Epos.
Untersuchungen zum Wortschatze Homers, Freiburg in der Schweiz,
Paulusdruckerei.
Willink, C. W. (hrsg.) (1986): Euripides Orestes, Oxford, Clarendon Press.
__________
13
Es wird Mastronarde, 451 und Hutchinson, 49 insofern nicht zugestimmt, als nach ihrer Darstellung der zweite Teil des Wortes (-δετος) ohne semantische Entwicklung geblieben sei. Sie lassen de facto das Epitheton wie
μέλας gleichbedeutend sein. Das Bild des schwarzen auf der Klinge geronnenen Opferblutes ist, meines Erachtens, jedoch genau in der Bedeutung von
μελάνδετος eingeschlossen. Es handelt sich dann um etwas viel konkreteres
und präziseres als das vage μέλας. Wenn die Übersetzung den mit der Idee
des Hüllens verbundenen Wortbestandteil vernachlässigt, wird die poetische
Wirksamkeit des Epithetons stark verringert.
ipabo_66.249.78.36
A constructionist approach to the potential
optative in classical Greek drama
By ANNEMIEKE DRUMMEN, Heidelberg
Abstract: The ancient Greek optative mood accompanied by the modal
particle ἄν, usually called the ‘potential optative’, has several possible
interpretations, dependent on the context. This paper presents a
construction-grammar description of the potential optative in classical
Greek drama. Following Boogaart (2009), it is argued that a constructionist approach combines the advantages of both a monosemy and a
polysemy view, without taking over their shortcomings. The potential
optative can be represented as a construction, i.e. a pairing of form
and meaning (see Croft & Cruse 2004). The features of this higherlevel construction are inherited by its daughter constructions, which,
however, also display more specific elements of form and meaning.
Construction grammar explains at the same time what all potential
optatives in the corpus have in common, how they differ, and why
they may receive such widely different interpretations.
1. Introduction
The combination of the verbal optative mood and the modal
particle ἄν in classical Greek1 is usually described as expressing
some kind of potentiality, with several other uses. In this
introductory section, I will summarize a few of these descriptions, and suggest some questions arising from them. Section 2
presents the theoretical background of my own description,
which is applied to the potential optative in section 3. Subsequently, section 4 describes all characteristics of the ‘daughter’
constructions in which the potential optative participates. Section 5, finally, will give a summary and conclusions.
Let us start with a brief overview of existing descriptions of
__________
1
This study is confined to the potential optative in main clauses, because
the optative in subordinate clauses frequently has other functions as well.
Glotta 89, 68–108, ISSN 0017–1298
A constructionist approach to the potential optative in classical Greek drama
69
the potential optative. First, Paley (1881: 4) notes, in his description of ἄν: ‘[c]onstructed with the optative (aorist or present) the
particle expresses the probable results of a certain condition
being fulfilled’. This construction, he remarks (p. 11), ‘is often
used as a mild or polite command or request.’ Some condition
such as ‘if you wished to please me’ may be implied in such
cases.
According to Kühner & Gerth (1963 [1898–1904]: 225–34),
the optative may be used to express a judgment or opinion, as
well as a desire. The potential meaning of the optative is used,
they argue, to present a state of affairs as ‘eventuell möglich’, or
to express a subjective uncertainty about the reality of a state of
affairs. Kühner & Gerth point out, furthermore, that in Attic the
potential optative may be exploited to give firm opinions the
colour of doubt. The construction can also be a mild expression
of will, or a milder form of a request. Finally, the authors mention that in tragedy, sentences starting with πῶς and containing a
potential optative have developed into a fixed formula for
wishing.2
Similarly, Schwyzer & Debrunner (1950: 327–9) mention
indirect wishes, mild requests and weakened assertions among
the uses of the potential meaning of the optative mood.
Slotty (1915: 82–3), furthermore, notes that the optative with
ἄν expresses all shades of potentiality. Beside the potential
sense, he recognizes an interpretation as an expression of will
(pp. 76–7, 93–7). The volitive optative usually involves negated
first person forms. In a second or third person, furthermore, this
nuance may express an order. Contrary to the view of Kühner &
Gerth, Slotty does not consider these orders to be milder than
imperative forms. He also disagrees with Kühner & Gerth on the
point of the interpretation of πῶς-questions as wishes. Slotty
argues (pp. 90–3) that the value of the potential optative in questions remains the same as in assertions. Therefore, he thinks a
πῶς-question should never be equated with an expression of wish.
__________
2
The use of a potential optative as a wish in a sentence starting with πῶς
is also noted by Moorhouse (1982: 231), who focuses on Sophocles only.
ipabo_66.249.78.36
70
Annemieke Drummen
Rijksbaron (2002: 41) writes that the classical optative
accompanied by ἄν ‘presents the realization of the state of
affairs as possible’ and ‘often serves as a cautious variant of the
indicative’. When accompanied by οὐ, the aorist optative with
ἄν ‘expresses an emphatic negation’, according to Rijksbaron:
‘it is not even possible that the state of affairs should occur’.
Beside these nuances, the second person, especially in the present stem, often serves as ‘a cautious variant of the imperative’
(p. 42).
The description of the optative mood by Willmott (2007:
113–52) should also be mentioned, although her investigation
involves Homeric Greek only. In conditional clauses, she writes
(p. 124), the optative expresses ‘negative epistemic stance’. This
means that the speaker views the state of affairs as ‘unreal’, but
not necessarily as unlikely (pp. 120–4). Willmott also mentions
(p. 118) the frequent expression of strong affirmation or denial
in this respect. Subsequently, two other optative meanings are
called into being: ‘dynamic’ and ‘objective deontic’. The former
(pp. 138–44) refers to inherent abilities of the subject. The latter
is described as a constative report of the existence of an
obligation (pp. 144–9). This means that the speaker does not
give an order, but points out that the state of affairs is a duty,
usually for the addressee.3
Thus, the potential optative appears to have several different
interpretations: the probable results of a condition being fulfilled, a mild request, a subjective uncertainty, an indirect wish,
a weakened assertion, all shades of potentiality, a cautious version of the indicative, an emphatic negation, negative epistemic
stance, and dynamic possibility. This observation leads us to
three related questions. First, what do these interpretations have
__________
3
I fail to see why the objective deontic category would be needed. Willmott does not explain in what way the Homeric cases cited are objective. In
all eight examples, the speaker is involved, in my view, which diminishes the
objective character of the utterances. Moreover, they can all be adequately
explained without assuming an objective deontic meaning. Cf. Allan (2010:
302), who has similar objections against Willmott’s objective deontic
category.
71
in common? Second, how do they differ from each other? And
third, how may we decide which interpretation is the most appropriate one for each instance? As has been shown by Boogaart
(2009), a monosemous approach may answer the first question,
whereas a polysemous view is most suitable for dealing with the
second one; however, construction grammar is able to illuminate
all three questions at the same time. I will argue that Boogaart’s
view is applicable to the description of the classical Greek
potential optative as well.
The corpus for this investigation consists of 230 potential
optatives from three comedies and six tragedies: Aristophanes’
Acharnenses, Ranae, and Vespae, Aeschylus’ Agamemnon and
Persae, Euripides’ Andromache and Hippolytus, and Sophocles’
Ajax and Antigone. The lively, diverse dialogues of drama texts,
displaying a variety of speaker intentions, can tell us much about
the possibilities of verbal communication.4
2. Monosemy versus polysemy versus construction grammar
The above-mentioned paper by Boogaart (2009) discusses the
advantages and problems of monosemy and polysemy, in particular with regard to a Dutch modal verb. He proposes a
constructionist approach instead, which combines the advantages but does not display the problems of both other theories. I
will apply the same strategy to the Greek potential optative.
First, in a monosemous account, it is believed that a certain
form, such as the combination of an optative mood and the
particle ἄν, has only one meaning. Possible other senses are seen
as belonging to the pragmatics of the expression, whereas only
its core meaning is contributed by its semantics. Monosemy is
especially suitable for showing what all different interpretations
of the potential optative have in common. It does not clarify,
however, how exactly they may differ, and which interpretation
__________
4
My observations and descriptions hold for these 230 instances only; I
am not arguing anything regarding the potential optative in other texts.
ipabo_66.249.78.36
72
Annemieke Drummen
will be the most appropriate in which context. Neither does this
approach explain how a speaker may know for which nuances
exactly a certain form can or cannot be used.
Polysemy, on the other hand, considers one form to have
possibly more than one distinct meaning. However, one meaning
may be more prominent or more prototypical, because it is more
frequent, or because it is, diachronically, the sense from which
one or more of the other senses derived. The polysemous approach shows how the different interpretations differ from
another: they are based on different meanings of the same form.
However, it is not specified in this model which contextual
elements make clear which meaning is the intended one in each
case.
As a third model, Boogaart (2009) applies construction
grammar, as described by Croft & Cruse (2004).5 In this
approach, all linguistic elements are seen in terms of constructions: symbolic pairings of form and meaning. Constructions can
be visualized as in Figure 1.
Figure 1. A construction (Croft & Cruse 2004: 258)
__________
5
Construction grammar started from the ideas of Fillmore, Kay, &
O’Connor (1988), and was further developed by, among many others, Croft
(2001) and Goldberg (1995). It has become an entire subfield of linguistics,
comprising several subtheories. See Östman & Fried (2004) for an overview
of the theory’s historical and intellectual background.
73
On the ‘form’ pole, the relevant phonological, morphological,
syntactic, prosodic and discourse information is represented. For
the potential optative, this would be the specific verbal mood
(morphological) and the co-occurence of ἄν (syntactic). Other
information, such as the person of the verb, the relative order of
the verb and ἄν, or the specific type of discourse, is not relevant
for the meaning of this construction, but might be relevant in
other constructions. On the ‘meaning’ pole, not only semantic,
but also discourse-pragmatic information can be represented. In
construction grammar, there is no strict distinction between
semantics and pragmatics, because it is believed that linguistic
information on any descriptive level can be represented as
constructions. Therefore, I will simply speak of ‘form’ and
‘meaning’ elements, regardless of the descriptive level.
Such constructions are considered to be organized in networks (Croft & Cruse 2004: 262‒5; Boogaart 2009: 230), in
which so-called ‘daughter’ constructions inherit all features, of
both form and meaning, of their ‘parent’ constructions. For
example, a negated first-person potential optative will show all
features of its possible parent constructions ‘negated finite
verbs’, ‘verbs in first person’, ‘optative mood with ἄν’, and so
on. On top of that, such form may also display features of its
own, which would make it into a separate construction.
I will show in this paper that we can indeed identify several
constructions in which the potential optative participates. A
constructionist approach, I will argue, is helpful to clarify why
this modal expression may have so widely different meanings,
and how we may distinguish the most suitable interpretation for
each instance.
ipabo_66.249.78.36
74
Annemieke Drummen
3. The potential optative as a network of constructions
The potential optatives in the corpus share two aspects of form:
the optative mood, and the presence of ἄν.6 Thus, in a constructionist view, they will normally also share some aspects of
meaning. In my view, this shared meaning can be described as
epistemic possibility, to be paraphrased as: ‘it is possible
(according to the speaker) that the state of affairs obtains’.
According to the presentation of a speaker, it is possible that the
state of affairs mentioned obtains; it is also possible that it does
not. The following example may make this meaning clear:
(1) γένοιτο μέντἂν πᾶν θεοῦ τεχνωμένου
‘Indeed anything can happen if a god contrives it.’7 (S. Aj. 86)
According to the speaker, it is possible that anything happens if
a god brings it about; it is however not certain. Such epistemic
modality has scope over the whole proposition, as opposed to
the more limited proposition-internal scope of dynamic modality.8 Epistemic modality specifies a subjective attitude of the
speaker towards the whole proposition, whereas dynamic modality is ‘participant-oriented’: it expresses a relation between a
participant and a state of affairs.
The epistemic possibility meaning is closely tied up with
conditional contexts. When the occurrence of a state of affairs is
dependent on a condition, this means it is not certain that it will
take place. In other words, dependency on a condition means
that a state of affairs does not necessarily occur, and, therefore,
that it is presented as no more than possible. Consequently, even
__________
6
There are 9 exceptions without ἄν in the corpus of 230 potential optatives, but in classical Greek its presence is the rule. The potential optatives
without ἄν are: Ar. Ra. 574; V. 72 (second optative); A. Ag. 620, 1049
(second optative); E. Andr. 929; Hipp. 211, 1186; S. Ant. 605, 1157 (second
optative).
7
The translations of the Ajax cited are those by Lloyd-Jones (1994a). For
clarity, the potential optatives are in bold in all examples.
8
For explanation of the differences between dynamic and epistemic modality, see e.g. van der Auwera & Plungian (1998: 81–2); Nuyts (2001: 193–4).
75
when no condition is explicitly mentioned, a potential optative
usually implies a conditional context.9
To see epistemic possibility as the meaning of the potential
optative makes sense in all 230 cases investigated. Of all
possible nuances expressed, only epistemic possibility is not
restricted to specific contexts. Furthermore, this view is in
accordance with the existing theories on the potential optative in
classical Greek (see § 1). In all descriptions of the optative with
ἄν, notions as ‘possibility’ or ‘potential’ are central. Thus, I
think we can describe the potential optative in classical Greek
drama as a construction which has on the ‘form’ pole no more
than the optative mood with ἄν, and on the ‘meaning’ pole the
notion of epistemic possibility. This construction can be considered as parent construction for several ‘daughters’, which inherit
its form and meaning, but also add more specific aspects of their
own. These constructions will be discussed in detail in the next
sections.
First, some specific contextual features – part of the ‘form’
pole – block any other interpretation than epistemic possibility
only. Therefore, this special daughter construction displays only
extra aspects of form with respect to the parent construction, no
additions to the meaning. Still, I think it is useful to distinguish
this daughter construction, because it differs from its ‘sisters’ in
specific form aspects.
Second, in specific contexts, a small-scope nuance of nonepistemic possibility is present alongside the epistemic meaning.
It can be paraphrased by: ‘(the subject) is able to carry out the
state of affairs’. I call this nuance ‘participant-oriented possibility’, because this is a relatively neutral term showing that a
participant is involved in the possibility expressed.10 In all cases
__________
9
The relation between protasis and apodosis, however, is not presented as
possible: the state of affairs in the apodosis arises as a certain consequence
from the condition described in the protasis. Cf. the similar observation of
Willmott (2007: 118) on the Homeric optative. Furthermore, a speaker may
not only use a subordinate clause (usually introduced by εἰ) for the expression of a condition, but also a participle, an adverb, or an adjective.
10
Similar notions, such as ‘dynamic modality’ (Nuyts (2001: 193; 2006:
ipabo_66.249.78.36
76
Annemieke Drummen
expressing this meaning, the ‘parent’ meaning of epistemic
possibility is present at the same time. The construction of
participant-oriented possibility has in its turn another, more
specific, ‘granddaughter’ construction: the expression of an
indirect wish.
Third, the potential optative may be used to mitigate the
illocutionary force of an utterance. This construction occurs in
direct and indirect requests, assertions, and promises. In this
case, specific properties of the relation between speaker and
hearer are part of the ‘form’ pole of the construction.
Fourth, we can distinguish the construction of a strengthened
refusal, when the potential optative is combined with other specific contextual features. Although the meanings of mitigation
and strengthening are diametrically opposed to each other, they
can both be derived from the parent construction’s meaning of
epistemic possibility. Describing the different uses of the potential optative in terms of different constructions is, in my view,
the key to the explanation of its seemingly contradictory nuances.
4. Daughter constructions of the potential optative
4.1 Epistemic possibility only
Since all potential optatives in the corpus express epistemic
possibility, in my view, no extra form aspects are necessary to
__________
2–4); Willmott (2007: 138–43)), ‘root possibility’ (Bybee, Perkins & Pagliuca
(1994: 178); Coates (1983: 145); Willmott (2008: 244–7)), and ‘facultative
modality’ (Hengeveld (2004: 1193–4)) are less clear, in my view, and are
often used in a more narrow sense than I am aiming at here. The term
‘participant-oriented possibility’ is meant to equal the categories of
‘participant-internal possibility’ and ‘participant-external possibility’ of van
der Auwera & Plungian (1998: 80–4) together. In the terminology of Hengeveld (2004: 1193–5), my term encompasses both ‘facultative participantoriented modality’ and ‘facultative event-oriented modality’. In short, this
nuance involves inherent abilities (participant-internal possibilities) as well as
circumstantial conditions (participant-external possibilities) enabling or
disabling a participant to engage in a state of affairs.
77
indicate that this meaning is intended. Some cases, however, are
accompanied by elements that block or weaken the interpretation of extra nuances on top of this meaning. Consequently,
these potential optatives will be interpreted as conveying epistemic possibility, and no more than that. Because these instances
show specific formal aspects, I consider them a separate
daughter construction.11
One of these formal features blocking the interpretation of
other nuances is the use of the adverbs ἴσως and/or τάχα, or the
particle που (all meaning ‘perhaps’, τάχα also ‘probably’).12 As
these words indicate the speaker’s uncertainty about the occurrence of a state of affairs, their function is very similar to that of
the potential optative itself. Therefore, they are harmonic with
the verbal construction.13 The adverbs or particle reinforce the
meaning of epistemic possibility, and therefore rule out the other
daughter constructions.
Another factor that leads to an interpretation as epistemic
possibility only, is a lack of emphasis on the subject of the verb,
which blocks an extra interpretation as participant-oriented
possibility (see § 4.2). The absence of emphasis on the subject is
especially conspicuous in the case of generic subjects, i.e. those
__________
11
I have classified 109 cases as conveying epistemic possibility only: Ar.
Ach. 801, 921, 924, 925, 940, 944, 1195; Ra. 81, 134, 533, 573, 574, 575,
1149, 1439, 1448, 1450, 1458, 1461, 1531; V. 72 (first optative), 72 (second
optative), 158, 277a, 348 (first optative), 348 (second optative), 350, 511,
928, 1159, 1176, 1367, 1456; A. Ag. 38, 323, 340, 347, 366, 1049 (second
optative), 1199, 1233, 1328, 1508, 1659; Pers. 212, 234, 438, 786, 789; E.
Andr. 351, 463, 679, 904, 929, 975, 1165; Hipp. 90, 227, 349, 472, 654, 796,
1017, 1062, 1063, 1154, 1186; S. Aj. 86, 155, 156, 171, 213, 265, 345, 378,
409, 456, 469, 524, 526 (second optative), 551, 692, 779, 816, 921, 963,
1009, 1043, 1074, 1075, 1122, 1123, 1130, 1149, 1247, 1305, 1330, 1342,
1344, 1371; Ant. 240, 505, 680, 686, 739, 884, 912, 926, 1194.
12
Ἴσως is found in Ar. Ra. 533, 1448; A. Ag. 1049 (with second optative); S. Aj. 963; τάχα is found in Ar. V. 277a; S. Aj. 171, 345, 779, 1043,
1149; both ἴσως and τάχα are found in Ar. V. 1456; S. Aj. 692; που is found in
S. Aj. 469, 1009.
13
See Coates (1983: 137–8) on harmonic modal adverbs in English. I do
not consider the adverbs pleonastic with the potential optative, as Slotty
(1915: 90) does.
ipabo_66.249.78.36
78
Annemieke Drummen
that do not have a unique referent, when they are used without
an identifiable referent being known.14
Furthermore, the lexical meaning of some states of affairs
may contribute substantially to the prominence of the epistemic
meaning. The meanings of verbs such as γίγνομαι, ‘become’,
‘happen’, are incompatible with the implication of participantoriented possibility and the strengthening use, because they do
not normally denote a controllable state of affairs.15 An example
is:
(2) φϱάσαι θέλω σοι πϱῶτα τἀμαυτοῦ· τὸ γὰϱ
πϱᾶγμ᾽ οὔτ᾽ ἔδϱασ᾽ οὔτ᾽ εἶδον ὅστις ἦν ὁ δϱῶν,
οὐδ᾽ ἂν δικαίως ἐς κακὸν πέσοιμί τι.
(‘First I want to tell you about myself; I did not do the deed,
nor did I see who did, and I could not with justice come to any
harm.’16 S. Ant. 238–40)
In this context, the messenger anxiously tries to prevent Creon
from punishing him. He gives some arguments for his innocence, in the hope that he may persuade the king that it would
not be possibe that he would ‘justly fall into any harm’. The subconstructions with the meanings of participant-oriented possibility and a strengthened refusal (see § 4.2 and 4.4) are ruled out
here because the speaker himself does not have power to decide
about the occurrence of the state of affairs.
__________
14
Cases in which an unidentifiable generic subject plays a role in
interpreting the potential optative as conveying epistemic possibility only,
are: Ar. Ach. 921, 940; Ra. 1458; A. Ag. 1328; E. Andr. 929; Hipp. 1186; S.
Aj. 86, 155, 156, 456, 524, 1149; Ant. 884, 912.
15
The influence of the lexical meaning of the state of affairs can be noted
in: Ar. Ach. 1195 (γίγνομαι); Ra. 1149 (εἰμί), 1439 (φαίνομαι); V. 350 (εἰμί),
928 (δύναμαι); A. Ag. 347 (γίγνομαι), 1199 (γίγνομαι), 1508 (γίγνομαι); Pers.
212 (γίγνομαι), 234 (γίγνομαι), 438 (γίγνομαι), 789 (πϱάσσω ἄϱιστα); E.
Andr. 351 (βούλομαι), 463 (πϱάσσω κακῶς), 904 (εἰμί), 1165 (εἰμί); Hipp. 90
(φαίνομαι), 227 (γίγνομαι), 349 (εἰμί), 472 (πϱάσσω εὖ), 654 (εἰμί); S. Aj. 86
(γίγνομαι), 378 (γίγνομαι), 524 (γίγνομαι), 551 (γίγνομαι), 779 (γίγνομαι),
816 (γίγνομαι), 1247 (γίγνομαι), 1330 (εἰμί); Ant. 240 (πίπτω); 686
(δύναμαι), 912 (βλαστάνω).
16
The translations of the Antigone cited are those by Lloyd-Jones
(1994b).
79
Similarly, the passive voice will normally lead to a purely
epistemic interpretation as well, which is conspicuous in case of
a negated occurrence. Although the negation typically makes an
interpretation as participant-oriented possibility or strengthened
refusal more likely (see § 4.2 and 4.4), these interpretations are
canceled by the passive voice. Example (3) shows this
influence.17
(3) εἰ νῦν γε δυστυχοῦμεν ἐν τούτοισι, πῶς
τἀναντί ᾽ ἂν πϱάξαντες οὐ σῳζοίμεθ᾿ ἄν;
(‘If we’re faring poorly with the current bunch, how wouldn’t
we find salvation if we did the opposite?’18 Ar. Ra. 1449–50)
Because of the presence of the question word πῶς (‘how?’) and
the negation oὐ, an interpretation as participant-oriented possibility (see § 4.2) becomes prominent in (3). However, the passive voice of σῳζοίμεθα (‘we would be saved’) makes this
interpretation less likely again. As there is normally no emphasis
on the subject in case of a passive verb, the relation between
subject and state of affairs is not relevant in this context. Moreover, a subject cannot influence a state of affairs of which he is
the Patient: passive states of affairs are not controllable. The
potential optative in (3) thus expresses large-scope epistemic
possibility only.
The presence or absence of negation is important in itself.
Although a purely epistemic reading is not excluded when the
utterance is negated, negation does usually favor the constructions conveying participant-oriented possibility or strengthening.
This can be explained by the emphatic effect of a negation:
saying ‘it is not possible that the state of affairs occurs’ is more
emphatic than saying ‘it is possible that the state of affairs
occurs (but it is also possible that it does not)’. By negating the
__________
17
Negated passive cases conveying epistemic possibility only, are: Ar. Ra.
1450; A. Ag. 340; Pers. 786; E. Hipp. 90; S. Aj. 1074, 1075, 1342; Ant. 680.
Although the form in A. Ag. 340 is strictly speaking active (ἀνθαλοῖεν), it
belongs to a passive verb (ἀνθαλίσκομαι, ‘to be captured in turn’) and has a
passive meaning.
18
The translations of the Ranae cited are those by Henderson (2002).
ipabo_66.249.78.36
80
Annemieke Drummen
epistemic possibility, the uncertainty is removed from the
utterance. Consequently, a speaker may convey emphasis on
proposition-internal relations, as is the case with the participantoriented possibility and strengthening nuance (see § 4.2 and
4.4). When a negated potential optative expresses epistemic
possibility only, other contextual features are present to make
this meaning prominent, such as an unidentifiable generic
subject or the lexical meaning of the verb.
In short, the features that I consider the ‘form’ pole of this
daughter construction are a non-controllable state of affairs,
unidentifiable generic subjects, the passive voice, and the cooccurrence of ἴσως, τάχα, or που. The presence of negation, in
contrast, is less compatible with a purely epistemic reading
without any extra nuance.
4.2 Participant-oriented possibility
Another daughter construction conveys a meaning of participantoriented possibility, next to the epistemic meaning of the parent
construction.19 This nuance is semantically different from epistemic possibility, because participant-oriented possibility expresses a proposition-internal relation between a participant and a
state of affairs. In the case of the potential optative, this participant is the grammatical subject, and the state of affairs concerned is indicated by the verb and its complements. Consider
an English example of participant-oriented possibility:
(4) Stephen can run faster than Michael.
Without any special context, a proposition-internal relation
between a participant (the subject, Stephen) and a state of affairs
__________
19
I interpret 54 cases as belonging to the daughter construction of
participant-oriented possibility: Ar. Ach. 307, 314, 797, 803, 918, 991; Ra.
68, 96, 296, 583, 906, 1406; V. 166, 172, 212, 726, 1195; A. Ag. 280, 620,
622, 1021, 1163, 1341, 1376, 1450, 1565; Pers. 243, 267, 430, 632; E. Andr.
85, 335, 596; Hipp. 209, 211, 345, 480, 557, 677, 961, 981; S. Aj. 391, 518,
537, 768, 917, 1051, 1137, 1262; Ant. 40, 552, 605, 668, 954.
81
(running faster than Michael) is conveyed in (4) by use of the
auxiliary verb ‘can’. In an expression of epistemic possibility, by
contrast, the commitment of the speaker to the truth of the whole
proposition is expressed. (5) presents an example of epistemic
possibility.
(5) One can be robbed any time in this city.
In this case, there is no relation between the subject (one) and
the state of affairs (being robbed). Rather, the speaker expresses
that in his view, the whole proposition (one being robbed any
time in this city) is possible. The two modalities thus differ in
scope: participant-oriented possibility has a more narrow scope,
inside the proposition, whereas epistemic possibility has scope
over the whole proposition.20 Participant-oriented possibility can
be divided into participant-internal possibility (ability), and
participant-external possibility, depending on the enabling or
disabling factors in the specific context.21 When these factors are
inherent in the participant who is said to have or lack the possibility to carry out the state of affairs, we speak of participantinternal possibility, or ability.22 When the enabling or disabling
factors lie in the circumstances, outside the participant, the
modality is participant-external.23 As the daughter constructions
inherit all features of their parent construction, the meaning of
epistemic possibility is still present in this daughter construction
as well.
Incidentally, it is remarkable that most English translators
usually feel the presence of a participant-oriented nuance as
well. These instances are most often translated by ‘could’ or
‘can’, while for potential optatives without this meaning,
__________
20
See van der Auwera & Plungian (1998: 81–2).
See van der Auwera & Plungian (1998: 80–1). The categories of
participant-internal and participant-external possibility roughly correspond to
the ones of facultative participant-oriented modality and facultative eventoriented modality of Hengeveld (2004: 1194–5).
22
An example of participant-internal possibility is E. Andr. 596.
23
An example of participant-external possibility is S. Ant. 552.
21
ipabo_66.249.78.36
82
Annemieke Drummen
translators more often choose ‘would’, ‘will’ or some other
translation to render them in English. The relevant numbers are
shown in table (6).24
(6) Relation between participant-oriented possiblity meaning and
translation ‘could’/‘can’
translation
with participant-oriented
possibility meaning
without participant-oriented
possibility meaning
‘could’/‘can’
35 (65%)
other25
19 (35%)
total
54
27 (15%)
149 (85%)
176
As can be seen in the table, translators choose ‘could’ or ‘can’ in
65% of the cases belonging to the daughter construction of
participant-oriented possibility (according to my analysis). In
the translations of potential optatives without this meaning,
‘could’ or ‘can’ is used in no more than 15% of the cases.26 This
means that these specific contexts must have one or more
features that lead me as well as the translators to interpret them
as conveying a participant-oriented possibility nuance. After all,
‘could’ and ‘can’ denote a relation between the subject and the
state of affairs, whereas ‘would’ and ‘will’ – the most frequent
translations in general – emphasize the conditional value of the
potential optative.
What exactly in the contexts concerned is it that makes the
meaning of participant-oriented possibility prominent? In other
__________
24
The following translations were used in the countings: Aristophanes’
Acharnenses: Henderson (1998a); Ranae: Henderson (2002); Vespae: Henderson (1998b); Aeschylus’ Agamemnon: Young (1974); Persae: Smyth
(1963); Euripides’ Andromache and Hippolytus: Kovacs (2005); Sophocles’
Ajax: Lloyd-Jones (1994a); Antigone: Lloyd-Jones (1994b).
25
The category ‘other’ includes translations with ‘(woul)’d’, ‘(wi)’ll’,
other auxiliary verbs, several periphrastic expressions, the past tense, imperatives, and zero translations, which means that the potential optative is
translated just as an indicative.
26
The relation is statistically significant. From the data of table (6), it
follows that Χ2 = 51,367. The probability P (Χ2 ≥ 51,367) < 0,001. So with a
significance level of 5% (or even 1%), this result is significant. See Rice
(1995: 384–506, esp. 387–9).
83
words, which elements are added to the ‘form’ pole of this
daughter construction with respect to its parent? The most
important contextual feature of all these cases is that the state of
affairs is presented as controllable by the subject. This is a
necessary contextual feature for the interpretation as participantoriented possibility: all cases share it. This means that it is also
necessary that the subject is a constituent able to exert some
control: it has to be an animate agent, or at least has to be
presented as such by the speaker.27
Since the value of participant-oriented possibility involves a
relation between the subject and the state of affairs, some
emphasis on the subject helps to convey this meaning. Therefore, this nuance is never found with dummy subjects, and
seldom with vague or unidentifiable ones.28 In the case of a first
or second person, the emphasis on the subject is usually conveyed by explicitly mentioning it with a personal pronoun.29 An
example is shown in (7).
(7) Φι. (...) ἀποδόσθαι βούλομαι
τὸν ὄνον ἄγων αὐτοῖσι τοῖς κανθηλίοις·
νουμηνία γάϱ ἐστιν.
Βδ. οὔκουν κἂν ἐγὼ
αὐτὸν ἀποδοίμην δῆτ᾽ ἄν;
Φι. οὐχ ὥσπεϱ γ᾽ ἐγώ.
(‘I just want to take the donkey and its panniers out and sell
them. It’s market day. # Surely I could do that, couldn’t I? #
Not the way I would.’30 Ar. V. 169–72)
As do all occurrences, the potential optative in (7) expresses
__________
27
Indeed most cases have human beings, divinities, or animals as their
subject. Of all 54 instances, there are 5 exceptions. These cases, which I
interpret as bearing a participant-oriented possibility meaning but which do
not have an actual animate agent, are: Ar. Ach. 918; E. Hipp. 557, 961; S.
Ant. 605, 954. See below, example (8).
28
Cf. the findings of Heine (1995: 35–6, 46) with regard to German
modal verbs: impersonal and dummy subjects are linked to an epistemic
interpretation of such verbs, whereas forms in the first and second person are
typically associated with an interpretation as participant-oriented modality.
29
Cases with first or second person pronoun bearing a meaning of
participant-oriented possibility are: Ar. Ach. 314; Ra. 583; V. 172; E. Hipp.
345; S. Aj. 1137, 1262; Ant. 40, 552, 668.
30
The translation of the Vespae cited is that by Henderson (1998b).
ipabo_66.249.78.36
84
Annemieke Drummen
epistemic possibility. In this case, the epistemic meaning can be
paraphrased by ‘surely it is possible, in some circumstances, that
I sell it?’. This instance may, however, convey a meaning of
participant-oriented possibility at the same time. The explicit
mentioning of the subject with ἐγώ and its implied contrast with
the second person create emphasis on this subject. Therefore, a
hearer or reader can be led to the interpretation in which the
relation between subject and state of affairs is relevant: a
participant-oriented interpretation. A paraphrase of this nuance
may run: ‘certainly I would have the possibility/ability to sell
it?’. In my view, this interpretation is more relevant in this
specific context, which contrasts the capacities of speaker and
addressee. It is not only relevant that there exists a possibility
that Bdelycleon sells the donkey: it is even more relevant that he
has the participant-oriented possibility (the ability) to do so.
When the subject is a third person, the use of a demonstrative
pronoun may similarly lead to the interpretation as participantoriented possibility; cf. (8).
(8) Νι.
ἐκ τῶν πολεμίων γ᾽ εἰσάγεις θϱυαλλίδας.
Δι.
ἔπειτα φαίνεις δῆτα διὰ θϱυαλλίδα;
Νι.
αὕτη γὰϱ ἐμπρήσειεν ἂν τὸ νεώϱιον.
(‘You’re importing lamp wicks from hostile territory. # So
you’re actually exposing him because of a lamp wick? # This
could burn up the shipyard!’ 31 Ar. Ach. 916–8)
Here, the sycophant Nicarchus wants to prove that the lamp
wicks of the character Dicaeopolis are truly dangerous. By using
a potential optative, he asserts that he considers it possible, in
some circumstances, that a wick sets a shipyard on fire: largescope epistemic possibility. In this context, it is also relevant
that the wick is inherently able to start such fire: small-scope
participant-internal possibility. It is the pronoun αὕτη (‘that
(wick)’) that lends a clear emphasis on the subject of the
potential optative. Note that this instance is exceptional with
__________
31
The translations of the Acharnenses cited are those by Henderson
(1998a).
85
regard to its subject: a lamp wick is no animate agent, and,
consequently, cannot actively decide to carry out an action.
However, this is precisely the pun of the accusation: the lamp
wick is presented as an active criminal item, which Dicaeopolis
finds absurd.
Another type of context in which a participant-oriented
interpretation is often appropriate is the one of a question. When
a speaker uses an interrogative sentence, he seeks information
about a state of affairs. Usually he does not want to know about
the possibility of the occurrence of the state of affairs in some
circumstances, but about the possibility or ability of the subject
to carrry it out. Heine (1995: 26), writing on German modal
auxiliaries, mentions interrogative sentences in a similar way.
(9) might be a clarifying example.
(9) (...)
Δι.
καὶ σκέλος χαλκοῦν ἔχει;
Ξα. νὴ τὸν Ποσειδῶ, καὶ βολίτινον θάτεϱον,
σάφ᾽ ἴσθι. Δι. ποῖ δῆτ᾽ ἂν τραποίμην; Ξα. ποῖ δ᾽ ἐγώ;
(‘And does she have a brazen leg? # Yes indeed, and the other
one’s made of dung, I swear. # Where can I run to? # Where
can I?’ Ar. Ra. 294–6)
Xanthias and Dionysus are frightened by a monster nearby. Both
of their questions – the first containing a potential optative, the
second with ellipsis of a similar verb – are used as desperate
cries for help. A paraphrase showing the epistemic possibility
meaning is: ‘where is it possible that I turn myself to?’ On top of
this meaning, a participant-oriented possibility reading is relevant as well. This nuance is of importance for asking the question, because Dionysus wants to know about specific escape
possibilities for himself. Furthermore, the participant-oriented
interpretation of ἂν τϱαποίμην is reinforced by the occurrence of
the first person pronoun in Xanthias’ following question.
The participant-oriented possibility meaning in such interrogative sentences, more specifically with πῶς (‘how?’), may be
further exploited to express an indirect wish. We can consider
this a more specific ‘granddaughter’ construction, with its own
ipabo_66.249.78.36
86
Annemieke Drummen
‘form’ and ‘meaning’ poles. Strictly speaking, in a sentence
starting with πῶς, the speaker asks about the means to achieve a
state of affairs, or about the circumstances accompanying it. In
some cases, the interrogative force of a πῶς-sentence seems to
be transformed into an expressive one.32
By using a πῶς-question containing a potential optative, a
speaker questions an element of the possibility of a state of
affairs, often the means to accomplish it. In this way, he may
imply that he desperately wants the state of affairs to occur. In
other words, an utterance such as ‘how could I do X?’ may
imply ‘I wish to do X!’. To convey an indirect wish in this way,
the construction conveying epistemic possibility is necessary,
and so is even the daughter construction of participant-oriented
possibility, in which the relation between the subject and the
carrying out of the state of affairs is relevant. The ‘granddaughter’ thus inherits all ‘form’ and ‘meaning’ elements of
these constructions, but also adds more specific elements.33
Usually the verb is a first person, or the involvement of the
speaker is otherwise clarified in the context. After all, one
usually wishes for things which concern oneself. As wishes
often have a desperate character, because the speaker presents
the vehemently desired state of affairs as beyond his own power,
the contexts are usually quite emotional. An example of this type
is (10).
__________
32
Kühner & Gerth (1963: 234; see § 1) observe that questions with πῶς
are a common formula to express a wish indirectly. I agree with Slotty (1915:
90–3; see § 1) that a sentence starting with πῶς and containing a potential
optative is not semantically equivalent to a direct wish. However, this view
does not exclude the option of an interpretation of a πῶς-question as an
indirect wish, provided it is seen as dependent on specific contextual features.
For example, the πῶς-question with potential optative in A. Pers. 243 does
not express an indirect wish, because the speaker (the queen) is not directly
involved in the state of affairs, and only asks a range of questions.
33
I classify 7 potential optatives as indirect wishes: Ar. Ach. 991; V. 166;
A. Ag. 622; E. Hipp. 209, 211, 345; S. Aj. 391.
87
(10) ὦ Ζεῦ πϱογόνων πάτωϱ,
πῶς ἂν τὸν αἱμυλώτατον,
ἐχθϱὸν ἄλημα, τούς τε δισσάϱχας ὀλέσσας βασιλῆς,
τέλος θάνοιμι καὐτός;
(‘O Zeus, father of my ancestors, if only I could destroy the
craftiest of all, the trickster that I detest, and the two brother
kings, and at last die myself!’ S. Aj. 388–91)
Here, the character Ajax questions, when read literally, the
means by which it would be possible for him to die, after having
killed Odysseus, Agamemnon, and Menelaus. In the desperate
situation at hand, Ajax wants nothing but killing these enemies,
and then die himself. He desires the realization of the state of
affairs mentioned, rather than an answer to his question. This
context thus indicates that the interrogative sentence is in fact
meant as a wish, albeit in an indirect form. Note that, although
the verb θάνοιμι itself does not designate a typically controllable
action, the state of affairs as a whole does. Therefore, this
potential optative may belong to the daughter construction of
participant-oriented possibility. Such potential optative in a πῶςquestion does not convey a wish in all cases, however, for the
state of affairs mentioned obviously has to be something desirable for the speaker in order to point to this construction.
Another element of the context that increases the prominence
of the participant-oriented possibility implication is the lexical
meaning of some state of affairs. For example, when the verb is
εὑϱίσκω (‘find’), a participant-oriented sense will usually be
relevant in the utterance, as shown in (11).34
(11) Ἀν.
(...) θέλεις οὖν ἄγγελος σύ μοι μολεῖν;
Θε.
τί δῆτα φήσω χϱόνιος οὖσ᾽ ἐκ δωμάτων;
Ἀν.
πολλὰς ἂν εὕροις μηχανάς· γυνὴ γὰϱ εἶ.
(‘Will you then go as my messenger? # What shall I say to ex-
__________
34
A similar instance is E. Hipp. 480.
ipabo_66.249.78.36
88
Annemieke Drummen
cuse my long absence from home? # You will find many
ruses: you are a woman.’35 E. Andr. 83–5)
In this dialogue between Andromache and her servant, it is not
only relevant that it is possible, in some circumstances, that the
servant finds contrivances, but also that she has the participantinternal possibility (the ability) to do so. The lexical meaning of
the verb εὑϱίσκω plays a role in this interpretation: one can
succeed or fail in finding something, and one’s own abilities, as
well as one’s specific possibilities in a situation, usually influence this success.
The participant-oriented nuance may also be made manifest
by the wider context, e.g. an explicative clause. This is the case
in (11) as well: the ability reading of ἂν εὕϱοις is strengthened
by the subsequent clause γυνὴ γὰϱ εἶ (‘for you are a woman’).
The servant’s gender constitutes an obviously inherent factor
enabling her to be inventive (according to the character Andromache). Incidentally, allowing the participant-oriented nuance to
be present in our interpretation increases, at the same time, the
relevance of this explicative clause. Both the participant-oriented
possibility implication and the clause γυνὴ γὰϱ εἶ emphasize the
role of the servant as the agent in this state of affairs.
To sum up, the elements at the ‘form’ pole of the potentialoptative daughter construction conveying participant-oriented
possibility are the following. The most important aspects are a
controllable state of affairs, and, consequently, a subject that is
(presented as) an animate agent. Furthermore, there is often an
explicit emphasis on the subject in these contexts. The construction of participant-oriented possibility is especially compatible
with interrogative sentences. The lexical meaning of a state of
affairs, furthermore, may reinforce the interpretation as participant-oriented possibility as well, just as elements in the wider
contexts. Some of the instances in interrogative sentences
pragmatically convey an indirect wish, which can be seen as an
__________
35
The translations of the Andromache and Hippolytus cited are those by
Kovacs (2005).
89
even more specific (grand)daughter construction. In these cases,
the physical and emotional involvement of the speaker is somehow indicated.
4.3 Mitigation
The potential optative is also used in a daughter construction
conveying a mitigation, which I have found in 43 cases in the
corpus.36 By presenting the realization of a state of affairs as
possible, instead of factual, a speaker indicates or pretends that
he is not completely sure about what he says, or implies that the
state of affairs mentioned only obtains when some condition is
fulfilled. In this way, he may diminish the illocutionary force of
his utterance. Negated potential optatives cannot be used in this
way, because presenting a state of affairs as impossible in all
circumstances is, on the contrary, very strong. This leads to the
strengthening construction of the potential optative, discussed in
§ 4.4 below.
The parent-construction meaning of epistemic possibility is
present in all mitigation cases as well. However, allowing a
pragmatic effect such as mitigation into the ‘meaning’ pole of a
daughter construction shows us what a speaker may do with this
epistemic value in the specific interactional context. The mitigation use is also mentioned by many of the previous accounts of
the potential optative.37
Dependent on the context, mitigation may have the following
communicative effects. First, a request can be made less
disturbing for the addressee. Both a direct request (mitigated
with respect to an imperative), and an indirect request (mitigated
__________
36
The 43 mitigation cases are: Ar. Ra. 283, 431, 437, 1401, 1467; V. 484;
A. Ag. 252, 263, 319, 552, 739, 838, 896, 930, 1049 (first optative), 1394,
1578; Pers. 706; E. Andr. 645; Hipp. 89, 270, 336, 469, 842, 904; S. Aj. 88,
161, 185, 218, 377, 526 (first optative), 715; Ant. 80, 218, 314, 444, 476, 646,
652, 687, 1108, 1244, 1339.
37
The mitigation use or some equivalent is explicitly mentioned by Kühner & Gerth (1963 [1898‒1904]), Paley (1881), Rijksbaron (2002), and
Schwyzer & Debrunner (1950). See § 1.
ipabo_66.249.78.36
90
Annemieke Drummen
with respect to an indicative) may show this use. In an assertion,
secondly, a speaker may cover himself up against possible contradiction, or appear more polite by using a mitigating potential
optative. In promises, finally, the use may have a similar safeguarding effect as in assertions.
Mitigation in direct requests
In the case of a mitigated direct request, we find a secondperson potential optative.38 Inheriting the epistemic-possibility
meaning of the parent construction, this form presents the state
of affairs as possible in some circumstances. On top of that, the
utterance may function as a mild request, given the appropriate
contextual features: a controllable state of affairs, the real or
pretended desirability of the state of affairs for the speaker, and
the real or pretended social dominance of the speaker over the
addressee.39 In such cases, we can think of an ‘original’ version
with an imperative, which is made less disturbing for the addressee(s) by the mitigating potential optative. Presenting a state
of affairs as possible is, after all, less strong than directly
ordering someone to carry it out.40 An example of a mitigated
direct request is:
__________
38
I found 9 mitigated direct requests: Ar. Ra. 437, 1401, 1467; V. 484; A.
Ag. 1049 (first optative), 1394; S. Ant. 80, 444, 1339.
39
The last feature might seem surprising, since the potential optative is
more cautious than the imperative. However, the speaker still needs some
reason to expect that the addressee will probably comply to his request. When
the speaker wants to present himself as socially lower, a mitigated indirect
request is a more appropriate way of formulating a request (see below).
40
The mitigation use in direct requests is mentioned by Kühner & Gerth
(1963 [1898–1904]), Paley (1881), Schwyzer-Debrunner (1950), and Rijksbaron (2002). Slotty (1915: 95–7), on the other hand, disputes the claim of
these scholars that directive optatives would be milder than imperative forms.
The expression of orders did not arise from the potential meaning of the
optative, according to Slotty, but was original as well. However, in view of
the meaning of the parent construction as outlined in this paper, it makes
more sense to follow the former group of authors in seeing these potential
optatives as more cautious than imperatives. Using this construction, a
speaker presents the state of affairs as possible, which is less disturbing for
the addressee than an imperative, and leaves more room for refusal.
91
(12) ἄγοιτ᾿ ἂν μάταιον ἄνδϱ᾽ ἐκποδών
(‘Lead me out of the way, useless man that I am’, S. Ant.
1339)
These words are spoken by Creon, the king, to the chorus. He
has an obvious dominance over his addressees. However, he
only utters a mild request by using a potential optative instead of
an imperative. This choice can be explained by the fact that
Creon has just realized that he has made some large mistakes,
which have led to the loss of his son and wife. Not surprisingly,
he now feels very miserable, and does not deem himself worthy
to give very severe orders.
Two cases of the mitigating construction in direct requests
seem to form an even more specialized ‘granddaughter’, conveying a concessive sense. This is illustrated in (13). The extra
‘form’ element for this granddaughter is a contrast between the
second and the first person.41 The semantic meaning ‘it is
possible that you do X, but I will do Y’ may imply a sarcastic
encouragement such as ‘go on doing X, I do not care; I will do Y
anyway’.42 A speaker may thus express his simulated indifference towards the actions of his addressee, while attributing
significance to his own decision to act differently.
(13) Ἰσ.
ἐγὼ μὲν οὐκ ἄτιμα ποιοῦμαι, τὸ δὲ
βίᾳ πολιτῶν δϱᾶν ἔφυν ἀμήχανος.
Ἀν.
σὺ μὲν τάδ᾽ ἂν προὔχοι᾽, ἐγὼ δὲ δὴ τάφον
χώσουσ᾽ ἀδελφῷ φιλτάτῳ ποϱεύσομαι.
(‘I am not dishonouring them, but I do not have it in me to act
against the will of the people of the city. # You may offer that
excuse; but I shall go to heap up a tomb for my dearest
brother!’ S. Ant. 78–81)
As noted by Griffith (1999) ad loc., the character Antigone uses
the potential optative ἂν πϱοὔχοιο in a sarcastic way. Antigone
__________
41
I have found such concessive value in two cases: A. Ag. 1394 and S.
Ant. 80.
42
The English auxiliary verb ‘may’ similarly has both a directive and a
concessive use: see Coates (1983: 131–45).
ipabo_66.249.78.36
92
Annemieke Drummen
pretends not to mind that Ismene refuses to bury Polyneices,
while in fact she is quite angry about this dishonour. The
utterance can be roughly paraphrased by: ‘it is possible that you
offer that excuse, in some circumstances (viz. if you choose to
do so), but I will bury him’. I think the most relevant interpretation in this context is one of a concessive request: ‘please do so,
if you like; I don’t care’. Antigone seems to encourage Ismene to
behave as she does, in order to emphasize her own, more noble
decision. In this way, Antigone stresses that she will bury her
brother anyhow, whatever her sister’s opinion.
Mitigation in indirect requests
Not only direct requests (containing a second-person verb
denoting the desired action), but also indirect requests may be
mitigated by the use of a potential optative. Here, we can
imagine an indicative, instead of an imperative, as the ‘stronger’
variant. A speaker may for example use a first person form of a
volitive verb in such requests.43 By describing one’s wishes, one
may indirectly ask the addressee to carry out some specific act.
In these environments, it is more polite to mitigate the volitive
expression, as a more straightforward indicative leaves less
room for refusal. A similar politeness strategy can be discerned
in modern languages, for example in the English expression ‘I
would like’ as opposed to the blunt ‘I want’. Likewise, the
Greek potential optative has a conditional implication, which
creates the impression that the speaker is not totally sure about
his own desires. However, since people normally do know what
they themselves want, it is not the general value of epistemic
possibility that comes to the fore with these volitive verbs, but
the more specific, pragmatic value of mitigation. An example is
βουλοίμην ἄν in (14).
__________
43
11 indirect requests with mitigation nuance were found: Ar. Ra. 431; A.
Ag. 263, 319; E. Hipp. 89, 270, 842, 904; S. Aj. 377, 526 (first optative); Ant.
218, 1244. Out of these cases, 4 have a first person form of a volitive verb: A.
Ag. 319; E. Hipp. 270, 904; S. Aj. 526 (first optative).
93
(14) κϱαυγῆς ἀκούσας σῆς ἀφικόμην, πάτεϱ,
σπουδῇ·τὸ μέντοι πϱᾶγμ᾽ ἐφ᾽ ᾧ στήσας ἔχεις
οὐκ οἶδα, βουλοίμην δ᾽ ἂν ἐκ σέθεν κλύειν.
(‘I heard your cry and came in haste, father. But what it was
that brought forth your groan, I do not know but would gladly
hear from your lips.’ E. Hipp. 902–4)
Hippolytus addresses his father Theseus, using the potential
optative βουλοίμην ἄν in an indirect request. Hippolytus is
worried because of the cry he has heard, and does not want to
upset his father even more. Therefore, he phrases his request for
information in a cautious way. Had Hippolytus used the indicative βούλομαι here, the request would have been much directer,
and hence less polite. The mitigating potential optative is thus a
means for the speaker to be less challenging to the addressee’s
face. In this scene, Hippolytus’ polite formulations increase the
contrast with the attitude of his father, who turns out to be
furious with him.
Mitigation in assertions
When a mitigating potential optative occurs in an utterance
meant as an assertion, its value may bring about that it is more
difficult for the addressee(s) to contradict the speaker.44 In other
words, a speaker may use the mitigation nuance to cover himself
up against possible contradiction. This is especially apparent
when the speaker considers himself socially higher than his addressee, as in (15), where king Creon is speaking to a guard.
(15) ἐκ τῶν γὰϱ αἰσχϱῶν λημμάτων τοὺς πλείονας
ἀτωμένους ἴδοις ἂν ἢ σεσωσμένους.
(‘One sees more people ruined than one has seen preserved by
shameful gains.’ S. Ant. 313–4)
By using the potential optative ἴδοις ἄν, the speaker lends an
__________
44
I have interpreted 20 cases as mitigated assertions: Ar. Ra. 283; A. Ag.
252, 552, 739, 838, 896, 930, 1578; Pers. 706; E. Andr. 645; Hipp. 469; S.
Aj. 161, 185, 218, 715; Ant. 314, 476, 646, 652, 687.
ipabo_66.249.78.36
94
Annemieke Drummen
even more general applicance to the described state of affairs
than would have been the case with an indicative. After all, there
is a larger chance that an utterance is true when formulated as ‘it
is possible that X obtains’ than when formulated as ‘X obtains’,
however general the state of affairs may be. The epistemic
possibility meaning of the potential optative makes the assertion
less strong, but at the same time more generally applicable. In
this case, it is much more difficult for the guard to contradict the
assertion ‘it is possible, if some condition is fulfilled, that you
see (people ruined, etc.)’ than the assertion ‘you see (people
ruined, etc.)’. Because of his higher social position, on top of
that, Creon is even more assured to be protected from any
contradiction. In this way, he may pretend to be in possession of
great wisdom, while in fact saying very little. The general nature
of the state of affairs, and the ‘cautious’ formulation with a
potential optative, implying some condition, make it very hard
for the addressee to contradict the speaker.
The self-confident character Creon in Sophocles’ Antigone is
especially fond of this mitigation use. As noted by Griffith
(1999: 33), Creon has an ‘obsession with his own authority’, and
(p. 36) he ‘habitually starts out and ends his speeches with
generalizations’. I think Creon’s abundant use of mitigating
potential optatives (7 in total45) can similarly be seen as part of a
rhetorical strategy. He utters many very general expressions,
which his interlocutor cannot deny to be true. The character is
thus presented by Sophocles as someone with high self-esteem,
who uses several rhetorical strategies to be always proved right.
Besides the effect of covering oneself up against possible
contradiction, a mitigating potential optative in an assertion may
also be used in order to be polite. Whereas the first effect is
especially apt for socially high characters, the latter interpretation will be more appropriate when the speaker is in a lower
position than his addressee. Often a first-person verb of saying is
__________
45
The character Creon uses a potential optative to mitigate an assertion in
S. Ant. 314, 476, 646, and 652; he also mitigates a direct request in 444 and
1339, and a promise in 1108.
95
used with this extra nuance: ‘I would say X’ implies that one
would only say X under some specific circumstances, but in
reality one has of course already said it. An example is (16).
(16) πάϱαυτα δ’ ἐλθεῖν ἐς Ἰλίου πόλιν
λέγοιμ’ ἂν φϱόνημα μὲν
νηνέμου γαλάνας
(‘At first, I would say, there came to Ilium the spirit of unruffled calm,’46 A. Ag. 738–40)
When the chorus would have used an indicative here (λέγω ‘I
say’), it would have been less polite. With the potential optative
λέγοιμ᾽ ἄν, a condition is implied, such as ‘if I were asked’. At
the same time, contradiction is made less probable, because the
speaker implies not to be totally sure of this statement.
Mitigation in promises
Furthermore, I found three mitigating potential optatives in
promises: Euripides’ Hippolytus 336, Sophocles’ Ajax 88 and
Antigone 1108. Because the realization of the state of affairs is
presented as something no more than possible, and as something
somehow dependent on a condition, the speaker does not totally
commit himself to the content of his utterance. Therefore,
although the speaker places himself lower than the addressee by
complying to some order or suggestion, he also tries to safeguard himself against unexpected circumstances. This coveringup effect of the mitigating potential optative in promises is also
noted by Moorhouse (1982: 230). He describes this first person
use as conveying ‘intention, but expressed with some reserve or
reluctance.’ On the instance given in (17), Moorhouse remarks it
is spoken ‘reluctantly’ as well.
(17) Ἀθ.
Ὀδ.
σίγα νυν ἑστὼς καὶ μέν᾽ ὡς κυϱεῖς ἔχων.
μένοιμ᾿ ἄν· ἤθελον δ᾽ ἂν ἐκτὸς ὢν τυχεῖν.
__________
46
The translation of the Agamemnon cited is that by Smyth (1957).
ipabo_66.249.78.36
96
Annemieke Drummen
(‘Then stand in silence and remain as you are. # I shall
remain; but I wish I were not here.’ S. Aj. 87–8)
As it is the goddess Athena who has just ordered Odysseus to
stay where he is, he has no choice but to obey. However, master
of language as this character is, he formulates his promise to do
so in a clever way. The potential optative implies that he will
only stay in case some condition is fulfilled, so that Odysseus
both pleases Athena and leaves some room for himself to escape
at the same time.
4.4 Strengthening
Another pragmatic ‘daughter’ of the epistemic meaning involves
the strengthening of a specific type of negated utterances: refusals.47 Refusing is different from simply denying: the will of the
speaker is decisive in refusals, because they involve controllable
states of affairs and first person forms. Thus, using this type of
expression, a speaker commits himself to refrain from acting in
a certain way. The role of the potential optative is to strengthen
the refusal: it implies that the speaker would refuse in all possible circumstances or under all possible conditions. It is the
negation that explains how the notion of possibility may be used
both to mitigate (‘merely possible, in some circumstances’) and
to strengthen (‘impossible under all conditions’).
An example of a strengthening potential optative is:
(18) (a stranger asks Dicaeopolis for a glass of peace in exchange
for some meat)
Δι.
ἀπόφεϱ᾽ ἀπόφεϱε τὰ κϱέα καὶ μή μοι δίδου,
ὡς οὐκ ἂν ἐγχέαιμι χιλίων δϱαχμῶν.
(‘Take the meat back, take it back and don’t offer it to me! I
wouldn’t pour a drop for a thousand drachmas.’ Ar. Ach.
1054–5)
__________
47
I interpret 24 cases as strengthening a refusal: Ar. Ach. 403, 1055; Ra.
135, 581, 585, 695, 1222; V. 509; A. Ag. 275, 1130, 1665; E. Hipp. 950; S.
Aj. 477, 1093, 1117, 1339; Ant. 69, 70, 185, 188, 731, 1157 (first optative),
1157 (second optative), 1171.
97
In this scene, the character Dicaeopolis gets angry because many
people want to take advantage of his own good idea to have
peace. He therefore vehemently denies the possibility that he
would give some of it to the eager guest of this moment. The
‘parent’ meaning of the potential optative, epistemic possibility,
is pragmatically exploited to strengthen the refusal. ‘It is not
possible that I pour’ thus comes to convey ‘I would under no
circumstances (be willing to) pour’. In this case an extreme
condition is already mentioned: even when he would get a
thousand drachmas, Dicaeopolis claims, he would still refuse to
share his peace. The use of the potential optative implies that
this extreme scenario is just one of the cases in which he would
refuse: ‘I would not (be willing to) pour, viz. not (even) for a
thousand drachmas’. An indicative would only have conveyed:
‘I do not pour, not (even) for a thousand drachmas’. Because the
epistemic possibility meaning implies conditionality, the negated
potential optative yields a stronger refusal than a negated
indicative would have done.
Some of the descriptions discussed in section 1 also mention
this strengthening use, but they do not define it accurately.48 For
example, Rijksbaron (2002) calls these utterances both ‘emphatic
negation’ (p. 41) and ‘emphatic refusal’ (p. 48), without pointing
out any difference.
The contextual elements of person and negation are the extra
‘form’ elements of the strengthening daughter construction. A
negation is present in all cases, and I have nearly exclusively
found this interpretation appropriate in instances with a grammatical first person singular.49 Furthermore, Rijksbaron (2002:
41) points out that such ‘emphatic negation’ occurs only with
aorist optatives, a correlation which is confirmed by my
findings.50 This is understandable, because it is to be expected
__________
48
A nuance similar to the strengthening nuance is explicitly mentioned by
Rijksbaron (2002), Slotty (1915), and Willmott (2007). See § 1.
49
There are two exceptions, in which the involvement of the speaker is
shown indirectly: A. Ag. 1665 and S. Ant. 70.
50
Of all 24 cases, 21 are aorists, as against only 3 in the present stem. The
ipabo_66.249.78.36
98
Annemieke Drummen
that a single, completed state of affairs is more often refused
than a non-completed or habitual one. Besides, passive forms, or
verbs with stative Aktionsart normally do not yield an interpretation as a refusal, and therefore are not strengthened by the
potential optative.51 After all, one can logically refuse only
actions over which one has some power.
Apart from the factors of a first person singular and the
presence of a negation, the strengthening construction is
compatible with specific communicative situations. It typically
occurs in emotional contexts, where the speaking character is
angry or indignant. Therefore, it is also compatible with other
emphatic elements, such as a personal pronoun, lexical means or
specific particles.52 For example, in (19), the emphatic expression μὰ τὴν Δήμητϱα (‘by Demeter!’) and the particle γε are
added to the strengthened refusal.
(19) (...)
Δι.
ὑφέσθαι μοι δοκεῖ.
τὸ ληκύθιον γὰϱ τοῦτο πνευσεῖται πολύ.
Εὐ.
οὐδ᾽ ἂν μὰ τὴν Δήμητϱα φροντίσαιμί γε·
νυνὶ γὰϱ αὐτοῦ τοῦτό γ᾽ ἐκκεκόψεται.
(‘I think you should reef your sails; that oil bottle’s blowing
up a gale. # Quite the contrary, I’m not at all worried. This
time it’ll be knocked right out of his hand.’ Ar. Ra. 1220–3)
During a poetic battle, the character Aeschylus attacks the verses
__________
3 present optatives are: Ar. Ra. 695; A. Ag. 1665; S. Ant. 70. These three
cases with present aspect are atypical in other respects, too. In Aeschylus’
Agamemnon 1665 and Sophocles’ Antigone 70, the involvement of the
speaker is indicated differently from the grammatical subject. These cases
are, therefore, indirect refusals. In the third exception (Aristophanes’ Ranae
695), the verb ἔχω is used, which does not have an aorist form. This verb is
normally uncontrollable, but in combination with its complement μὴ οὐ
καλῶς φάσκειν ἔχειν I think we may interpret this utterance as a refusal.
51
Negated first person potential optatives which I do not consider as
strengthening refusals for these reasons are found in A. Pers. 786 (plural), E.
Hipp. 90 (plural), S. Aj. 1330, Ant. 240, 680 (plural). The fact that three of
these instances are plurals also makes them less suitable to be interpreted as
refusals.
52
Strengthened refusals with other emphatic elements (a personal pronoun, lexical means and/or specific particles) are: Ar. Ra. 695, 1222; V. 509;
E. Hipp. 950; S. Aj. 1339; Ant. 70, 185, 1171.
99
of Euripides with a fictional oil flask. However, the latter refuses
to give up, which he stresses by using a potential optative: ‘in no
circumstances would I (be willing to) consider it (viz. to give
up)’. I think an interpretation of this instance as refusal is more
relevant than as negation only, because it is Euripides’ own
decision to carry on with the battle. A potential optative, implying a refusal in all circumstances, is stronger than an indicative, which would have concerned the present situation only.
The pragmatic strengthening is made even more emphatic with
the swear-word and the particle γε, which demarcates the
relevance of the utterance to the verb form φϱοντίσαιμι and
conveys an even more emotional tone.
In short, all strengthening cases contain the following ‘form’
elements: a negation, the involvement of the speaker, and a
controllable state of affairs. Furthermore, this daughter construction is compatible with an emotional communicative situation,
and the co-occurrence of emphatic elements.
An additional element that stands out in relation to the daughter
construction of strengthening is the correlation with repeated ἄν.
Of the 24 cases classified as strengthening, six cases are
accompanied by two occurrences of the modal particle instead
of one.53 These constitute 25% of all strengthened refusals,
whereas only 13% of all potential optatives in the corpus as a
whole has double ἄν (viz. 29 of all 230 cases). To make a more
meaningful comparison, I have looked at the co-occurrence of
repeated ἄν and negation in general. The results are shown in
tables (20) and (21).
(20) The relation between negation and repeated ἄν
no repeated ἄν
without negation 157 (92 %)
with negation
44 (73 %)
repeated ἄν
13 (8 %)
16 (27 %)
total
170
60
__________
53
Strengthened refusals accompanied by double ἄν are: Ar. Ra. 581, 585;
V. 509; S. Aj. 1339; Ant. 70, 1157 (first optative).
ipabo_66.249.78.36
100
Annemieke Drummen
As can be seen in this table, repeated ἄν occurs in 16 cases of all
60 negated potential optatives. These constitute 27% of the
negated cases, against the much lower frequency of 8% with
double ἄν in cases without negation.
(21) The inverse relation between negation and repeated ἄν
no repeated ἄν
repeated ἄν
without negation with negation
157 (78 %)
44 (22 %)
13 (45 %)
16 (55 %)
total
201
29
Table (21) shows that the relation holds the other way around as
well: whereas the 44 negated cases make up 22% of the instances with single or omitted ἄν, as many as 55% of the cases with
double ἄν is negated. In short, the corpus data suggest a correlation between negation and repeated ἄν.54 How may this correlation be explained? As Slings (1992: 102‒5) describes it, ‘the
doubling of ἄν is connected with the complexity of the sentence’, which can be a syntactic or a pragmatic complexity. In less
specific terms, the same had already been remarked by Paley
(1881: 6), who writes: ‘[w]hen there are two emphatic words, or
when the verb follows long after the introductory ἄν, the particle
may be repeated.’ Although neither Paley nor Slings notes a
higher frequency of repeated ἄν in case of negation, I think their
explanation can be seen as including this correlation. That is,
Slings (1992: 103) explains pragmatic complexity as ‘a high information load: the quantity of words that by their very meaning
carry a certain emphasis, like question words and adjectives denoting quantitative intensity.’ No doubt negation is one of these
emphatic elements as well. Therefore, negated potential optatives are extra compatible with the repetition of ἄν, which may
__________
54
To test the significance of this correlation, I applied a χ2 test for
independence to the data of tables (20) and (21). From the data, it follows
that Χ2 = 14,56. The probability P (Χ2 ≥14,56) = 0,0001. So with a
significance level of 5% (or even 1%), this result is significant. See note 26.
However, a correlation between the strengthening use and repeated ἄν cannot
be indicated on the basis of my data.
101
increase the emphasis even more.55 Incidentally, this emphatic
character of negation also leads to its incompatibility with the
daughter construction of mitigation, as shown above. Therefore,
a mitigated interpretation is ruled out when a potential optative
is negated.
The ‘pragmatic complexity’ probably due to negation is
illustrated in (22). It shows the most beautiful corpus example of
a strengthened refusal twice accompanied by ἄν, but without a
complex syntax.
(22) (Dionysus, dressed up as Heracles, wants Xanthias to take
over his costume)
Δι.
κάκιστ᾽ ἀπολοίμην, Ξανθίαν εἰ μὴ φιλῶ.
Ξα.
οἶδ᾽ οἶδα τὸν νοῦν· παῦε παῦε τοῦ λόγου.
οὐκ ἂν γενοίμην Ἡϱακλῆς ἄν. (...)
(‘May I die a miserable death if I don’t love Xantias! # I know
what you’re thinking, I know. Stop talking, stop it. I’m not
going to be Heracles again.’ Ar. Ra. 579–81)
In this example, Xanthias uses a potential optative to strengthen
his refusal: ‘I would under no circumstances (be willing to)
become Heracles.’ Note that in this special case, the verb
γίγνομαι (‘become’, ‘happen’) denotes a controllable action, in
contrast to its normal use. Xanthias may decide himself to
‘become’ Heracles (viz. to put on the Heracles costume), or,
consequently, to refuse to do so. As the syntax of οὐκ ἂν
γενοίμην Ἡϱακλῆς ἄν is anything but complex, the repetition of
ἄν has to be explained in pragmatic terms. It is the negation that
makes this context suitable for such repetition.
In short, a correlation between negation and the repetition of
ἄν can be observed in the corpus of potential optatives, and this
__________
55
Goldstein (2010: 176, 193) also discusses repeated ἄν, but only in
Herodotus, where the repetition can be explained by the combination of
modal and clause-level semantics of participial phrases. In a different context
(p. 59), Goldstein notes that when οὐ(κ) is stressed for pragmatic or semantic
reasons, ‘its consequent prosodic body makes it a licit clitic host’ for clitics
such as ἄν. Thus, the presence of ἄν after οὐκ can be an indication of pragmatic or semantic emphasis on the negation.
ipabo_66.249.78.36
102
Annemieke Drummen
correlation can be explained with Slings’ (1992) observation,
already hinted at by Paley (1881), that emphatic words lead to
‘pragmatic complexity’.
5. Summary and conclusions
Admittedly, the same network of different semantic and
pragmatic meanings can be described using a monosemous or
polysemous approach, instead of construction grammar. The
monosemous view may be represented as in Figure 2; the
polysemous one as in Figure 3.
Figure 2. A monosemous description of the potential optative in
tragedy and comedy (based on Boogaart 2009: 218)
103
Figure 3. A polysemous description of the potential optative in tragedy
and comedy (based on Boogaart 2009: 221, 224)
As described by Boogaart (2009), a monosemous approach
clarifies what the different meanings or uses of a form have in
common, while a polysemous description shows how they
differ, and how they are related. However, a constructiongrammar approach deals with both problems at the same time,
and makes it clear, on top of that, which contextual features
exactly lead to which interpretation. In such constructionist
approach, forms and meanings are connected in conventionalized associations, called constructions, at any level of
linguistic description. These constructions are again connected
in networks. Figure 4 gives a complete overview of my constructionist description of the potential optative. The daughter
constructions are presented in order of their frequency in the
corpus. Features of form given in parentheses are not necessary
for individual instances to be identified as a specific construction, but do show compatibility with it.
ipabo_66.249.78.36
104
Annemieke Drummen
Figure 4. A constructionist description of the potential optative in
tragedy and comedy
We can distinguish one ‘parent’ for several more specific
daughter constructions, which inherit all features of the parent,
but also each have their own features of form and meaning. The
‘form’ of this parent construction is the combination of a verb in
the optative mood and the particle ἄν; its ‘meaning’ can be
described as epistemic possibility. This meaning may be para-
105
phrased by: ‘it is possible (according to the speaker) that the
state of affairs obtains’. It implies that some condition is operative for the occurrence of the state of affairs. All 230 potential
optatives in the corpus belong to this parent construction.
Of these, 109 cases do not show additional features of
meaning. However, we can still distinguish them as a separate,
more specific construction, because particular features of form
indicate that the other daughter constructions are canceled.
Instead, these cases are characterised by elements which
emphasize the wide-scope meaning of the potential optative.
The next daughter construction shown in Figure 4 has
‘participant-oriented possibility’ at its ‘meaning’ pole. This may
be paraphrased by: ‘the subject has the possibility to carry out
the state of affairs’. Individual instances may involve
participant-internal possibility, participant-external possibility,
or a mixture of both, depending on the relevant enabling or
disabling factors. English translators of the plays stress the
proposition-internal relation of participant-oriented possibility
by choosing more often ‘could’ or ‘can’ in these cases, while
‘would’ or ‘will’ is the most frequent translation for cases
without this implication. Other contextual features that may
make this implication prominent are emphasis on the subject
(for example with a personal pronoun), an interrogative sentence
type, negation, the wider context, and the lexical meaning of the
state of affairs. This meaning of participant-oriented possibility
is sometimes further exploited, as an even more specific
‘granddaughter’ construction, to convey an indirect wish.
The daughter construction mitigation was identified in 43
cases. Presenting a state of affairs as possible (epistemic possibility) may be interactionally used to be more polite to an
addressee, or to cover oneself up. The specific effect of the
mitigation is dependent on the illocution of the utterance. The
contextual features that are important for this use are the status
relationship between the speaker and the addressee, some
specific states of affairs (for example, volitive verbs), and the
first or second person. Furthermore, negation blocks the mitiga-
ipabo_66.249.78.36
106
Annemieke Drummen
tion reading, because of its emphatic nature. When the first and
second person are constrasted with each other, an extra specific
‘granddaughter’ construction may be created, conveying a concessive nuance.
24 cases belong to the ‘strengthening’ daughter construction.
Necessary contextual ingredients are a negation, the
involvement of the speaker, and a controllable state of affairs.
By negating the possibility of himself carrying out some state of
affairs, the speaker implies that he would under no circumstances be willing to do so. Potential optatives exploited to
strengthen a refusal may be accompanied by a emphatic elements,
and are especially compatible with contexts of anger.
Incidentally, the emphatic nature of negation was confirmed
again by its correlation with double ἄν, which turned out to be
present in the corpus. Negated potential optatives are significantly more often accompanied by two instances of the modal
particle than non-negated cases.
Remarks on all these nuances are already found in the secondary
literature on the potential optative (§ 1). This constructionist
description adds more, however. It explains at the same time
what the different instances have in common, how they differ,
and how they are related. Moreover, by making it explicit which
features of form are relevant for each meaning, it shows why a
certain meaning makes most sense in an individual case. In this
way, construction grammar may decrease the subjectivity of our
interpretation.56
__________
56
I would like to thank Rutger J. Allan, Anna Bonifazi, and Elizabeth
Koier for helpful comments on earlier versions of this paper.
107
Bibliography
Allan, R. J. (2010): Review of J. Willmott, 2007, The Moods of Homeric
Greek. Mnemosyne 63: 298–303.
Auwera, J. van der & Plungian, V. A. (1998): Modality’s semantic map.
Linguistic Typology 2: 79–124.
Boogaart, R. (2009): Semantics and pragmatics in construction grammar: the
case of modal verbs. In: Bergs, A. & Diewald, G. (eds.). Contexts and
Constructions. Amsterdam-Philiadelphia: 213–41.
Bybee, J. L., Perkins, R. D. & Pagliuca, W. (1994): The evolution of grammar. Tense, aspect and modality in the languages of the world. ChicagoLondon.
Coates, J. (1983): The semantics of the modal auxiliaries. London-Canberra.
Croft, W. (2001): Radical Construction Grammar. Syntactic theory in typological perspective. Oxford.
Croft, W. & Cruse, D. A. (2004): Cognitive Linguistics. Cambridge.
Fillmore, C., Kay, P. & O’Connor, M. C. (1988): Regularity and idiomaticity
in grammatical constructions: the case of let alone. Language 64: 501–38.
Goldberg, A. (1995): Constructions. A construction grammar approach to
argument structure. Chicago.
Goldstein, D. M. (2010): Wackernagel’s Law in Fifth-Century Greek. PhD
dissertation, University of California, Berkeley.
Griffith, M. (1999): Sophocles: Antigone. Cambridge.
Heine, B. (1995): Agent-Oriented vs. Epistemic Modality: Some Observations on German Modals. In: Bybee, J. L. & Fleischman, S. (eds).
Modality in grammar and discourse. Amsterdam-Philadelphia: 17–53.
Henderson, J. (1998a): Aristophanes: Acharnians, Knights. LondonCambridge.
– (1998b): Aristophanes: Clouds, Wasps, Peace. London-Cambridge.
– (2002): Aristophanes: Frogs, Assemblywomen, Wealth. LondonCambridge.
Hengeveld, K. (2004): Mood and modality. In: Booij, G., Lehmann, C. &
Mugdan, J. (eds.). Morphology: A handbook on inflection and wordformation. Berlin-New York, 1190–1202.
Kovacs, D. (2005 [1995]): Euripides: Children of Heracles, Hippolytus,
Andromache, Hecuba. London-Cambridge.
Kühner, R. & Gerth, B. (1963 [1898–1904]): Ausführliche Grammatik der
griechischen Sprache. Zweiter Teil: Satzlehre (erster Band). Darmstadt.
Lloyd-Jones, H. (1994a): Sophocles: Ajax, Electra, Oedipus Tyrannus.
London-Cambridge.
– (1994b): Sophocles: Antigone, The women of Trachis, Philoctetes,
Oedipus at Colonus. London-Cambridge.
ipabo_66.249.78.36
108
Annemieke Drummen
Moorhouse, A. C. (1982): The Syntax of Sophocles. Leiden.
Nuyts, J. (2001). Epistemic modality, language, and conceptualization.
Amsterdam-Philadelphia.
Östman, J.-O. & Fried, M. (2004): Historical and intellectual background of
Construction Grammar. In: Fried, M. & Östman, J.-O. (eds.). Construction Grammar in a cross-language perspective. Amsterdam-Philadelphia:
1–10.
Paley, F. A. (1881): A short treatise on the Greek particles and their combinations. Cambridge-London.
Rice, J. E. (19952): Mathematical Statistics and Data Analysis. Belmont.
Rijksbaron, A. (20023): The syntax and semantics of the verb in Classical
Greek. Amsterdam.
Schwyzer, E. & Debrunner, A. (1950): Griechische Grammatik: auf der
Grundlage von Karl Brugmanns Griechischer Grammatik. Munich.
Slings, S. R. (1992): Written and Spoken Language: an Exercise in the Pragmatics of the Greek Language. Classical Philology 87, 95–109.
Slotty, F. (1915): Der Gebrauch des Konjunktivs und Optativs in den
griechischen Dialekten. Göttingen.
Smyth, H. W. (1957 [1926]): Aeschylus: Agamemnon, Libation-bearers,
Eumenides, Fragments. London-Cambridge.
– (1963 [1922]): Aeschylus: Suppliant maidens, Persians, Prometheus,
Seven against Thebes. London-Cambridge.
Willmott, J. (2007): The moods of Homeric Greek. Cambridge.
– (2008): The “potential” optative in Homeric Greek. The Cambridge
classical journal: Proceedings of the Cambridge Philological Society 54,
237–51.
Young, D. (1974): Aeschylus: The Oresteia. Norman.
Un costrutto discusso: participio pro verbo
di modo finito?
Di TRISTANO GARGIULO, Cagliari
Abstract: This paper re-examines a very peculiar and discussed use of
the Greek participle, which has been recognized by some scholars and
defined as ‘participle used as a finite verb’ or, better, as ‘participle
used instead of a finite verb’. However, this viewpoint has not been
generally accepted. The pattern in question occurs mainly, although
not exclusively, in late Greek prose, when a participle is found where
we expect, and the syntax would require, a finite verb, both in independent and subordinate clauses. In most cases editors alternatively
either choose a more regular reading, if witnessed, or simply emend
the text. Therefore we can find not a scanty amount of examples of
this construction relegated to (exhaustive) critical apparatuses, as
discarded variants, where we see that a participle is actually the only
reading or a better attested variant than the correspondent finite verb.
In order to strongly argue in favor of the real existence and legitimacy,
sometimes disputed, of this use of the participle, the author of the
present article intends, firstly, to offer a significant number of new
examples gathered from various works of Greek prose. In particular
the recensio vetusta of the Alexander Romance, represented by the codex unicus A, will be thoroughly scrutinized, where such examples,
amounting to the high concentration of nearly twenty ones, have almost
all been corrected and removed from the text by previous editors.
Secondly, he aims to show how we can manage to assemble and
classify all the known examples in a few broad categories repeatedly
recurring with a certain consistency. The author reviews the explanations (ellipsis, anacoluthon) proposed so far, adding some considerations on the prominent role of the participle in the Greek sentence.
Si riconosce per lo più a J. H. Moulton il merito di aver fatto
cadere l’attenzione su una serie di luoghi dove sembra affiorare
un rarissimo e del tutto particolare impiego del participio. Egli
diede al paragrafo ad esso dedicato il titolo di “Participle for
Glotta 89, 109–125, ISSN 0017–1298
ipabo_66.249.78.36
110
Tristano Gargiulo Indicative” e analizzò alcuni esempi neotestamentari, che venivano apparentemente confermati da papiri di epoca sia tolemaica che romana: da questo complesso di testimonianze, a suo
avviso, si poteva dedurre che in taluni casi “the participle can be
used for indicative or imperative”.1
Mostrandosi d’accordo sull’esistenza dell’inusitato costrutto,
L. Radermacher2 ne tentò una spiegazione collegandolo alle
espressioni verbali perifrastiche costituite da participio più
ausiliare eijmiv, in cui capita talvolta, già nel greco classico, che
quest’ultimo sia sottinteso. Tra i nuovi esempi che portava, ve
n’era uno particolarmente interessante perché proveniva da un
autore ascrivibile alla letteratura classica (benché di nascita
ebraica), il che estendeva il campo dell’indagine, non più
dunque limitato solo ai testi cristiani e alla lingua spesso popolare dei papiri: Ios. Fl. Iud. Ant. IV 181 mhvt’ eujsebeiva", h|" nu'n
peri; to;n qeo;n e[conte", katafronhvsante" eij" a[llon
metasthvshsqe trovpon.3 Radermacher aveva dovuto rintracciare il participio e[conte", pur unanimemente tràdito, negli
apparati critici, perché esso era stato rimosso dal testo in tutte le
edizioni e sostituito dall’indicativo (congetturale) e[cete.
Contro Moulton e Radermacher si levò la voce di E. Mayser,4
il quale negò con decisione la presenza del fenomeno nei papiri
tolemaici, che erano stati chiamati in causa, senza tuttavia
escluderlo nei testi neotestamentari o della tarda grecità.
Nel 1929 uscì un articolo di H. Frisk5 specificamente dedicato al problema. In esso l’autore per prima cosa rettificava il
nome da dare al costrutto: non “das Partizipium a l s Verbum
__________
1
J. H. Moulton, A Grammar of New Testament Greek, Edinburgh 19062,
pp. 180 ss., 222–5.
2
L. Radermacher, Neutestamentliche Grammatik, Tübingen 19252, p. 205.
3
Più incerto il caso di XVII 129, dove il participio eijwqovte", posto
all’inizio di un periodo lungo e complesso, che esso sembra governare, a mio
avviso assomiglia piuttosto ad un nominativus pendens. E forse è per questo
che gli editori non lo correggono.
4
E. Mayser, Grammatik der griechischen Papyri aus der Ptolemäerzeit
II. Satzlehre, Berlin-Leipzig 1926, pp. 340–6.
5
H. Frisk, Partizipium und Verbum finitum im Spätgriechischen, Glotta
17, 1929, pp. 56–66.
Un costrutto discusso: participio pro verbo di modo finito?
111
finitum”, ma “das Partizipium s t a t t eines Verbum finitum”.
Egli discuteva anche una diversa configurazione, non ancora
segnalata, in cui il participio non funge da unico verbo principale della frase ma è coordinato con kaiv ad un indicativo, e ne
adduceva un nitido esempio di letteratura popolare in un passo
dei Miracula di Cosma e Damiano (48, 13–5 Deubner): h[kousen
de; kai; ta; qauvmata tw'n ejndovxwn ajnarguvrwn Kosma' kai;
Damianou', kai; ejlqw;n ejn tw'/ oi[kw/ aujtw'n kai; levgei pro;" tou;"
aJgivou". Frisk notava, inoltre, come il fenomeno sembrasse
circoscritto, nel Nuovo Testamento, a Paolo e all’Apocalisse6 e,
soprattutto, negava le attestazioni che di esso Radermacher
aveva indicato in Giuseppe Flavio: in particolare, nell’esempio
che abbiamo sopra citato, riteneva inevitabile la correzione di
e[conte" in e[cete. Frisk, in definitiva, giudicava discutibile «der
Gebrauch des Partizipiums statt eines Verbum finitum» e
riteneva che si dovesse ricorrere di volta in volta a diverse
spiegazioni, per lo più extralinguistiche, per tale apparente
costrutto: tentativo mal riuscito di innalzare il livello stilistico,
‘deragliamento’ sintattico dovuto a spinte psicologiche (sul tipo
dell’anacoluto), corruttela della tradizione manoscritta.
Col tempo, una volta venuta in luce la possibile esistenza di
questo particolarissimo uso del participio, ne sono stati prodotti
ulteriori esempi di varia provenienza. U. Ursing7 ne ha selezionati dai testi favolistici, T. Wikström8 da Vettio Valente (Wikström è stato anche il primo, a quanto mi consta, a segnalare
l’uso di questo participio anche come verbo di proposizioni
__________
6
G. Mussies, The Morphology of Koine Greek as used in the Apocalypse
of St. John. A Study in Bilingualism, Leiden 1971, pp. 324–8, ha ricondotto il
sintagma ad un influsso dell’ebraico, ammesso anche da M. Wilcox,
Semitisms in the New Testament, in: ANRW II 25.2, Berlin-New York 1984,
p. 1016.
7
U. Ursing, Studien zur griechischen Fabel, Lund 1930, pp. 68–9.
8
T. Wikström, Ad Vettium Valentem syntactica et critica, Eranos 47,
1949, pp. 35–8.
ipabo_66.249.78.36
112
Tristano Gargiulo dipendenti, in luogo di un congiuntivo o un ottativo), B. Mandilaras9 dai papiri non letterari (soprattutto di epoca romana).
Non è stato, però, tuttora raggiunto un unanime consenso, e
anzi sussistono forti incertezze, sulla legittimità e sul preciso
statuto sintattico di un tale impiego del participio.10 Si oscilla tra
il vedervi una struttura ellittica (Radermacher; Ursing) oppure
una assoluta di tipo anacolutico (Mayser; Blass-Debrunner;11
Frisk). Esiste tuttavia, a mio avviso, anche un’altra via che si
può tentare di percorrere, alla ricerca di raffronti che possano
aiutarci a spiegare o anche solo ad inquadrare meglio quest’uso
idiomatico.
È ampiamente attestata, fin dal greco omerico, e debitamente
riconosciuta dagli studiosi, una configurazione sintattica in cui
“geht die Partizipialkonstruktion in eine Form des Verbi finiti
über”.12 Ciò avviene per lo più in forme di coordinazione
__________
9
B. Mandilaras, The Verb in the Greek Non-Literary Papyri, Athens
1973, pp. 372–3.
10
Vi accennano appena A. N. Jannaris, An Historical Greek Grammar,
London 1897, § 2168b, e R. Kühner - B. Gerth, Ausführliche Grammatik der
griechischen Sprache II 2, Hannover-Leipzig 19043, p. 109 Anm. 3. Più
esplicitamente contrario E. Schwyzer, Griechische Grammatik II, München
1959, p. 407, che lo giudica frutto di una ipotesi “nicht genügend begründet”.
Almeno per l’epoca dal V sec. d. C. in poi, tuttavia, R. Browning, Medieval
and Modern Greek, Cambridge 19832, p. 64, ne parla come di un fenomeno
ben stabilito: “We also find participles used in coordination with finite verbs,
e. g. dexavmeno" ou\n oJ basileu;" ta; gravmmata para; ∆Epifanivou kai;
ejpoivhsen tuvpon toiou'ton”. Pur riconoscendo la giustezza di alcune delle
obiezioni di Frisk, anche G. Björk, HN DIDASKWN. Die periphrastischen
Konstruktionen im Griechischen, Uppsala 1940, pp. 116–7, concludeva che
l’esistenza di quest’uso del participio andava accettata “als Symptome des
schwingenden Gefühls für den idiomatischen Gebrauch des Partizips”. Nulla
si trova al riguardo in: W. J. Aerts, Periphrastica, Amsterdam 1965.
11
F. Blass - A. Debrunner - F. Rehkopf, Grammatik des neutestamentlichen Griechisch, Göttingen 197614, § 468 (= Grammatica del greco del
Nuovo Testamento, ed. it., Brescia 1982, pp. 570–1).
12
Kühner-Gerth, Ausführliche Grammatik, cit., p. 100, che ne offrono
esauriente documentazione (alla quale ho attinto). Ved. anche J. D. Denniston, The Greek Particles, Oxford 19542, p. 369 nt. 1 (“transition from
participial to finite construction”), e H. W. Smyth, Greek Grammar, Cambridge 19562, pp. 477 § 2147c e 656 § 2904a. Il primo a notare il fenomeno,
nel testo omerico, sembra sia stato H. Bekker, in: Monatsberichte der Akademie der Wissenschaften zu Berlin 1864, p. 141, seguito da O. Schneider,
Coniectanea in Colluthum, Philologus 23, 1866, pp. 441–2, che aggiunse altri
113
apparentemente simmetriche (me;n ... de; ... / te ... kai; ... / te ...
te ... / h] ... h] ... / ei[te ... ei[te ...), ma in realtà sensibil-mente
sbilanciate verso il secondo membro, costituito dalla frase con
verbo di modo finito, su cui cade l’enfasi. Questo fa sì che si
crei una sorta di equivalenza tra i due membri, tale da far sentire
quasi come autonomo il costrutto participiale. Non potremmo
trovarci qui di fronte a un antecedente del fenomeno che stiamo
indagando, che ne sia stato all’origine o ne abbia almeno
favorito lo sviluppo? Come esempi rappresentativi di questa
tipologia (ed anche della sua ampia diffusione) si possono citare:
– per la coordinazione me;n ... de; ... (la più diffusa): Hom. Il. V
145–7 to;n me;n ... balw;n ... / to;n d∆ e{teron ... / plh'x∆,13 592–4
“Arh" kai; povtni∆ ∆Enuwv, / hJ me;n e[cousa Kudoimo;n ... / “Arh" d∆
... e[gco" ejnwvma, XVIII 172–5 oiJ d∆ ajllhvlou" ojlevkousin / oiJ
me;n ajmunovmenoi nevkuo" pevri teqnew'to", / oiJ de; ejruvssasqai
poti; “Ilion hjnemovessan / Trw'e" ejpiquvousi, Od. XI 82–3 ejgw;
me;n ... favsganon i[scwn, / ei[dwlon d∆ ... povll∆ ajgovreuen, Aesch.
Sept. 758 to; me;n pivtnon, a[llo d∆ ajeivrei, Eur. Ba. 1132–3 oJ me;n
stenavzwn... / aiJ d∆ hjlavlazon; e, in prosa, Hdt. V 37, 214 tou;" me;n
ejxelauvnwn tw'n turavnnwn, tou;" d∆ e[labe turavnnou", VI 25, 2
kai; Karivhn e[scon oiJ Pevrsai, ta;" me;n ejqelonth;n tw'n polivwn
uJpokuyavsa", ta;" de; ajnavgkh/ proshgavgonto, Thuc. III 87, 1 hJ
novso" to; deuvteron ejpevpese toi'" ∆Aqhnaivoi", ejklipou'sa me;n
oujdevna crovnon to; pantavpasin, ejgevneto dev ti" o{mw" diokwchv,
Xen. Mem. II 1, 30 i{na me;n hJdevw" favgh/", ojyopoiou;"
mhcanwmevnh, i{na de; hJdevw" pivh/", oi[nou" te polutelei'"
paraskeuavzh/ kai; tou' qevrou" ciovna periqevousa zhtei'", Dem.
__________
esempi e se ne avvalse per difendere il testo tràdito in un luogo di Colluto
(vv. 257–62).
13
Già notato da scholia vetera (ad V 147b Erbse: ajpo; metoch'" ... eij"
rJh'ma) ed Eustazio (532, 27–8 van der Valk).
14
Sia gli esempi erodotei che quelli tucididei di seguito citati si trovano
già segnalati e raggruppati in due note dei commenti, rispettivamente, di H.
Stein, Herodotos, I, Berlin 19016 (rist. 1962), p. 104, e di J. Classen - J.
Steup, Thukydides, I, Berlin 19195, p. 169, da cui li ho desunti.
ipabo_66.249.78.36
114
Tristano Gargiulo 57, 11 ou|to" ejblasfhvmei kat∆ ejmou' ... mavrtura me;n w|n
kathgovrhsen oujdevna parascovmeno" ..., parekeleuveto de;
toi'" dhmovtai" ajpoyhfivzesqai. Anche senza mevn: Hom. Od.
XVII 65–6 ajmfi; dev min mnhsth're" ajghvnore" hjgerevqonto /
e[sql∆ ajgoreuvonte", kaka; de; fresi; bussodovmeuon;
– per la coordinazione te ... kai; ...: Thuc. IV 100, 1 prosevbalon tw/' teicivsmati, a[llw/ te trovpw/ peiravsante" kai;
mhcanh;n proshvgagon, VII 47, 2 novsw/ te ga;r ejpievzonto
kat∆ ajmfovtera, th'" te w{ra" ... ou[sh" ejn h|/ ajsqenou'sin
a[nqrwpoi mavlista, kai; to; cwrivon ... eJlw'de" kai; calepo;n
h\n, Xen. Cyr. V 4, 29 parh'n oJ Gadavta" a[lla te dw'ra
polla; kai; pantoi'a fevrwn kai; a[gwn ... kai; i{ppou" de;
h\ge pollouv". In Erodoto più volte anche te ... kai; dh; kai; ...:
I 85, 1 oJ Kroi'so" to; pa'n ej" aujto;n ejpepoihvkee a[lla te
ejpifrazovmeno" kai; dh; kai; ej" Delfou;" peri; aujtou'
ejpepovmfee crhsomevnou", III 152 a[lloisiv te sofivsmasi
peirhsavmeno" kai; dh; kai; tw/' Ku'ro" ei|lev sfea", kai;
touvtw/ ejpeirhvqh;
– per la coordinazione te ... te ...: Hom. Il. III 79–80 tw'/ d∆
ejpetoxavzonto ... ∆Acaioi; / ijoi'sivn te tituskovmenoi lavessiv
t∆ e[ballon,15 Hdt. VIII 136, 2 lewvn te pollo;n a[ra ajkouvwn
ei\nai kai; a[lkimon, tav te kata; th;n qavlassan
suntucovnta sfi paqhvmata katergasamevnou" mavlista
∆Aqhnaivou" ejpivstato;
– per la coordinazione h] ... h] ... o ei[te ... ei[te ...: Hom. Od.
IX 339 h[ ti oji>savmeno", h] kai; qeo;" w}" ejkevleusen, Hdt. I
19, 2 pevmpei ej" Delfou;" ... ei[te dh; sumbouleuvsantov"
teu ei[te kai; aujtw/' e[doxe pevmyanta to;n qeo;n ejpeirevsqai
peri; th'" nouvsou. Anche con ouj mei'on h[: Arr. Anab. II 11, 3
ouj mei'on ajp∆ ajllhvlwn katapatouvmenoi h] pro;" tw'n
diwkovntwn polemivwn ejblavptonto;
__________
15
Lo scolio propone in alternativa una soluzione (schol. vet. ad III 80a
Erbse: ajpo; metoch'" eij" rJh'ma, h] ajpo; koinou' to; e[ballon) seguita da
qualche interprete (così, p. es., M. M. Willcock, The Iliad of Homer, London
1978, p. 217: “Take both ijoi'sin and lavessi with e[ballon: ‘and they were
aiming at him and trying to hit him with arrows and stones’”).
115
– in asindeto: Hom. Il. XVIII 535–7 ojloh; Khvr, / a[llon zwo;n
e[cousa neouvtaton, a[llon a[outon, / a[llon teqnew'ta ...
e{lke podoi'i>n;
– con altra avversativa: Hom. Od. XXIII 350–3 polevwn
kekorhvmeq∆ ajevqlwn / ajmfotevrw, su; me;n ejnqavd∆ ejmo;n
polukhdeva novston / klaivous∆, aujta;r ejme; Zeu;" ... kai;
qeoi; a[lloi / iJevmenon pedavaskon.
Una diversa forma di ‘asimmetria’, che sembra ugualmente
testimoniare una predisposizione della lingua greca a un uso
sciolto e autosufficiente del participio, si può altresì individuare
in certe strutture dialogiche (che qualificherei più come idiomatiche che come propriamente ellittiche), frequenti in Platone, in
cui un personaggio ripete un’affermazione già fatta, per
ribadirla, o un suo interlocutore per approvarla, e ciò avviene, ad
inizio di battuta, tramite una frase participiale sintatticamente
autonoma e compiuta: p. es. Crat. 399 a (SW.) th'/ tou'
Eujquvfrono" ejpinoiva/ pisteuvei", wJ" e[oika". (ERM.) dh'la
dhv. (SW.) ojrqw'" ge su; pisteuvwn, Soph. 231 b–c (QEAI.) ...
ajporw' de; e[gwge ... (XE.) eijkovtw" ge su; ajporw'n, Symp.
174 e ei\pon ou\n o{ti kai; aujto;" meta; Swkravtou" h{koimi,
klhqei;" uJp∆ ejkeivnou deu'r∆ ejpi; dei'pnon. kalw'" g∆, e[fh,
poiw'n suv.
Gli esempi veri e propri del particolare fenomeno che stiamo
analizzando sono ancora relativamente radi e di disparata
provenienza. Sembrano quasi affiorare isolati nei testi in cui
vengono ritrovati.16 C’è, però, qualche ragione di credere che
__________
16
Vi sono tuttavia indizi che, in alcuni casi, gli studiosi che si sono
occupati di questo tema si siano limitati ad una campionatura molto ridotta.
Si può osservare, infatti, che nella stessa pagina (p. 207 dell’edizione di L. Deubner, Kosmas und Damian. Texte und Einleitung, Leipzig-Berlin 1907), da cui
è stato tratto l’esempio di Cosma e Damiano riportato supra a p. 111, si trovano
altri due casi che si potevano ad esso affiancare a pieno titolo: Mir. 48, 24–5
kai; hJmei'" deovmenoi tou' qeou', kai; hJmw'n deivxei oJ kuvrio", 26–7 kai; th;n
e{kthn w{ran th'" nukto;" ejlqw;n oJ a[ggelo" ÔRafah;l kai; levgwn pro;"
aujtouv". Analogamente, ad un esempio di Malalas (220, 22 Dindorf ajpo; th'"
ipabo_66.249.78.36
116
Tristano Gargiulo proprio il sospetto circa la reale esistenza di tale costrutto abbia
spinto gli editori ad eliminarne molte occorrenze correggendo i
testi, e che quindi un lavoro di recupero effettuato sugli apparati
critici possa dare i suoi frutti.
A riprova di ciò, un rapido scandaglio, condotto sul Bellum
Iudaicum di Giuseppe Flavio, mi ha portato a individuare tre
esempi piuttosto interessanti, perché ricoprono esattamente le
diverse tipologie che sono quelle più rappresentative di tale
struttura sintattica. In un caso abbiamo il participio in una
subordinata (causale): II 590 ejpei; de; to;n jIwvshpon oJrw'n
(PALVRC oJrwv/h M) aujtou' sfovdra caivronta tw'/ drasthrivw/,
peivqei prw'ton me;n aujtw'/ pisteu'sai to; tei'co" ajnoikodomh'sai th'" patrivdo". In un altro, abbiamo, in proposizione
indipendente, un participio futuro coordinato con un indicativo
futuro: III 483 prokinduneuvsw gavr ... kai; prw'to" eij" tou;"
polemivou" ejmbalw'n (PAMLacVRC ejmbalw' Lpc). Nel terzo
passo troviamo un participio presente che funge da verbo
principale e corrisponde ad un imperfetto: IV 302 nu'n de;
prwvtou" tou;" zhlwta;" speuvdonte" (PAMLVR e[speudon C)
th'" froura'" ejxelevsqai. In tutti e tre i casi il participio è
lezione quasi concordemente attestata.17 Dopo B. Niese e J. v.
Destinon (che difendevano il participio in II 590 e IV 302, ma
non in III 483),18 gli editori19 hanno sistematicamente confinato
i participi in apparato.
__________
kwvmh" ∆Hpeivrou ejxormhvsa" kai; h\lqen), che, citato isolatamente da Björk
(op. cit. alla nt. 10), sembrava quasi un unicum in questo autore, si possono
aggiungere quelli elencati da A.-J. Festugière, Notabilia dans Malalas II,
RPh 53, 1979, pp. 232–3, la maggior parte dei quali è accolta dagli editori
(sia Dindorf sia il recente J. Thurn, Berlin-New York 2000).
17
Al punto che i più regolari oJrwv/h, ejmbalw' e e[speudon si direbbero quasi
frutto di congettura normalizzante dell’unico codice (sempre diverso nei tre
casi) che li tramanda. Così sembra pensare anche Niese quando, a proposito
di M che ha oJrwv/h, annota: «qui sententiam uidit non recte procedere». Il
sospetto è ancora più forte per ejmbalw', che è lezione post correctionem.
18
Flavii Josephi Opera VI, ed. I. a Destinon et B. Niese, Berolini 1895
(19552).
19
H. St. J. Thackeray (Cambridge Mass.-Harvard 1971), G. Vitucci
(Milano 1974), A. Pelletier (Paris 1975–1982).
117
Altre volte l’esitazione dell’editore nelle scelte testuali appare
significativa. Nella Vita Aesopi G (codex unicus), pubblicata da
B.E. Perry,20 troviamo due casi in cui lo stesso Perry (seguito
dai successivi editori) elimina per correzione costrutti participiali del tipo che stiamo indagando: 2 o{tan ga;r oJ despovth"
ejpizhthvsh/ ta; su'ka kai; mh; e[conte" (G: e[cwmen Perry)
dou'nai, tiv e[stai, 28 o{tan de; sunestw;" oujrw' to; e[dafo"
zevon tou;" povda" mou katakaivei, kai; hJ drimuvth" tou'
ou[rou ajnatrevcei mou eij" ta;" ojsfrhvsei" ªkai; ta;" ejkroa;"
mou fravsseiº, kai; oJ h{lio" de; th;n kefalhvn mou flevgwn
(G: flevgei Perry). Ce ne sono però altri due in cui Perry o non
corregge: 33 hjrwvthsen to;n Diva katallagh'nai aujtw'/ kai;
mh; ajkurw'sai aujtou' th;n mantikhvn. oJ de; katallagei;" (G,
conservato da Perry, ma emendato in kathllavgh da Papathomopoulos) aujtw/', kai; ou{tw" ojneivrou" eJtevrou" e[plasen eij"
tou;" ajnqrwvpou" o{pw" aujtoi'" kaq∆ u{pnou" yeudh' deiknuvwsin, o, in frasi contigue, opta, pur dubbiosamente, per due
soluzioni diverse: 100 oJ de; Ai[swpo" quvsa" tai'" Mouvsai"
iJero;n kataskeuavsa" (G: kateskeuvasen Perry, che però in
apparato scrive, della lezione tràdita: fort. retinendum) aujtai'",
sthvsa" mevson aujtw'n Mnhmosuvnhn, oujk ∆Apovllwna. oJ
∆Apovllwn ojrgisqei;" (G, conservato da Perry, ma emendato in
wjrgivsqh da Papathomopoulos) aujtw/' wJ" tw/' Marsuva/.
La possibilità che di tale costrutto si possano rinvenire
esempi anche in epoca classica non è del tutto esclusa e
meriterebbe forse di essere esplorata. In poesia non sarebbe così
sorprendente (ved. infra, nt. 39). Per limitarmi ad un solo caso
prosastico in cui tale eventualità sarebbe da prendere in attenta
considerazione, adduco un luogo di Demade. Il testo comunemente accettato in Demad. Dodec. 6 eja;n tauvth" katatuvcw,
pavsa" ajpoluvsomai ta;" diabolav" è quello dei codici C (Laur.
__________
20
Aesopica I, Urbana 1952. Nessuno dei luoghi considerati ha un preciso
equivalente, con cui essere utilmente messo a confronto, nell’altra redazione
della Vita Aesopi denominata W.
ipabo_66.249.78.36
118
Tristano Gargiulo LVII 4) e B (Ambr. H 52 sup.), che hanno katatuvcw, laddove i
codici della famiglia di X (Pal. 88) danno invece katatucwvn.
Senza voler assegnare crismi di autenticità a quest’ultima lezione, si può almeno osservare come sia oggi accettato che X è
ramo poziore nella tradizione di vari oratori (tra cui Demade e
Lisia) e che le lezioni di C e B si devono considerare, in non
pochi casi, come felici lezioni congetturali21 (anche katatuvcw
potrebbe dunque, in teoria, esserlo).
Ma la testimonianza che qui vorrei soprattutto addurre è
quella di un testo letterario, la cui composizione si colloca in
epoca romana ma ha radici ellenistiche, e che presenta lingua e
stile compositi (accanto a preziosismi come ejpevfra, I 14, 8,
troviamo forme postclassiche quali gastevran, I 18, 10) ma di
registro non basso (occasionalmente, l’anonimo autore impiega
anche termini poetici nella sua prosa). Mi riferisco all’opera un
tempo conosciuta come Pseudo-Callistene e oggi meglio nota
come Romanzo di Alessandro, e in particolare alla cosiddetta
‘recensione’ A,22 tramandataci da un codex unicus, il Par. Gr.
1711.
L’immensa fortuna di cui ha goduto quest’opera naturalmente
non ne garantisce il valore letterario, ma vi sono in essa elementi
che permettono di qualificarla come sorvegliata e, nelle inten__________
21
Prima si riteneva che esse si dovessero al copista di C, cfr. L. Gernet in
Lysias. Discours I (I–XV), texte établi et traduit par L. Gernet et M. Bizos,
Paris 19553, p. 17: «il est certain que C est transcrit de X; mais le copiste, ici,
était intelligent et informé: telles de ses corrections sont heureuses; il mérite
de figurer parfois comme l’auctor d’une conjecture, ni plus ni moins qu’un
éditeur moderne». F. Donadi, Esplorazioni alla tradizione manoscritta
dell’Encomio di Elena gorgiano II: I mss.Laur. LVII.4 (C) e Ambr. H 52 sup.
(Am4), BIFG 3, 1976, pp. 225–50, ha mostrato come C dipenda in realtà da B,
e che si debbano attribuire al copista di quest’ultimo (che egli identifica in
Andronico Callisto) le caratteristiche e i meriti che prima si riconoscevano al
copista di C.
22
Per le vicende di quest’opera e le sue varie redazioni (‘recensioni’),
ved. da ultimo C. Franco, Vita di Alessandro il Macedone, Palermo 2001, pp.
15–29; C. Jouanno, Naissance et métamorphoses du Roman d’Alexandre.
Domaine grec, Paris 2002, pp. 5–55; Il Romanzo di Alessandro, vol. I, testo,
traduzione e commento a c. di R. Stoneman e T. Gargiulo, Milano 2007, pp.
LXXIII–LXXXIII.
119
zioni di chi l’ha ideata,23 di qualche ambizione.24 È un
prosimetrum, con sequenze anche lunghe di versi (per lo più
coliambi); oltre a vere e proprie citazioni, sono riconoscibili
riecheggiamenti, che paiono deliberati, della poesia classica,25 e
– cosa a mia conoscenza non ancora messa nel dovuto rilievo –
vi si può identificare una rete di significativi rimandi e richiami
a distanza26 e un fil rouge costituito da termini e concetti
chiave:27 tutto ciò denota un’accurata, e relativamente complessa, costruzione.
Ebbene, nel testo della pur meritoria edizione della recensione A curata da W. Kroll28 non vi è quasi traccia di un uso del
__________
23
Sia pure su materiali preesistenti, secondo la classica teoria di R.
Merkelbach, Die Quellen des griechischen Alexanderromans, München
19772 (1954).
24
Dell’anonimo autore di quest’opera sono stati dati spesso giudizi troppo
severi. Si pensi a quello di Merkelbach (op. cit., p. 89): “Der Verfasser des
Alexanderromans war zweifellos ein sehr ungebildeter und unwissender
Mensch”. Solo di poco più indulgente quello di A. Ausfeld, Der griechische
Alexanderroman, Leipzig 1907, p. 5, che pensa “dass der Alexanderroman im
ganzen und grossen nicht das Produkt der Volksüberlieferung, sondern einer
halb gelehrten Schriftstellerei sei”. Negli studi più recenti si tende ad una
almeno parziale rivalutazione (Franco, Jouanno, Stoneman).
25
Ved. p. es. G. Ieranò, Il barbaro in fuga: un’eco dei Persiani di Eschilo
nel Romanzo di Alessandro, Aevum Antiquum 9, 1996, pp. 217‒34. Non
sarebbe l’unico caso. Il racconto della vittoria di Alessandro giovinetto nella
gara dei carri ad Olimpia (I 19) ricorda da vicino, in molti dettagli, l’episodio
della gara delfica in cui si finge che abbia perso la vita Oreste nell’Elettra
sofoclea (El. 698 ss.).
26
P. es. I 1, 3 ta; ... kosmika; stoicei'a ~ I 13, 1 kosmikw'n stoiceivwn;
I 3, 5 kovsmon kukleuvsa" ~ I 11, 2–4; un importante rimando interno
(verbatim!), da un libro a un altro, è in II 4, 5 a I 30, 5.
27
In tutta l’opera, Alessandro è ripetutamente descritto come frenhvrh" (I
16, 5; 19, 5; 37, 5; II 4, 6; 13, 2; 16, 1; III 3, 3; 19, 8; 23, 5); ma soprattutto
egli ha il ruolo di e[kdiko" (‘vendicatore’): lo è, di volta in volta, della madre
Olimpiade per i torti subiti da parte di Filippo (I 4, 8; 5; 21, 4), di Filippo
stesso (I 24, 5), degli Ateniesi (II 3, 10), e perfino del suo nemico Dario (II
20, 3 e 8). Alessandro è consacrato kosmokravtwr – il termine che più lo
definisce – da Nectanebo (I 12, 8), da Aristotele (I 16, 5), da Filippo (I 17, 5;
24, 9), dagli Egiziani (I 34, 2).
28
W. Kroll, Historia Alexandri Magni (Pseudo-Callisthenes). Recensio
Vetusta, Berlin 1926.
ipabo_66.249.78.36
120
Tristano Gargiulo participio come quello che abbiamo prima delineato. Eppure, ad
un più attento esame, si scopre che di potenziali casi di
participio impiegato come verbo finito (o al posto di un verbo
finito, se preferiamo la precisazione di Frisk sopra ricordata) ve
ne sarebbero quasi una ventina, cioè la più alta concentrazione
che sia stato dato fino ad oggi di trovare in uno stesso testo. Solo
che sono stati pressoché tutti sistematicamente eliminati per
correzione29 e sono fortunatamente rintracciabili solo grazie
all’esaustività dell’apparato critico dell’editore (in altri apparati
critici, è bene ricordarlo, delle correzioni considerate palmari
spesso non rimane nessuna traccia).
Ecco l’elenco dei passi in questione:
(1) I 13, 2 ou{tw" eijpovnto" aujtou' to; paidivon labw;n (e[labe
Kroll) th;n devousan ejpimevleian
(2) I 32, 3 h[negkan aujto;n eij" to; nu'n kalouvmenon hJrw'/on kai;
ejpideivxante" (ejpevdeixan Kroll) th;n lavrnaka
(3) I 39, 4 ejgw; ga;r ejkduvsa" aujto;n th;n porfuvran kai; plhga;"
dou;" ajpostevllwn (ajpostelw' Kroll) aujto;n eij" th;n aujtou'
patrivda Makedonivan pro;" th;n aujtou' mhtevra ∆Olumpiavda, dou;"
krovtala kai; ajstragavlou"
(4) I 39, 8 par∆ ejmou' mhdevpote e[conte" (e[cete Kroll) ejlpivda
tinav, eja;n metabh'te th'" cwvra" ktl.
(5) II 5, 6 deilw'" e[cete, mh; th'/ basilikh'/ dunavmei ejmauto;n ejx
uJmw'n ajmunovmeno" (ajmuvnwmai Kroll)
__________
29
Può essere di qualche interesse osservare che lo stesso Kroll è autore
anche dell’edizione di Vettio Valente (Berlin 1908), dove, emendando il testo
o proponendo emendamenti in apparato, mostra di essere convinto che tale
costrutto sia pressoché sempre sospetto (seguito in questo dalla recente
teubneriana di D. Pingree, Leipzig 1986); ved. supra nt. 8. Il lavoro di K.
Wyss, Untersuchungen zur Sprache des Alexanderromans von PseudoKallisthenes (Laut- und Formenlehre des codex A), Freiburg 1942, che non
raramente prende le distanze dalle scelte linguistiche di Kroll e del quale
sarebbe stato interessante conoscere il parere, purtroppo non contempla la
sintassi.
121
(6) II 7, 9 e[sti gavr soi e[qnh Persw'n kai; Pavrqwn ..., i{na mhv
soi ta; Bavktrwn kai; ta; ∆Indw'n h] tw'n Semiravmew" melavqrwn
eijpwvn (ei[pw Kroll)
(7) II 9, 7 o[mnumi ... th;n ejpavnodon th;n ejn Makedoniva/ ginomevnhn
moi, wJ" eij nikhvsante" (wJ" ªeijº nikhvsante" Kroll) tou;" barbavrou" eij" th;n ÔEllavda uJpostrevyomen
(8) II 11, 2 boulovmeno" (bouvlomai Kroll) uJma'" citw'na" kataskeuavsanta" ajna; cilivou" pevmpein ejn ∆Antioceiva/
(9) II 14, 8 o{ti30 ou\n pavreimi a[ggelo" aujtou', ejgwv soi mhnuvwn
(mhnuvw Kroll)
(10) II 14, 10 oJ de; eijselqw;n (questo è l’unico caso in cui Kroll non
ha corretto; il perché non è chiaro: il contesto esclude che oJ ...
eijselqw;n sia un participio sostantivato; si ha qui una configurazione
simile all’esempio n. 19) eij" ta; mevlaqra Dareivou kai; ejpi; to;
dei'pnon eujqevw" ejkhruvcqh
(11) II 16, 2 kai; ga;r pollai; muriavde" mevllousai (muiw'n eijsin
Kroll), (12) ai} skopou'sai (skevpousi Kroll) to;n ajevra:(13) ejpa;n
de; aujtai'" ejpista'sai (ejpistw'si Kroll) sfh'ke", sobou'sin aujtav"
(14) II 16, 10 i{na Darei'o", oJ tosauvta" povlei" kai; e[qnh
uJpotavxa" kai; nhvsou" katadoulwsavmeno", ou|to" fuga;"
genovmeno" (Kroll, che nel suo testo ha effettuato una non necessaria
posposizione, propone in apparato di supplire ãejrhmwqh'/Ã, che desume
dalla recensione b e dall’Armena;31 sarebbe sufficiente, e molto più
__________
30
La crux che Kroll appone prima di o{ti non pare necessaria. Per la
proposta di una costituzione testuale di questo luogo e di II 16, 2 diverse da
quelle fatte da Kroll, ved. Il Romanzo di Alessandro, vol. II, testo, traduzione
e commento a c. di R. Stoneman e T. Gargiulo, Milano 2012, pp. XLVIII–
XLIX, L–LI, 42, 48.
31
Che non pochi interventi di Kroll, miranti ad avvicinare il testo delle
varie recensioni del Romanzo, siano discutibili, ho cercato di dimostrare in T.
Gargiulo, Un caso particolare di intertestualità nel Romanzo di Alessandro:
i fraintendimenti di B e G rispetto ad A, in: Tragico e comico nel dramma
attico e oltre: intersezioni e sviluppi parateatrali. Atti dell’Incontro di studi,
Cagliari 4–5 febbraio 2009, a c. di P. Mureddu, G. F. Nieddu e S. Novelli,
Amsterdam 2009, pp. 201–21.
ipabo_66.249.78.36
122
Tristano Gargiulo economico, supplire semplicemente ãh\/Ã,32 ma il confronto, piuttosto
puntuale, con l’esempio n. 6, sopra citato, lascia incerti se intervenire
o no) oJ hJlivw/ sunanateivla"
(15) III 17, 32 sunelqovnte" (suneish'lqon Kroll) dev moi ejk th'"
dunavmew" a[ndre" t∆ ajmavcairoi
(16) III 19, 4 ejxetavsa" (ejxhvtasen Kroll) aujto;n oJ Ptolemai'o"
(17) III 27, 5 ejnteu'qen de; ajnazeugnuvwn (ajnazeugnuvw Kroll) di∆
ejrhvmou kai; krhmnwvdou" cwvra"
(18) III 32, 6 proselqw;n de; plhsivon perieblevyato kai; ijdw;n (kai;
ei\de Kroll kai; ãejpevscenÃ ijdwvn Ausfeld) th;n eJautou' gunai'ka
ÔRwxavnhn prostrevcousan aujtw'/
(19) III 32, 12 aujtoi; me;n eijselqovnte" (ou|toi me;n eijsh'lqon Kroll),
kai; ejxaivfnh" ejk tw'n Makedovnwn pavntwn ejgevneto boh; kai;
33
sundromh; pro;" th;n aujlh;n ∆Alexavndrou .
Al termine di questa rassegna, si possono formulare le
seguenti considerazioni, che parlano in favore del participio
usato come o al posto di un verbo di modo finito:
__________
32
Costrutto perifrastico che ricorre anche in I 19, 4 h\n ajnaireqeiv"; II 15,
6 h\n genovmeno" e 17, 2 h\n lei'pon.
33
Più dubbi, anche se non sicuramente da scartare, appaiono altri quattro
casi. Nei primi due, Kroll risolve facilmente ogni problema sintattico eliminando un kaiv coordinativo e subordinando il participio al verbo principale: I
1, 3 ajll∆ eijshvrceto eij" ta; basivleia kai; ejlavmbane calkh'n lekavnhn
kai; gemivsa" aujth;n u{dato" ojmbrivou ªkai;º e[platten ejk khrou' ploiarivdia mikra; kai; ajnqrwpavria kai; ejnevballen aujta; eij" th;n lekavnhn; I 18,
6 proselqw;n hjspavsato to;n ∆Alevxandron, a{ma de; ejpi; tiv pavrei qevlwn
maqei'n ªkai;º ei\pe. Negli altri due, l’incertezza se correggere, con Kroll, o no
è ancora più grande: I 36, 5 eij de; mh; peisqh/'" (così Kroll; peisqei;" A)
toi'" keleuomevnoi" uJp∆ ejmou', ejkpevmyw katadivkou" sullhyomevnou" se;
II 12, 3 h[kousav se sunaqroivzonta o[clon kai; boulovmenon eJtevran
mavchn sunista'n pro;" ∆Alevxandron (così Kroll; A reca: ajkouvsasa ou\n
ajqroivzonta o[clon ktl.). Ho prudentemente escluso dal novero due luoghi
(II 17, 7 ajlla; qaumavzw, eij tw'n ejmw'n crhmavtwn Darei'o" lutrwvsetai
mhtevra kai; gunai'ka kai; tevkna, II 21, 23 tou;" ajnelovnta" to;n ejcqrovn
mou mevllw tima'n), dove Kroll dà in apparato, rispettivamente, qaumavzwn e
mevllwn come lezioni di A (da lui emendate in qaumavzw e mevllw) che, se fossero lezioni sicure, offrirebbero altri due limpidi esempi del nostro costrutto:
in realtà, il codice reca in entrambi i casi un’abbreviazione soprascritta
(qaumavz~ e mevll~) che, nello stesso ms., vale ora wn ora semplicemente w.
123
(a) i passi non mostrano anomalie eccentriche (il tempo del
participio, p. es., presente o aoristo, risulta sempre appropriato al
contesto, così come il numero e il genere sono senza eccezioni
in accordo con il soggetto della frase) o segnali di corruttela del
testo, ma rientrano tutti nelle tre tipologie che gli esempi finora
ritrovati hanno evidenziato come proprie di questa struttura (1. il
participio è unico verbo della frase o comunque verbo principale
[1, 3, 4,34 8, 9, 11, 15, 16, 17]; 2. il participio è coordinato35 con
un verbo di modo finito, seguendolo [2, 18] o precedendolo [10,
19]; 3. il participio è il verbo di una proposizione dipendente e
sostituisce, secondo quanto sintatticamente richiesto, un indicativo (7, 12) o un modo diverso dall’indicativo [5, 6, 13]);36
(b) in taluni casi sussistono precisi parallelismi con gli esempi
già ricordati di altri autori, il che fa apparire piuttosto improbabile che si tratti di coincidenza in errore di scribi diversi;37
__________
34
In questo caso non è facile dire se e[conte" funga da imperativo o da
futuro.
35
Tramite kaiv, che non assume mai, in questi esempi, valore di intensificatore o focalizzatore.
36
Dal momento che «il participio preceduto da ben definiti connettori
frasali rappresenta una nota strategia per esprimere alcune subordinate (w{",
a{te, kaivper)», Pierluigi Cuzzolin, che qui ringrazio per l’attenta lettura di queste
pagine, pone la questione se non sia «proprio a partire da costruzioni di
questo tipo che il participio si è esteso anche ad altre strategie di subordinazione». Si può forse rafforzare e allargare questo interessante suggerimento
ricordando come non raramente in greco il peso semantico di un enunciato
cada più sul participio che sull’effettivo verbo principale: cfr. A. Cavallin,
Zum Verhältnis zwischen regierendem Verb und Participium coniunctum,
Eranos 44, 1946, pp. 280–95; Kühner - Gerth, Ausführliche Grammatik, cit.
pp. 98–9; A. C. Moorhouse, The Syntax of Sophocles, Leiden 1982, pp. 254–
5 (§ 6. The participle expressing the leading idea).
37
Si confronti, ad es., il n. 10 con il passo di Cosma e Damiano cit. a p.
111 (48, 13–5 Deubner h[kousen de; kai; ta; qauvmata tw'n ejndovxwn ajnarguvrwn Kosma' kai; Damianou', kai; ejlqw;n ejn tw'/ oi[kw/ aujtw'n kai; levgei
pro;" tou;" aJgivou") e con quello di Malalas cit. alla nt. 16 (220, 22 Dindorf
ajpo; th'" kwvmh" ∆Hpeivrou ejxormhvsa" kai; h\lqen), oppure i nn. 4 e 12 con
il passo di Giuseppe Flavio cit. a p. 110 (Iud. Ant. IV 181 mhvt∆ eujsebeiva",
h|" nu'n peri; to;n qeo;n e[conte", katafronhvsante" eij" a[llon metasthvshsqe trovpon).
ipabo_66.249.78.36
124
Tristano Gargiulo (c) le correzioni da un participio presente nominativo a un
indicativo 1a persona sono molto facili, ma il loro numero mette in
sospetto [3, 6, 9, 17];
(d) ancora più numerosi sono i casi in cui la correzione, per
eliminare il participio, richiede un intervento più pesante ed è
perciò meno convincente [1, 2, 4, 5, 8, 10, 11, 12, 13, 15, 16, 18,
19];
(e) un indizio, che mi sembra di qualche utilità, a conferma del
fenomeno, lo si può anche vedere laddove, nel più tardo
passaggio dalla recensione b alla recensione g,38 il testo è
sintatticamente riconfigurato in modo da mostrare, tramite
l’equivalenza dei due enunciati, come il fulcro della frase
potesse risiedere senza difficoltà nel participio e il verbo di
modo finito mancare. In II 15, 5 il testo di b: tw/' lovgw/ ou\n
∆Alexavndrou oiJ Pevrsai ejkplagevnte" ejqauvmazon diventa in
g: to;n lovgon ∆Alexavndrou oiJ Pevrsai ejkplagevnte"; in II 15,
14 il testo di b: oJ de; ∆Alevxandro" pezeuvsa" ajpo; tou'
potamou' eu|ren Eu[mhlon ... kai; uJfhghvsato aujtw/' pavnta
ta; pracqevnta diventa in g: oJ de; ∆Alevxandro" pezeuvsa"
ajpo; tou' potamou' eu|ren Eu[mhlon ... kai; uJfhghsavmeno"
aujtw/' pavnta ta; pracqevnta.
Nel Romanzo di Alessandro (in A come nelle altre
recensioni) vi sono altri casi di participio usato assolutamente ad
inizio di frase, che, anche se in qualche modo affini al tipo che
stiamo considerando, configurano però una diversa, più classica
__________
38
In g le frasi participiali del tipo che ci interessa non sono rare (cfr., p.
es., II 33, 12 oJ de; frenhvrh" ∆Alevxandro" prostavssei a[fnw pu'r ajnavyai
kai; ejpidramw;n ejp∆ aujtouv", 39, 1 ejnqavde ∆Alevxandro" eijselqw;n h[geiren
aJyi'da ejn h/| panstrati; diabav"), ma non così significative come in A,
perché g, testo notevolmente più tardo (è posteriore, forse anche di parecchio,
al VI sec. d. C.), mostra ormai uno stadio avanzato nell’uso ‘assoluto’ del
participio, come si vede anche da una elevata frequenza e libertà di nominativi pendentes. Su alcune modalità di trasformazione con cui sono avvenuti i
passaggi da una recensione più antica ad una più recente del Romanzo di Alessandro, ved. T. Gargiulo, Un caso particolare di intertestualità nel Romanzo
di Alessandro, cit. alla nt. 31.
125
struttura di nominativus pendens,39 già ben attestata nella poesia
greca (ma anche nella prosa) di ogni periodo: I 35, 9 meta; de;
hJmevra" sumbalw;n oJ jAlevxandro" to;n povlemon ... h[noixan
ta;" puvla" aujtw'n ktl.; II 22, 1 oJ ∆Alevxandro" ajpokatasthvsa" ejn eijrhvnh/ th;n Persivda kai; keleuvsa" (Kroll
congettura ejkevleuse in luogo di kai; keleuvsa") ..., oiJ de;
ejxebovhsan; III 3, 3 polla; ou\n ejn eJautw'/ skeyavmeno" ...
poiei' de; (Kroll espunge il dev) tiv oJ frenhvrh" ∆Alevxandro"…40 È incerto, tuttavia, a mio avviso, se essi possano
aiutare a definire, nel senso di ‘costrutto assoluto’, lo statuto
sintattico del participio pro modo finito.
Può essere, infine, significativo osservare che la lingua del
Romanzo non presenta alcun influsso dell’ebraico, e che quindi
in essa l’adozione del costrutto che abbiamo discusso è da
inquadrare in àmbito autenticamente ed esclusivamente greco.
__________
39
Su cui ved., da ultimo, S. Novelli, Anomalie sintattiche e costrutti
marcati: l’anacoluto in Eschilo, Tübingen 2012 (in particolare p. 166 s., per
un esempio col participio iniziale in Aesch. Eum. 100 ss., che, almeno in
apparenza, sembrerebbe avere affinità col costrutto qui studiato; ma ved.
anche p. 91 ss. su Ag. 423, che sembra presentare il modulo del participio in
proposizione dipendente temporale, e pp. 111 ss., 160 ss. su Ag. 1096 e Eum.
68, dove apparentemente compare il modulo del participio unico verbo della
frase con valore di verbo principale).
40
Anche II 17, 3 ajpodou;" th;n mhtevra kai; gunai'ka kai; tevkna, ajnti;
de; touvtwn uJpiscnou'maiv soi tou;" qhsaurou;" dei'xai, sembrerebbe potersi
inquadrare in questa medesima struttura anacolutica; ma, poiché la frase si
apre con una formula deprecatoria (pro;" Dio;" iJkesivou), appare preferibile
accogliere l’imperativo congetturato da Kroll (ajpovdo") che mantenere il
participio tràdito da A.
ipabo_66.249.78.36
Orthographicum quoddam: reccido*
By TOM KEELINE, Cambridge / MA
For centuries Latin lexica have had a lemma rěcĭdo for the
word derived from re + cado meaning “I fall back”. Such lexica
then include a note about forms like reccido,1 implying that the
spelling with two cs is a special case. In fact the opposite is true:
recc- is the normal form, as the metrical evidence makes clear,
and should be printed in our classical texts and given as the
lemma in our dictionaries.
We would not expect re + cado to yield reccido. The
weakening of a to i is a standard effect of the word-initial stress
accent of Old Latin,2 but the gemination of c seems unmotivated. We do know, however, that the perfect tense has two cs:
re + cecidi > *rececidi > reccidi under the influence of Exon’s
Law, which states that “in a PItal. tetrasyllable with two light
medial syllables the second vowel regularly syncopates”;3 one
thinks for example of refero rettuli (< *retetuli). We might then
initially suspect that when the manuscripts transmit reccidit in
the present tense, they do so simply from scribal confusion with
the perfect.
__________
*
I should like to thank Christopher Krebs, John Blundell, and Jeremy Rau
for their helpful comments on this note.
1
A. Forcellini, Lexicon totius Latinitatis s.u. “prima syllaba a poetis
interdum producitur”, Lewis & Short s.u. “rĕcido”, OLD s.u. “FORMS: reccin pres. stem, usu. metri gratia, e.g. ...”. The linguists appear to be in
agreement as well: F. Sommer and R. Pfister, Handbuch der lateinischen
Laut- und Formlehre (Heidelberg 1977) §119 n. 3, “[the lengthened e in
recido] bei Dichtern vielfach zu prosodischen Zwecken verwendet”; cf. A.
Walde and J. B. Hofmann, Lateinisches etymologisches Wörterbuch (Heidelberg 1938–56) s.u. re-: “[the short e is normal;] anderes ist spät und
analogisch”.
2
Cf. e.g. attingo, conficio, recipio.
3
A. Sihler, New Comparative Grammar of Greek and Latin (Oxford
1995) §74.5; cf. M. Leumann, Lateinisch Laut- und Formenlehre (Munich
1977) 96, 587.
Glotta 89, 126–129, ISSN 0017–1298
© Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen 2013 Orthographicum quoddam: reccido 127
The manuscripts, however, are worthless as witnesses in such
questions of orthography. In the perfect tense, where reccidi is
unquestionably the right form, the manuscripts frequently offer
recidi.4 By the same token, even in the present tense the scribes
have a tendency to mix the two forms indiscriminately.5 While
to be certain the rec- forms seem to predominate in the present
by a ratio of about 2:1, at least as represented by our printed
editions of texts down to the time of Apuleius, we must weigh
our manuscripts, not simply count them, and amid such confusion it should be plain that no medieval manuscript has any
weight at all.
The only reliable testimony comes from what is guaranteed
by meter, and here the evidence is decisive: the present recc- is
metrically guaranteed in all but one instance in classical Latin
poetry,6 and that instance shall presently be shown to be
__________
4
E.g. Hisp. 17.2, Cic. Verr. II 5.163, Planc. 90, diu. 1.123, Att. 1.1.2, ad
Brut. 1.16.6, Nep. Alc. 7.1, Liu. 2.7.8, Liu. 24.34.11, Plin. nat. 5.17, Sen. nat.
3.pr.11, Val. Max. 9.9.2, Gell. 10.3.13 (= Cic. Verr. II 5.163), Suet. Iul. 17.1,
Tib. 72.2, Tac. ann. 6.10. I have no doubt that there are many other examples,
but since editors typically offer standardized forms and omit orthographical
trifles from their apparatuses, the only way to determine exact proportions
would be to collate all the manuscripts afresh – a pointless exercise.
5
For recc-, see e.g. Sall. rep. 1.3.2, Curt. 5.2.14, 9.5.25, and passim,
Quint. inst. 5.2.2 (but cf. 9.2.49), Sen. dial. 4.36.4, 5.39.3, Apul. met. 3.13,
8.5 (but cf. 1.11, 1.16), Fest. p. 277 M s.u. remorbesco, Porph. ad serm. 1.2.2,
1.6.85–7. As observed in the previous note, there are doubtless many more
such examples: to cite one such possibility, Winterbottom’s OCT prints
reccidere throughout, Radermacher’s Teubner recidere, neither with any note
in the apparatus; cf. Winterbottom, Problems in the Text of Quintilian
(London 1970) 56, which mentions that the mss agree on reccido at inst.
5.2.2, on recido at 9.2.49, and otherwise vary “a lot”. John Blundell kindly
informs me that in a sampling of manuscripts up to the 8th century reccpredominates by a wide margin.
6
Recc- guaranteed by meter in the present stem in Latin poetry: Plaut.
Poen. 485, Lucr. 1.857, 1.1063, 5.280, Ou. met. 6.212, 10.18, Phaedr. 3.18.5,
Prec. Terr. 14, Iuu. 12.54 (probably present in light of the surrounding
presents, but possibly perfect), Prud. apoth. 878, Orac. Sibyll. 8.243 (= Aug.
ciu. 18.23), carm. epigr. 1495.2 (= CIL 6.26003); the tense is unclear in Enn.
Var. 29 V. and Plaut. Poen. 1369; at Ter. Hec. 47 the word could conceivably
be scanned with a short or long initial syllable (the Bembinus [saec. iv-v]
actually offers recc-). The only metrically secure instances of rěcido in the
TLL material are found in the late 5th-century Dracontius (e.g. laud. dei
2.161) and, later still, once in the 6th-century Verecundus (satisf. 72).
ipabo_66.249.78.36
128
Tom Keeline illusory. This is the case from the time of Plautus all the way
through the late fifth century, and in a variety of meters. If the
true form of the verb were rĕcido, it would have been easy for
poets to use it in hexameters or trimeters,7 but they seem on the
contrary to have avoided it entirely. By contrast recc- is less
convenient for the dactylic poet, frequently yielding intractable
cretics, and yet it invariably appears.
What about the one place where rĕcido8 seems metrically
guaranteed (Plaut. Men. 518–21)?9
Pen.:
Numquam edepol quisquam me exorabit quin tuae
uxori rem omnem iam ut sit gesta eloquar;
omnes in te istaec recident contumeliae;
faxo haud inultus prandium comederis.
Since this is the sole instance of rĕcido in all of Latin poetry,
one is inclined toward suspicion, and with good reason, as it
turns out. Verse 520 contributes nothing to the sense of Peniculus’s speech; indeed it is merely a doublet of 521. Moreover,
if the verse were excised, the speech would suffer not a whit of
damage. The line thus bears the twin stamp of an interpolator: it
contains unexampled prosody and merely repeats the sense of
another verse. Though modern editors have ignored these problems, as long ago as 1869 Müller had condemned the line as
spurious,10 and we should do the same today.
There is thus not a single fully reliable instance in Latin
poetry of recido. Does it seem likely that Roman poets always
__________
7
Indeed anapaestic words like recido recidis recidit recidunt would have
been useful in hexameters.
8
Note that even here the Ambrosian palimpsest (saec. iv) appears to give
recc-, contra metrum.
9
Lewis & Short misleadingly state s.u. “with e long ... prob., also, Plaut.
Men. 3, 2, 54 (sic, should read 3, 2, 53, i.e. the present passage)”! Their
article contains some other lapses as well: one can expunge Prop. 4.8.44 and
Ou. rem. 611 from their list of instances of reccin the present, since these
are in fact examples of the perfect. Furthermore, at Cic. nat. deor. 2.66 one
should read reccidunt and oriuntur (indicatives); at Cic. Sull. 91 ad luctum
(not ludum); at Lucr. 1.857 ad nihilum.
10
C. F. Müller, Plautinische Prosodie (Berlin 1869) 228–9, on grounds of
prosody and superfluousness.
Orthographicum quoddam: reccido 129
chose to take poetic license and employ an exceptional form,
even though they could have used the normal one without any
difficulty? Hardly. Relying on the metrical evidence, we can
ignore the testimony of the manuscripts and write reccido everywhere. This not only accords with all the poetic evidence but
also results in better clausulae in Cicero, as Zielinski pointed
out.11 Indeed, it even obeys the precepts of the sixth-century
grammarian Cassiodorus, which, though somewhat confused,
nevertheless preserve a grain of truth: re praepositio non numquam, cum ad consonantes accedit, geminat illas, quod plerumque apud antiquos est, ut … cado reccido.12
That reccido is the standard present tense form appears a
secure fact, but its origin is harder to explain. As already noted,
it is not predicted by etymology. Nevertheless, speakers may have
been led to produce the form by two convergent analogies. First,
there may have been the influence of back-formation from the
perfect tense, where reccidi had arisen as the expected outcome
of linguistic change long before the beginnings of Latin literature. Such a process helps explain the future participle recasurus:
since the verb has no supine form, the future participle was
formed directly from a recomposed re + casurus. Secondly,
speakers may have felt an analogy with other compounds of
cado, all of which have a geminated c arising from assimilation
(e.g. accido, occido, succido). These two processes taken
together perhaps account for the form, which is otherwise quite
hard to parallel. Whatever its origin, however, the metrical
evidence should compel us to print reccido in our texts and
dictionaries.
__________
11
T. Zielinski, Das Clauselgesetz in Ciceros Reden (Leipzig 1904) 179,
noting that recc- never produces an inferior clausula and usually yields a
better one. Similar evidence exists for other authors; see e.g. on Lactantius T.
Stangl, “Lactantiana”, RhM 70 (1915) 246.
12
Cassiodorus, de orthographia (GLK 7.203.5). In full: re praepositio non
numquam, cum ad consonantes accedit, geminat illas, quod plerumque apud
antiquos est, ut duco redduco, cado reccido, tuli rettuli, pello reppello, do
reddo, lego rellego; unde relliquias et relligionem scribunt. quod apud poetas
ita oportere scribi concesserim, tametsi apud oratores quoque antiquos est;
nobis iam decor et leuitas obtinenda est, qui maximus fructus est latinitatis.
ipabo_66.249.78.36
Zur Verwendung des Perfektfuturs im Altlatein1
Von MARTIN JOACHIM KÜMMEL, Freiburg
1. Präliminarien
Anstelle der sonst üblichen Begriffe Futur II oder Futurum
exactum werde ich hier der Deutlichkeit halber Perfektfutur
sagen, während ich das Futur I als Präsensfutur bezeichne, und
dies gilt auch für die übrigen Kategorien. (Noch besser wäre es
wohl, die in der lateinischen Grammatik bekannten Begriffe
infectum und confectum zu verwenden, doch sind diese in der
Praxis zu wenig vertraut.) Den Konjunktiv Präsens/Perfekt möchte ich aufgrund des Gebrauchs in Hauptsätzen als Konjunktiv
vom Konjunktiv Imperfekt/Plusquamperfekt als Konditionalis
differenzieren (auch wenn dies nicht zur consecutio temporum
passt, vgl. Meiser 1993: 172ff., 181ff., 191ff.). Unberücksichtigt
bleibt hier generell der morphologisch abweichende Typ faxo,
amasso, der zumindest synchron nicht (mehr?) zum lateinischen
Perfektstamm gehört.
2. Das Problem
Bekanntlich wird das Perfektfutur im Altlatein und besonders
bei Plautus oft so verwendet, dass man vom Standpunkt der
späteren Sprache aus ein Präsensfutur erwarten würde, in den
Worten von Lindsays Spezialsyntax: „the Tense sometimes has
its true function, e.g. Bacch. 708 hoc ubi egero, tum istuc agam,
but often has practically the same function as the Future“.
(Lindsay 1907: 60).
__________
1
Dieser Aufsatz geht zurück auf meinen Probevortrag beim Habilitationskolloquium vor der Philologischen Fakultät der Universität Freiburg am
01.06. 2005. Die Vortragsform wurde im Wesentlichen beibehalten.
Glotta 89, 130–145, ISSN 0017–1298
Zur Verwendung des Perfektfuturs im Altlatein
131
Ein Beispiel: In Plautusʼ Cistellaria, Vers 498f. stehen in
einer doppelten Selbstverfluchung im Nebensatz jeweils die
Perfektfutura duxero und dedero, die beide nicht eigentlich als
vorzeitig zu deuten sind und somit vom klassischen Standpunkt
aus im Gebrauch von den normalen Präsensfutura ducam und
dabo nicht abweichen:
Plautus, Cistellaria, Vers 497–499:
Alc. Di me perdant -- Mel. Quodcumque optes, tibi uelim contingere.
Alc. Si illam uxorem duxero umquam, mihi quam despondit pater.
Mel. Et me, si umquam tibi uxorem filiam dedero meam.
‚Die Götter mögen mich verderben … – Was auch immer du
wünschst, möchte ich, dass es dir geschehe. – … wenn ich jemals jene
zur Frau nehmen werde, die mir der Vater ausgesucht hat. – Und
mich, wenn ich dir jemals meine Tochter zur Frau geben werde’.
3. Historische Erklärung: perfektiv, nicht perfektisch
Man kann zwar hier eine synchrone syntaktische Bedingung
finden, wie man wiederum bei Lindsay nachlesen kann: „It [the
Future Perfect] is normal after si in threats, e.g. si attigeris,
vapulabis, where an Aoristic sense is perhaps conveyed; while
after nisi the [Future] Present is normal, e.g. nisi abis, vapulabis“. (Lindsay 1907: 60f.), und Ähnliches gilt auch für Verfluchungen und Versicherungen (Sjögren 1906: 186–192), doch
bleibt es vom Standpunkt der späteren Sprache aus auffällig.
Und dies gilt auch für andere Stellen, sogar in Hauptsätzen.
3.1. Vergleichende Rekonstruktion
Wie schon bei Lindsay angedeutet, hat man schon lange vermutet, dass hier Reste einer älteren perfektiven (nicht perfektischvorzeitigen) Funktion vorliegen (vgl. die Literatur bei van Wijk
1905), wie man sie auch für die Konjunktive und den Infinitiv
des Perfektstammes anzunehmen hat, da ihre sprachhistorischen
Vorläufer dem indogermanischen und griechischen Aorist
zugehörig waren. Denn das lateinische Futur geht auf den indo-
ipabo_66.249.78.36
132
Martin Joachim Kümmel
germanischen Konjunktiv zurück (*eseti > erit = ved. ásati), und
das lateinische Perfekt beruht auf einem Synkretismus des
indogermanischen Aorists mit dem Perfekt, klare Fortsetzer von
Aoristen z. B. in lat. dīx-it = gr. é-deiks-e, fēc-it = é-thēk-e. Für
den indogermanischen Aoriststamm (inklusive der Modalformen) wird allgemein eine perfektive Funktion rekonstruiert, die
ihn vom sogenannten Präsensstamm als dem imperfektiven
Stamm unterschied (vgl. Tichy 2004: 82f., 121ff.; Jasanoff
2003: 30f., 43f.). Im Lateinischen ist diese Funktion ja beim
Indikativ des Perfekts in Opposition zum Imperfekt auch deutlich erhalten und lebt noch heute in den romanischen Sprachen
weiter.
Dürfen wir daher wie beim Indikativ Reste einer älteren
perfektiven (nicht perfektischen) Funktion auch beim zugehörigen Konjunktiv/Futur vermuten?
4. Die implizierte Chronologie und das Lateinische
Wenn man die Texte analysiert, kann man eine durchgehende
aspektuelle Unterscheidung nicht feststellen: Sjögren (1906) hat
für das Futurum gezeigt, dass das Perfektfutur in vielen Fällen
mit dem Präsensfutur vertauschbar zu sein scheint, also nicht
generell funktional von diesem unterschieden ist. Das Präsensfutur könne zudem auch „aoristische Aktionsart“ haben. Dies ist
bemerkenswert: Im vorhistorischen Latein (oder seinen Vorläufern) muss es einmal eine scharfe Trennung des Konjunktivs
Aorist (und Perfekt) vom Konjunktiv Präsens gegeben haben,
und im klassischen Latein werden die gleichen Kategorien – nun
als Futur des Perfekts bzw. des Präsens – ebenfalls klar unterschieden. In der dazwischenliegenden Zeit soll aber diese
Differenzierung nicht recht funktioniert haben, schematisch
nach folgender Periodisierung:
- Vorhistorisch: klare Differenz
Konjunktiv Präsens (imperfektiv) vs. Konjunktiv Aorist/Perfekt
(perfektiv)
133
- Altlatein: keine klare Differenz
Präsensfutur ~ Perfektfutur (weitgehend) vertauschbar
- Klassisches Latein: klare Differenz
Futurum (allgemein) vs. Futurum exactum (anterior)
Wie kann das sein? Hätten die beiden Kategorien nicht in der
mittleren Periode der „Unklarheit“ zusammenfallen müssen?
Diese Darstellung der altlateinischen Situation ist nun freilich
überzeichnet, denn die Unklarheit betraf eben nicht den gesamten Funktionsbereich, die beiden Futura waren keineswegs in
allen Fällen vertauschbar:
- Das Perfektfutur hatte sicherlich schon immer den Anwendungsbereich der eigentlichen Vorzeitigkeit für sich: Diese wird ja
im Griechischen, das hier als Modell für die allen italischen
Sprachen zugrundeliegende Sprachstufe dienen kann, durch die
perfektiven Kategorien ausgedrückt, die durch den Kontext diese
Lesart erhalten. Ähnliches gilt auch sonst für Sprachen mit einer
perfektiven Aspektkategorie. In diesem ihm auch später zugehörigen Bereich wäre das Perfektfutur also schon immer unvertauschbar gewesen.
- Umgekehrt müsste allein das Präsensfutur ursprünglich für
imperfektive Kontexte verwendbar gewesen sein, und in solchen
bleibt es ja auch später als normales Futur erhalten. In diesem
Bereich sollte also alles beim alten geblieben sein.
Bei Sjögren (1906) allerdings wird nur festgestellt, dass die
Kategorien bei nichtvorzeitigem Gebrauch grundsätzlich vertauschbar sind, einen unvertauschbaren Bereich des Präsensfuturs hat er nicht ermittelt. Man würde also den Eindruck
gewinnen, als habe nur das Perfektfutur den gesamten Funktionsbereich des Futurs, während das Präsensfutur sich davon
dadurch unterscheide, dass es nicht bei Vorzeitigkeit stehen
könne.
ipabo_66.249.78.36
134
5. Lesarten und Kontextdifferenzierung
Wenn wir uns fragen, wie die Entwicklung im Lateinischen
verlaufen sein müsste, kann eine Differenzierung verschiedener
Lesarten oder Gebrauchstypen nützlich sein. Und dabei müssen
wir Sjögrens Auffassung schon deshalb erneut prüfen, weil sein
Modell von „aoristischer Aktionsart“ veraltet ist. Denn erst zwei
Jahre nach ihm schlug Sigurd Agrell (1908) vor, grammatischen
Aspekt und Aktionsart zu differenzieren, was sich danach
zumindest in der kontinentalen Sprachwissenschaft durchsetzte.
Unter Aspekt verstehe ich hier nur „ASPECT1“ im Sinne von
Sasse 2001, also eine grammatische Unterscheidung von imperfektiv/perfektiv (und auch nicht progressiv, habituell o.ä.), die
erst auf Satz- und Textebene wesentlich wird. Dagegen nenne
ich intrinsische Eigenschaften der Prädikate (Sasses „ASPECT2“)
„Aktionsart“. Wenn ich also „imperfektiver Kontext“ sage,
meine ich nicht wie Sjögren (und nicht wenige andere auch
später noch) eine irgendwie „durative“ Art der Handlung, sondern einen Kontext, in dem z.B. im Neugriechischen das
imperfektive Futur (formal Konjunktiv Präsens mit θα) stehen
könnte, ähnlich auch sonst in Sprachen mit Aspektdifferenzierung in futurischen Kontexten. Das entscheidende Merkmal
ist unbounded (unabgeschlossen genügt nicht) – zu solchen
Kontexten vgl. auch Raible 1991: 197–199.
6. Überlegungen zur Entwicklung
Als Parallelfall, in dem wir einen Teilzusammenfall getrennter
Aspektkategorien mit Verschiebung zugunsten der einen beobachten können, können die indoiranischen Präterita gelten, die
Tichy (1997: 595–606) diskutiert hat. Dort wurde das ältere
aspektuelle System der Präterita durch ein Tempussystem abgelöst, in dem das neue Merkmal der Proximalität/Distalität
dominiert (vgl. auch Kümmel 2002: 42f.; Dahl 2010). Der ursprünglich einfach perfektive Aorist wurde (im Wesentlichen)
135
auf die Verwendung in aktuellen Kontexten, also als proximales
Präteritum eingeschränkt, während das ursprünglich nur imperfektive „Imperfekt“ als neue Funktion die Darstellung präteritaler Sachverhalte auch bei perfektivem Kontext gewann, wenn
diese nicht proximal zur Gegenwart waren. Schematisch kann
man die Entwicklung nach Kontext-Lesarten so darstellen:
imperfektiv
voruriir.
uriir.
ved.
Imperfekt
Imperfekt
Imperfekt
Perfektiv
distal
proximal
Aorist
Aorist
Aorist/Imperfekt
Aorist
Imperfekt
Aorist
Bei perfektiven Kontexten der aktuellen Gegenwart blieb der
Aorist also stets unvertauschbar, desgleichen das Imperfekt bei
imperfektiven Kontexten (bei denen Proximalität offenbar nicht
so relevant war). Dagegen muss es in einer Zwischenphase
partielle Vertauschbarkeit in perfektiven Kontexten mit distaler
Zeitlage gegeben haben (nach Dahl 2010 noch im Frühvedischen).
7. Hypothese über die lateinische Entwicklung
Eine grundsätzlich ähnliche Entwicklung kann man sich auch
für das italisch-lateinische System der Modalkategorien vorstellen, wozu hier wegen seiner Herkunft aus dem alten Konjunktiv
auch das Futur gerechnet wird. Nach dem alten aspektuellen
System (Zusammenfall von Aorist und Perfekt vorausgesetzt)
muss primär zwischen imperfektiven und perfektiven Kontexten
unterschieden werden; bei letzteren kann die vorzeitige (anteriore) Lesart als Sonderfall unterschieden werden. Nehmen wir den
Endbefund der klassischen Sprache, so sehen wir, dass das
Präsensfutur im Lateinischen praktisch alle Lesarten des
Perfektfuturs übernommen hat, außer dem Gebrauch bei
Vorzeitigkeit, d.h. es hat zusätzlich zu dem Gebrauch in allen
imperfektiven Kontexten auch alle nicht-vorzeitigen perfektiven
ipabo_66.249.78.36
136
Lesarten hinzugewonnen. Das bedeutet für die Übergangszeit,
dass das Perfektfutur in diesen Kontexten immer mehr (in
relikthafter Verwendung) mit dem Präsensfutur vertauschbar
geworden sein muss, bis es von diesem in allen Vertauschbarkeitsfällen verdrängt wurde:
imperfektiv
vorlat.
alat.
klass.
Präsens
Präsens
Präsens
Perfektiv
nicht anterior
Perfekt
Präsens/Perfekt
Präsens
Anterior
Perfekt
Perfekt
Perfekt
8. Implikationen der Hypothese
Dies impliziert eine Verdrängung des Perfektfuturs in perfektiven Kontexten durch das Präsensfutur, erlaubt aber kein
Eindringen des Perfektfuturs in die Domäne des expansiven
Präsensfuturs, also in imperfektive Kontexte. Hier scheint sich
nun aus der bisherigen Forschung Widerspruch zu erheben:
Explizit haben sowohl van Wijk (1905: 470) als auch Sjögren
(1906: 194) behauptet, das Perfektfutur erscheine auch (selten)
in imperfektiven Kontexten. Dagegen lässt sich folgendes
einwenden:
Den Sonderfall fuero hat schon Sjögren beseitigt. Hier liegt
Vermischung zweier Verben oder zweier Aktionsarten vor
(stativ/inzeptiv, s. Sjögren 1906: 173–185 gegen die „Tempusverschiebung“ von Blase 1898; 1903: 188–193): Tatsächlich
bedeutet fuero meist ‘ich werde werden’ und ist noch Futur zu
fieri. Ähnliches gilt modifiziert auch für habuero, das dann eher
‘ich werde festhalten/bekommen’ bedeutet.
Die anderen Gegenbeispiele betreffen alle einen Bereich, in dem
sich Aspektsprachen unterschiedlich verhalten können, nämlich
voll durative Aktionsarten, besonders Zustände: Während die
von van Wijk verglichenen slavische Sprachen hier oft eine
Klasse von Imperfektiva tantum haben, können das Griechische
137
und noch mehr das Lateinische (und seine romanischen Tochtersprachen) oft perfektive Formen dieser Verben bilden.
9. Aktionsart und Aspekt
Darauf, dass gerade solche Aktionsarten eine Rolle spielen
können, gibt schon die folgende Beobachtung Lindsays (1907:
60) einen Hinweis: „Why Plautine (and Terentian) Latin should
use only the Fut., never the Fut. Perf., of oportet, possum, volo,
is not clear“. Warum haben gerade diese Modalverben kein
Perfektfutur?
Die genannten Verben sind Zustandsverben mit deutlich
stativ-durativer Aktionsart, die ein überwiegendes Vorkommen
in imperfektiven Kontexten und weitgehendes Fehlen in perfektiven Kontexten zur Folge haben sollte. Denn es ist eine
bekannte Tatsache, dass man zwar Aktionsart (als intrinsisch,
lexikalisch bzw. propositional) von Aspekt (grammatisch, mehr
auf Satz- und Textebene) unterscheiden sollte, dass aber
andererseits auch eine gewisse Korrelation existiert: Verben mit
atelischen Aktionsarten (besonders stativischen) werden tendenziell („unmarkiert“) vor allem im imperfektiven Aspekt gebraucht, während bei Verben telischer Aktionsarten der
Gebrauch im perfektiven Aspekt dominiert (Raible 1991: 199f.;
Sasse 2001: 8–9).
Die Grundaktionsarten kann man verschieden einteilen; ich
gebe hier einerseits Vendlers Klassifikation (1967), andererseits
die Klassifikation nach Breu/Sasse (bei Leluda-Voß 1997: 22–
27 für das Griechische, weiteres bei Raible 1991: 208–210):
Atelisch: states = total stativ (TSTA), inzeptiv-stativ (ISTA);
activities = Aktionsverben (AKTI), Telisch: accomplishments =
graduell-terminative Verben (GTER); achievements = totalterminative Verben (TTER).
Zu erwarten ist eine relative Distributionslücke der Perfektkategorien atelischer Verben (vor allem „states“, TSTA), oder
deren Vorkommen nur vorzeitig und bei perfektivem Kontext
ipabo_66.249.78.36
138
(besonders „activities“, AKTI), d.h. als Relikt in ihrem traditionellen Anwendungsbereich. Begünstigend könnten noch dichterische Bedürfnisse wirken: Schon Sjögren stellt fest, dass die
Mehrzahl der Formen am Versende steht, wofür sie wegen ihrer
metrischen Struktur besonders geeignet sind.
10. Überprüfung der Hypothese
Für die Überprüfung dieser von der Hypothese vorausgesagten
Erwartung habe ich atelische Verben bei Plautus hinsichtlich
ihres Perfektfuturs untersucht. Dabei habe ich mich auf die
verba simplicia beschränken, da bei diesen atelische Aktionsarten sehr viel häufiger sind als bei Komposita. Berücksichtigt
wurden Verben, wenn a) ein Perfektfutur belegt ist oder sie b)
über 20 finite Belege aufweisen und unter diesen wenigstens ein
Präsensfutur erscheint. Die Komödien des Plautus sind ein
Kleincorpus mit etwa 165.000 Wörtern.
26 stative Verben (TSTA/ISTA) sind ohne Beleg eines
Perfektfuturs: amāre, audēre, cēnsēre, cessāre, cōgitāre,
cubāre, cupere, dēbēre, dormīre, egēre, gaudēre, habitāre,
licēre, lubēre, mālle, metuere, patī, posse, quīre, sapere, stāre,
tacēre, tenēre, ualēre, uelle, uīuere. Das Gleiche gilt für 28
durative und iterative Prozessverben (AKTI): auscultāre,
cēdere, cēlāre, clāmāre, currere, discere, errāre, ēsse, fābulārī,
loquī, memorāre, ōrāre, peccāre, petere, pōtāre, properāre,
pultāre, putāre, quaerere, quaeritāre, rogitāre, rogāre, scrībere,
seruīre, simulāre, spectāre, uāpulāre, uerberāre. Belegt war ein
Perfektfutur 42 mal bei insgesamt 28 Verben, 13 mit stativer
Aktionsart und 15 Aktionsverben, davon 21 vorzeitig, 21 nicht
vorzeitig.
Als Gegenstichprobe wurden häufige telische Verben (> 100
Belegstellen) überprüft. Bei diesen ergaben sich durchschnittlich
mindestens doppelt so viele Belege des Perfektfuturs (1,2–6%
der Formen gegenüber 0–1,1%). Die Seltenheit der Perfektfutura
bei atelischen Verben dürfte also kaum reiner Zufall sein.
139
Betrachtet man die Belege atelischer Verben im Einzelnen
(siehe Anhang), ist zunächst das „echte“ Futurum exactum
auszuscheiden, also klar vorzeitige Verwendung, so Nr. 10, 15b,
17, 21.
Wenigstens sachlich vorzeitig sind auch die Belege in Nebensätzen, die ein für den Hauptsatz gegebenes und vorauszusetzendes Faktum nennen und durch si, ubi oder ein Relativpronomen
subordiniert werden, z.B. Nr. 25a aus Miles gloriosus 565f.
egone si post hunc diem muttiuero, ..., dato excruciandum me
‘wenn ich nach diesem Tag muckse, … dann übergib mich der
Folter’. Dies trifft auch zu in Nr. 5c, 6, 7a, 11, 13, 19, 22, 24,
25b, 26c und 27.
Desgleichen können als vorzeitig auch die Fälle betrachtet
werden, wo ein teilweise statives Verb inzeptive Aktionsart
haben kann (also zur ISTA-Klasse nach Breu/Sasse gehört) und
diese vorliegt (wie auch in den Perfekta dieser Verben). Hierher
gehören Nr. 1, 2, 3 und 12 (sowie mit klar vorzeitiger Funktion
auch Nr. 10). Alle diese Fälle setzen also die boundedness der
jeweiligen Situation voraus und bieten damit einen perfektiven
Kontext.
Die nächste ebenfalls perfektive Umgebung ist das, was
Sjögren „Nachsatz einer hypothetischen Periode“ nennt: Hier
steht das Perfektfutur, um eine aus der gegebenen Nebensatzhandlung resultierende Handlung zu bezeichnen, die eine
Handlungsfolge abschließt – in der Vergangenheit würde hier
Perfekt stehen, also das perfektive Präteritum (vgl. Tichy 1999:
136f.). So verhält es sich in Nr. 13b Amphitruo 313f. Me. quid
si ego illum tractim tangam, ut dormiat? So. seruaueris ‚Was,
wenn ich jenem eins überziehe, so dass er schläft?‘ – ‚Du wirst
[mich] retten‘, ähnlich auch Nr. 6b und 23.
Damit verwandt ist der Fall, wenn mehrere gleichzeitige
Handlungen durch die letzte resümierend abgeschlossen werden
(vgl. Tichy 1999: 141), so in Nr. 14: Asinaria 280 erum in
opsidione linquet, inimicum animos auxerit ‚er wird den Herrn
in der Blockade verlassen, und [dadurch] den Mut der Feinde
stärken’.
ipabo_66.249.78.36
140
Es bleibt eine Gruppe von Belegen, bei denen Sjögren keine
„aoristische“ Aktionsart zu erkennen vermochte – zu Recht,
denn es handelt sich um durative Verben. Dies bedeutet aber
nicht, dass kein perfektiver Aspekt vorgelegen haben könnte
(was auch van Wijk wegen seiner zu stark vom Russischen
abhängigen Sichtweise entgehen konnte – dort sind solche
Verben nämlich bisweilen ohne Aspektpartner und scheinbar
„imperfektiv“, s. o.). Vielmehr ist der perfektive Aspekt auch
bei dauernden Handlungen ohne inhärente Grenze angezeigt,
wenn die dauernde Handlung als Einzelereignis zusammengefasst wird (vgl. Raible 1991: 200f.), weil a) nur das Faktum
genannt werden soll, ohne Rücksicht auf die Dauer oder b)
durch eine Zeitangabe eine Begrenzung gegeben wird. In beiden
Fällen würde in der Vergangenheit ein perfektives Präteritum
stehen, im Lateinischen also ein Indikativ Perfekt. Als Beispiel
für a) kann dienen Nr. 7b Casina 787 cras habuero ... conuiuium
‘morgen werde ich ein Festmahl halten’. Ähnlich verhalten sich
Nr. 7b, 9, 13, 15a, 16, 18, 20 und 26ab und vielleicht 4, 6b.
Eine explizite Angabe des Zeitraums findet sich z.B. in Nr. 8:
Asinaria 327f. age, age, mansero tuo arbitratu, uel adeo usque
dum peris ‘Wohlan, ich werde nach deinem Belieben bleiben,
oder sogar, bis du vergehst’ („durativ“ nach Sjögren 1906: 150,
194). Hier kann auch ein bei Sjögren explizit genanntes „duratives“ Gegenbeispiel angeschlossen werden, das ein Kompositum betrifft, nämlich das Perfektfutur von dēlectāre ‘erfreuen;
sich erfreuen’ (Nr. 29), bei dem in beiden Belegen ausdrücklich
eine begrenzte Zeitspanne genannt wird, in der die Handlung
stattfindet.
11. Ergebnis
Die aus der Hypothese folgende Erwartung hat sich damit bestätigt: Nichtvorzeitige Perfektfutura von Verben mit atelischen
Aktionsarten kommen im Altlateinischen des Plautus nur in eng
begrenzten Kontexten vor, die typisch für perfektive Kategorien
141
sind und daher als Relikte eines vorlateinischen Zustands
verständlich sind. Definiert man Aspekt wirklich als Aspekt und
berücksichtigt man die Korrelation von Aspekt und Aktionsart,
so bietet der Gebrauch des Perfektfuturs im Altlatein also
durchaus das Bild, das wir erwarten, wenn ein Übergangsstadium von einem ehemals einfach perfektiven Futur zu einem
vorzeitigen Futur vorliegt.
Literatur
Agrell, S. (1908): Aspektänderung und Aktionsartbildung beim polnischen
Zeitworte: ein Beitrag zum Studium der indogermanischen Praeverbia
und ihrer Bedeutungsfunktionen. Lunds Universitets Årsskrift n.s. I,IV.2.
Blase, H. (1898): Zur Geschichte der Futura und des Konjunktivs des
Perfekts im Lateinischen. Archiv für lateinische Lexikographie und
Grammatik 10, 313–343.
– (1903): Tempora und Modi; Genera Verbi. In: G. Landgraf (Hrsg.),
Historische Grammatik der lateinischen Sprache. Band 3,1: Syntax des
einfachen Satzes, 1. Heft, Leipzig: Teubner, 97–312.
Comrie, B. (1976): Aspect: an Introduction to the Study of Verbal Aspect and
Related Problems. Cambridge.
– (1985): Tense. Cambridge.
Dahl, Ö. (1985): Tense and aspect systems. Oxford.
Dahl, E. (2010): Time, Tense and Aspect in Early Vedic Grammar. Exploring
inflecional semantics in the Rigveda. (Brill’s Studies in Indo-European
Languages & Linguistics, 5). Leiden-Boston: Brill.
Galton, A. (1984): The Logic of Aspect. An Axiomatic Approach. Oxford.
Hoffmann, K. (1970): Das Kategoriensystem des indogermanischen Verbums. Münchener Studien zur Sprachwissenschaft 28, 19–41.
Kühner, R. – Stegmann, C. (1912): Ausführliche Grammatik der lateinischen
Sprache. Zweiter Teil: Satzlehre, 2 Bde. Hannover, Hahnsche Buchhandlung [Nachdruck 1966].
Kümmel, M. J. (2002): Temporalität und Aspektualität in der neueren germanistischen Forschung. Rezensionsaufsatz zu W. Abraham, Th. Janssen
(Hrsgg.): Tempus - Aspekt - Modus. Die lexikalischen und grammatischen
Formen in den germanischen Sprachen, Tübingen 1989, und M. Herweg,
Zeitaspekte: Die Bedeutung von Tempus, Aspekt und temporalen Konjunktionen, Wiesbaden 1990. Philologia Fenno-Ugrica 7 (2001/2002), 35–45.
ipabo_66.249.78.36
142
Leluda-Voß, Chr. (1997): Die Perfekterfassung und das neugriechische
Perfekt. Frankfurt a. M.-Berlin-New York usw.
Leumann, M. (1977): Lateinische Laut- und Formenlehre. (Handbuch der
Altertumswissenschaft 2. Abteilung, 2. Teil: Lateinische Grammatik von
Leumann-Hofmann-Szantyr, 1. Band). Neuausgabe. München: C. H. Beck.
Lindsay, W. M. (1907): Syntax of Plautus. Oxford: Parker.
Meiser, G. (1993): Uritalische Modussyntax: zur Genese des Konjunktiv
Imperfekt. In: Oskisch-Umbrisch. Texte und Grammatik. Arbeitstagung
der Indogermanischen Gesellschaft und der Società Italiana di Glottologie vom 25. bis 28. September 1991 in Freiburg, ed. Helmut Rix, Wiesbaden: Reichert, 167–195.
Pinkster, H. (1984): Latijnse syntaxis en semantiek. Amsterdam: Grüner.
Raible, W. (1991): Types of Tense and Aspect Systems. In: Toward a Typology of European Languages, ed. J. Bechert, Berlin-New York: Mouton
de Gruyter, 195–214.
Sasse, H.-J. (2001): Recent activity in the theory of aspect: accomplishments,
achievements, or just non-progressive state? (Arbeitspapiere des Instituts
für Sprachwissenschaft 40). Köln.
Sjögren, H. (1906): Zum Gebrauch des Futurums im Altlateinischen. Leipzig:
Harrassowitz.
Tichy, E. (1997): Vom indogermanischen Tempus/Aspekt-System zum vedischen Zeitstufensystem. In: E. Crespo, J. L. García Ramón (Hrsgg.),
Berthold Delbrück y la sintaxis indoeuropea hoy. Madrid-Wiesbaden,
589–609.
– (1999): Nicht nur Vordergrund und Hintergrund: Zum Aspektgebrauch im
Neuen Testament und im homerischen Epos. In: N. Nebes (Hrsg.), Tempus und Aspekt in den semitischen Sprachen. Jenaer Kolloquium zur
semitischen Sprachwissenschaft, Wiesbaden: Harrassowitz, 127–145.
– (2004): Indogermanistisches Grundwissen für Studierende sprachwissenschaftlicher Disziplinen. 2., überarbeitete Auflage. Bremen: Hempen.
Vendler, Z. (1967): Linguistics in Philosophy. Ithaca (NY).
Wijk, N. van (1905): Die altitalischen Futura. Indogermanische Forschungen
17, 465–483.
143
Anhang: Belege des Perfektfuturs atelischer Verben
a) Stative Verben
1. crēdere ‘vertrauen, glauben’: Trinummus 607 si hoc non credis, ego
credidero ‘wenn du das nicht glaubst, ich werde es glauben!’; sonst
telisch ‘anvertrauen’
2. esse ‘sein’: Vorzeitig Casina 547f. ubi nuptiae fuerint, tum istam
conuenibo ‘wenn die Hochzeit gewesen ist, dann werde ich sie besuchen’.
Sonst als periphrastisches Hilfsverb oder in telischer Bedeutung ‘werden’
(zu fieri, Sjögren 1906: 179–83): z.B. Trinummus 157f. si quid eo fuerit,
... habeo dotem ‘wenn ihm etwas geschehen wird, habe ich die Mitgift.’
3. scīre ‘wissen, erfahren’: a) Persa 218 nisi scieró prius ex te, tu ex me
numquam hoc quod rogitas scies ‘wenn ich es nicht vorher von dir
erfahre(n habe), wirst du von mir niemals erfahren, wonach du fragst’
b) Bacchides 358f. quom se excucurrisse illuc frustra sciuerit ‘wenn er
erfährt, dass er vergeblich dorthin hinausgereist ist’
4. nescīre ‘nicht wissen’: Miles gloriosus 572f. posthac etiam illud quod
scies nesciueris nec uideris quod uideris ‘danach wirst/sollst du auch das,
was du weißt, nicht wissen, und nicht sehen, was du siehst.’ (präventive
Illokution)
5. cauēre (ne ...) ‘aufpassen, sich hüten’: a) Menaechmi 270f. id utrumque,
..., cauero, ne tu delinquas ... ‘Auf diese beiden werde ich achten, …, dass
du keinen Fehler machst…’
b) Pseudolus 478 iratus sit: ego ne quid noceat cauero ‘Mag er wütend
sein: ich werde aufpassen, dass er keinen Schaden tut’
c) Menaechmi 347 tum demum sciam recte monuisse, si tu recte caueris
‘Dann erst werde ich wissen, dass mein Rat richtig war, wenn du richtig
aufgepasst hast’
6. cūrāre ‘(be)sorgen’ (Aktionsverb?): a) Amphitruo 741 tua istuc refert – si
curaueris ‘das betrifft d i c h – wenn du dich darum kümmerst’ (Antwort
auf eine Drohung)
b) Captivi 314: is, uti tu me hic habueris, proinde illum illic curauerit
‘Wie du mich hier behandel(t habe)n wirst, genauso wird er sich um jenen
dort kümmern’ (formale Assimilation? Sjögren 1906: 169f.)
7. habēre ‘halten, haben’: a) Captivi 314 uti tu me habueris (sachlich
vorzeitig)
b) Casina 787 cras habuero ... conuiuium ‘morgen werde ich ein Festmahl haben’.
8. manēre ‘bleiben’: Asinaria 327f. age, age, mansero tuo arbitratu, uel
adeo usque dum peris ‘Wohlan, ich werde nach deinem Belieben bleiben,
oder sogar, bis du vergehst’ („durativ“ nach Sjögren 1906: 150, 194)
9. placēre ‘gefallen’: Menaechmi 670 si tibi displiceo, patiundum; at
placuero huic Erotio, ... ‘wenn ich dir missfalle, ist das zu dulden; aber
gefallen werde ich dieser Erotium hier.’
ipabo_66.249.78.36
144
Grenzfälle (eher ingressiv/terminativ?):
10. merēre ‘verdienen, erwerben’: Asinaria 377: quin promitto, inquam,
hostire contra ut merueris ‘wirklich verspreche ich, sag’ ich, zu vergelten,
wie du es verdient hast.’
11. optāre ‘auswählen, wünschen’: Asinaria 720 LE. opta id quod ut
cóntingat tibi uis. ARG. quid si optaro? LE. eueniet. ‘Wünsche das, was du
willst, dass es dir zufallen soll! – Was, wenn ich es (ge)wünsch(t hab)e? –
Es wird geschehen.’
12. sedēre ‘sitzen’ (inzeptiv?) oder sīdere ‘sich setzen’ (Perfekt homonym):
Stichus 93 non sedeo isti, uos sedete, ego sedero in subsellio ‘Dort sitze
ich nicht, sitzt i h r da! Ich werde mich auf der Bank hinsetzen.’
13. seruāre ‘bewahren, retten’: a) Pseudolus 630 uinctam potīus sīc
seruāuerō ‘gefesselt werde ich es [das Geld] so besser bewahren’
b) Amphitruo 313f. Me. quid si ego illum tractim tangam, ut dormiat?
So. seruaueris ‘Was, wenn ich jenem eins überziehe, so dass er schläft?’
– ‘Du wirst [mich] retten‘
c) Rudens 1350 etsi tu fidem seruaueris ‘wenn du deinen Eid hältst’
b) Aktionsverben
14. augēre: Asinaria 280 erum in opsidione linquet, inimicum animos auxerit
‚er wird den Herrn in der Blockade verlassen, und den Mut der Feinde
stärken.’
15. bibere: a) Bacchides 49 eadem biberis, eadem dedero ‘dann wirst du
trinken, dann werde ich dir … geben.’ b) Bacchides 49 eadem dedero tibi
ubi biberis sauium ‘dann werde ich dir, wenn du getrunken hast, einen
Kuss geben.’
16. cēnāre ‘speisen’: a) Casina 780f. Si sapitis, uxor, uos tamen cenabitis,
cena ubi erit cocta; ego ruri cenāuero ‘wenn ihr klug seid, Frau, werdet
ihr dennoch speisen, wenn das Essen gekocht sein wird; ich werde auf
dem Lande speisen’
b) Mostellaria 1007 cras ... apud te cenauero ‚morgen werde ich bei dir
speisen’
17. coquere ‘kochen, braten’: vorzeitig Aulularia 367 comedent, sī quid
coxerint
18. ferre ‘tragen; (tragend) bringen’: Cistellaria 650 ibo, hanc tetulero intra
limen ‘ich gehe nun, diese werde ich über die Schwelle tragen.’
19. flēre ‘weinen’: Pseudolus 100 nisi tu illi drachumis fleueris argenteis
‘wenn d u ihr nicht silberne Drachmen weinst [und nicht etwa echte,
nützt es nichts]’
20. īre ‘gehen’: Stichus 484: sed – quoniam nihil processi sat ego hac, iuero
apertiore magi' uia ‘aber, da nun auf diese Weise i c h nicht genug
vorangekommen bin, werde ich auf einem offeneren Weg vorgehen.’
(aber telisch-terminativ Captivi 194: ad fratrem, quo ire dixeram, mox
145
iuero ‘zum Bruder, wohin ich gesagt hatte, dass ich gehen werde, werde
ich bald gehen’)
21. lauere/lauāre ‘baden’ nur vorzeitig: a) Stichus 569 post ubi lauero,
otiosus uos opperiar ‘sobald ich gebadet habe, werde ich euch in Muße
erwarten’
b) Poenulus 702 ubi tu laueris ‘sobald du gebadet hast’
22. legere ‘lesen, sammeln’: a) Amphitruo 326 alia forma | esse oportet quem
tu pugno legeris [spricht zu seinem Arm:] ‘Von einer anderen Form muss
sein, wen d u mit der Faust „(ge)sammel(t habe)n“ wirst’
b) Pseudolus 25f. has quidem pol credo nisi Sibulla legerit, interpretari
alium potesse neminem ‘glaube ich, wenn nicht die Sibylle sie lesen kann,
dass sie niemand sonst interpretieren kann’
23. lūdere ‘spielen’: Captivi 344 at nihil est ignotum ad illum mittere:
operam lūseris ‘Aber es bringt nichts, einen Unbekannten zu ihm zu
senden: Du wirst (dann) die Mühe verspiel(t hab)en.’ (terminative Spezialbedeutung)
24. mouēre ‘bewegen’: Mercator 311f. sei mouero me seu secari sensero,
Lysimache, auctor sum ut me amando – enices. ‘wenn ich mich bewege
oder spüre, dass ich geschnitten werde, Lysimachus, ermächtige ich dich,
dass du mich durch Liebe – zu Tode quälst.’
25. muttīre ‘mucksen, murmeln’: a) Miles gloriosus 565f. egone si post hunc
diem muttiuero, ..., dato excruciandum me ‘wenn ich nach diesem Tag
muckse, … dann übergib mich der Folter’
b) Bacchides 800 impinge pugnum, si muttiuerit ‘Hau die Faust rein,
wenn er muckst’
26. narrāre ‘erzählen’: a) Mostellaria 1039 eademque opera <ego> haec tibi
narrauero; hi sciunt qui hic adfuerunt ‘diese Sachen werde ich dir noch
erzählen; die wissen es, die hier dabei gewesen sind.’
b) Pseudolus 721 uobis post narrauero ‘euch werde ich es danach
erzählen.’
c) Stichus 144: probiores credo arbitrabunt, si probis narraueris ‘sie
werden [euch] für besser halten, wenn du es guten Leuten erzähl(t ha)st’.
27. plaudere ‘klatschen’: Casina 1017 qui non manibus clare quantum poterit
plauserit ‘wer nicht mit den Händen laut, so wie er kann, klatschen wird
…’
28. prandere ‘frühstücken’: vorzeitig Pseudolus 664 quin ubi prandero, dabo
operam somno ‘nachdem ich gefrühstückt habe, werde ich mich dem
Schlaf hingeben’
29. dēlectāre ‘(sich) erfreuen’: a) Mercator 547f. breue iam relicuom uitae
spatiumst: quin ego uoluptate, uino et amore delectauero ‘schon kurz ist
der restliche Zeitraum des Lebens, da werde i c h mich wahrlich an Lust,
Wein und Liebe erfreuen’
b) Pseudolus 573 tibicen uos interibi hic delectauerit ‘dieser Flötenspieler wird euch inzwischen erfreuen’
ipabo_66.249.78.36
Ionic and Attic: on temporal constructions
in
Thucydides, Xenophon and Plato
By ANTONIO LILLO, Murcia
During the fifth century BC there is a strong Ionic influence
on the Attic dialect, as is suggested by the Old Oligarch’s assertion ([X.] Ath. 2.8),
(1) ἔπειτα φωνὴν πᾶσαν ἀκούοντες ἐξελέξαντο τοῦτο μὲν ἐκ τῆς,
τοῦτο δὲ ἐκ τῆς· καὶ οἱ μὲν Ἕλληνες ἰδίᾳ μᾶλλον καὶ φωνῇ καὶ διαίτῃ
καὶ σχήματι χϱῶνται, Ἀθηναῖοι δὲ κεκϱαμένῃ ἐξ ἁπάντων τῶν
Ἑλλήνων καὶ βαϱβάϱων. «Further, through hearing every kind of
dialect, they have taken something from one, something else from
another; the Greeks individually tend to use their own dialect, way of
life and style of dress, but the Athenians use a mixture from all the
Greeks and non-Greeks».
As we can see, by the 5th century BC. the Old Oligarch is
already entertaining doubts about the “purity” of the Attic, when
this dialect begins to become a literary dialect because of the
range of features taken from other Greek dialects, especially
from Ionic. As a consequence, we can assume a certain degree
of adaptation of those Ionic features to the system of Attic.
These features are more easily detectable when they are of a
phonetic, morphological of lexical type, but this is more complicated when they are of a syntactic type, because they can
more easily be masked. Accordingly, our focus in this article
will be the use of certain marks of temporal construction, marks
made, on the one hand, by the representative par excellence of
Ionian, Herodotus, and, on the other, by that other great Attic
historian, Thucydides. We shall go on to make a comparison
__________

This work is part of the research project HUM2007–62333/FILO. It is a
revision and enlargement of a previous study, published in F. Cortes Gabaudan & J. V. Méndez Dosuna (eds.), DIC mihi, MUSA, Virum. Homenaje al
professor Antonio López Eire, 2010 Salamanca, p. 389–396, dedicated to the
memory of Prof. López Eire, to whom this article is also dedicated.
Glotta 89, 146–169, ISSN 0017–1298
Ionic and Attic: on temporal constructions in Thucydides, Xenophon and Plato
147
with the use of these same marks by Xenophon and Plato, so
that we can see to what extent these traits are integrated into
their respective language systems.1
We shall begin with the procedures for the expression of
simultaneity by means of subordinate sentences and the distribution of markers. In Herodotus these markers are ἕως, ἐν ᾧ / ὅσῳ,
μέχϱι and ἔστ’ἄν, and in Thucydides they are the same, with the
exception of ἔστ’ἄν, which does not appear. If we consider that
μέχϱι is not used as a temporal conjunction in older Attic
inscriptions2 (nor, of course, is the phrase ἐν ᾧ) and that ἕως is
the only marker in Attic oratory, it is clear that the presence of
these procedures in Thucydides requires a thorough analysis to
determine whether or not these are Ionismus and to what extent
they depend upon the text of Herodotus.
Let us, therefore, analyze the distribution of procedures used
by the Ionian historian to establish a comparison with the use of
these same procedures by Thucydides. Fuchs (1902: 66–88) and
Monteil (1963: 299–308, 316–25) have analyzed them in a
generic way and, apart from particular aspects of their own
methodologies, do not tackle the question of the distribution of
markers in those corpora. The conclusion to be drawn from their
works is thus that, in this particular case, these procedures are all
equivalent and, hence, interchangeable for the expression of
simultaneity. A different approach is adopted by Rijksbaron
__________
1
Rosenkranz (1930: 164–5) minimizes the Ionian influence in Thucydides and believes that the area covered by Attica, Euboea and Boeotia presents
doublets in their dialect forms. It is generally believed that the Ionic influence
in Thucydides is only noticeable in the conservation of-σσ-and-ϱσ-; cf. Meillet (1975: 239), Hoffman-Debrunner-Scherer (1969: 134, § 222), Hiersche
(1970: 212), Palmer (1980: 167). Adrados (1999: 143) considers Thucydides
to be the author of a prose transition. Concerning Xenophon, it is worth
stressing the enumeration of features of a phonetic and morphological type,
though mainly lexical, attributable to Ionic, but also to Doric and to poetic
uses, which appears in Gautier (1911), Hoffmann-Debrunner, Scherer (1969:
137–9, § 228–230) and Hiersche (1970: 216–21). Meillet (1975: 241) points
out that Plato comes to use poetic words, but it can be said that he reproduces
the conversational tone of the learned men of Athens; cf. also Adrados (1999:
138, 144 and 146).
2
It is used only μέχϱι as a preposition; cf. Meisterhans-Schwyzer (1900:
219, 251), Threatte (1996: 669–79).
ipabo_66.249.78.36
Antonio Lillo
148
(1976: 139–45), for whom the difference between ἕως and ἐν ᾧ
constructions in Herodotus, both indicating simultaneity, is that
the use of ἕως implies not only that the duration of both actions
is the same, but that the process described in the main clause has
its raison d’être in the subordinate, so that, from a semantic
viewpoint, ἕως + imperfect constructions have similarities with
the conditionals; in a different way, the marker ἐν ᾧ introduces
only constructions, which coincide at some point with the main
action. Both values are exemplified by the following texts:
(2) Οἱ δὲ στϱατιῶται ἕως μέν τι εἶχον ἐκ τῆς γῆς λαμβάνειν,
ποιηφαγέοντες διέζωον “And the soldiers, while they could get
anything from the earth, kept themselves alive by eating grass” (Hdt.
3.25.6)
(3) Ἐν ᾧ δὲ οὗτοι ταῦτα ἔπασχον, ὁ Δημοκήδης ἐς τὴν Κϱότωνα
ἀπικνέεται “While they were in this plight, Democedes made his way
3
to Croton” (Hdt. 3.136.2)
But these are not the only marks used by Herodotus to point
out the simultaneity of actions by means of sentence constructions, as he also uses μέχϱι and ἔστ’ἄν. As examples:
(4) Μέχϱι μὲν γὰϱ ὥϱων ἡμέας ὅπλα ἔχοντας, οἱ δὲ ἐνόμιζον
ὅμοιοί τε καὶ ἐξ ὁμοίων ἡμῖν εἶναι “But, as long as they see us armed,
they imagine that they are our equals and the sons of our equals” (Hdt.
4.3.4)
(5) Νῦν τε ἀπάγγελλε Μαϱδονίῳ ὡς Ἀθηναῖοι λέγουσι, ἔστ’ ἂν ὁ
ἥλιος τὴν αὐτὴν ὁδὸν ἴῃ τῇ πεϱ καὶ νῦν ἔϱχεται, μήκοτε ὁμολογήσειν
ἡμέας Ξέϱξῃ “Now carry this answer back to Mardonius from the
Athenians, that as long as the sun holds the course by which he now
goes, we will make no agreement with Xerxes” (Hdt. 8.143.2)
The form μέχϱι works as an adverb, preposition and conjunction with the meaning of “until”, “so long as”, like ἔστε.4
Consequently, the action of the sentence introduced by μέχϱι in
__________
3
4
Equivalent to ἐν ᾧ is the phrase ἐν ὅσῳ; cf. Hdt. 1.174.3.
Cf. P. Chantraine, DELG, s.v.; Monteil (1963: 316–318).
149
(4) covers the whole duration of the action in the main clause up
to the point at which it ends,5 so that the exact meaning of the
text would be: “But, right through to the end, when they see us
armed, they imagine that they are our equals and the sons of our
equals”. It is necessary to think, therefore, that the construction
introduced by μέχϱι is of a ἕως type, indicating that there are
two simultaneous actions, that of the main clause and that of the
subordinate, where the process described in the main clause has
its raison d’être in the subordinate, but also that this type of
construction involves as its specific mark the act of emphasizing
that both actions conclude at the same time. And, as occurs in
(4), the construction introduced by ἔστ’ἄν in (5) points to the
fact that the action of the subordinate clause covers the whole
duration of the main clause right up to the moment at which it
ends;6 nevertheless, the difference with regard to μέχϱι is that
the action expressed in the main one does not have its raison
d’être in that of the subordinate, but rather that both actions, that
of the main sentence and that of the subordinate, are simply
simultaneous. In contrast to these procedures, the sentences
introduced by ἐν ᾧ indicate the mere coincidence at some point
with the action of the main one, without any additional mark of
cause, condition, limit or consequence. Consequently, we can
draw a parallel between ἕως and μέχϱι in relation to ἐν ᾧ and
ἔστ’ἄν, based on the fact that the actions expressed by the subordinates introduced by ἕως and μέχϱι are not just simultaneous,
but the condition or raison d’être of the main clause, as opposed
to those introduced by ἐν ᾧ and ἔστ’ἄν, which are simply simultaneous. On the other hand, actions of the sentences introduced
by μέχϱι and ἔστ’ἄν would cover the whole duration of the
respective main sentences up to the moment at which they end,
as opposed to the subordinates introduced by ἕως and ἐν ᾧ,
where these sentences would not have this specific mark.
__________
5
6
Cf. also Hdt. 8.3.2.
Cf. also Hdt. 3.53.5, 4.201.2, 8.61.2, 8.142.4, 8.144.3.
ipabo_66.249.78.36
150
Antonio Lillo
In connection with this use of μέχϱι as a conjunction we have
the prepositional expression μέχϱι τῆς ἐκείνου ζοῆς, “for life”,
“while living”, “till the end of life”, which appears in (6):
(6) καὶ τὴν Βαβυλῶνά οἱ ἔδωκε ἀτελέα νέμεσθαι μέχϱι τῆς ἐκείνου
ζόης “and let him govern Babylon all his life with no tribute to pay”
(Hdt. 3.160.2)
This locution, corresponding to Attic ἕως ἂν ζῇ, also appears
as μέχϱι (οὗ) ἂν / κα ζῇ in Attic and in other areas. That this
prepositional turn μέχϱι τῆς ἐκείνου ζόης is a typical procedure
of the phraseology of East Ionic is also corroborated by its
appearance in inscriptions in this Ionic area and in the nearby
Doric one:
(7) τίθεσ|θαι ἐν τῆι ὀϱχήστϱαι τϱίποδα καὶ θυμίατϱον, τοὺς δὲ
ἱεϱεῖς… καὶ τὸν ἱεϱοκήϱυκα καὶ τοὺς… παιδονόμους καὶ μέχϱι μὲν
ζωιῆς Εὔδημον “that the priests place in the orchestra a tripod and a
censer… and the herald of the sacred ceremonies, the education
officials… and lifetime Eudemus” (SIG 577.34, decree in honor of
7
Eudemus Talionis, Milet, 200 / 199 B.C.)
(8) ἐὰν8 δέ τινες τῶν ἐσομέν[ων κληϱο]|νόμων μὴ ὁμολογήσωσιν
ὀφ[είλειν τὸν φό]|ϱον μέχϱι τῆς ἑαυτῶν ζωιῆς [καθὼς γέγϱα]|πται
“And, if some of the future heirs didn’t agree, they will pay a tribute
during their life, as it is written” (IMylasa 421.9, Mylasa, date
unknown)
(9) [καὶ ἀπα]ϱ<ε>νοχλή-τ[ους ἀπό τ‛ ἐμοῦ]|[καὶ πα]ντὸς
κληϱονό[μου — — — ἐπὶ]|[παϱαμ]ονῇ μέχϱι τῆς ζω[ῆς μου — —
]|[— — — —] εὐάϱεστοι τῇ μ[ητϱί μου]|[— — — —]τάδι (CIRB 73.
9
10
11, Pantikapaion , ca. 100–150 A.C.?)
__________
7
The expression is repeated in the same inscription, in l. 73.
This form is properly Attic and of koiné; nevertheless, the locution is
Ionic.
9
Greek settlement of the Crimean peninsula founded by settlers from
Milet, which corresponds to current Kerch, Ukraine.
10
The fragmentary state of the text does not allow us to give a translation
here.
8
151
(10) πα]|ϱαμίνασα Διοκλῖ μέχϱι ζωᾶς εὐαϱέστως·“after having
11
remained next to Diocles during his life in a pleasant way”
(Tit.Calymn. 154.3, Calymna, ca. 14–54 A.C.)
We find, therefore, a typical East Ionic phraseology, which
extends over a very defined area, with testimonies in Milet and
Mylasa and the nearby Doric island of Calymna.12
Let us now consider the situation of Attic. Concerning the
expression of the idea “while living”, in contrast to Herodotus
and part of the Ionic and Doric of Calymna, where the μέχϱι
locution is used, Lysias (5.1, 17.3, 19.46), Xenophon (HG
3.1.19; An. 3.1.43; Cyr. 8.3.38, 5.26), Isaeus (6.9), Aristophanes
(Eq. 395) and Demosthenes (20.64) always use the phrase ἕως
ἔζη, ἕως ἂν ζῇ, “while living”, “while alive”, and never the
nominal locution with μέχϱι, introducing either a sentence with
μέχϱι or a sentence with ἂν ζῇ or ἔζη. It should be noted also
that in Lysias, Isaeus, Aristophanes and Demosthenes, authors
whose “purity” in the use of Attic is beyond doubt, μέχϱι
appears only as a preposition, as in the more ancient Attic
inscriptions,13 while its use in Xenophon as a preposition
accounts for four fifths of the evidence, with only one fifth as a
conjunction. On the other hand, Plato, more interested in an
artistic prose than in the “purity” of Attic, presents the sentence
ἕως ἂν ζῇ three times (Smp. 192.e, Lg. 755.a, Phd. 84.b) and
μέχϱι ἂν ζῇ twice (Prt. 325.c, Epin. 973.c).
This locution does not appear in any form in Thucydides, but,
as we have indicated before, we do find the same procedures in
Herodotus, ἕως, μέχϱι and ἐν ᾧ / ὁσῳ, except ἔστ’ἄν, to indicate
simultaneity by sentence construction. Since neither μέχϱι nor ἐν
ᾧ / ὁσῳ nor ἔστ’ἄν introducing sentence constructions appear in
__________
11
This pattern is repeated in another eight inscriptions of the same type in
Calymna, circa 14–54 A.C., Tit.Calymn. 156.3, 157.5, 161.3, 163, 7, 164a.6,
166a.4, 166b.8, 178.4. This is a set of inscriptions in which slave owners give
them freedom after having remained next to them throughout their lives.
12
It is important to bear in mind that in Calymna the locution ἐς ὅ κα also
appears, in Tit.Calymn. 79A.74, beginning of the 3rd century B.C., to indicate limit, “till”, as in Herodotus.
13
Cf. note 2.
ipabo_66.249.78.36
152
Antonio Lillo
Attic oratory, it is reasonable to think that the appearance of
these procedures in Thucydides, Xenophon and Plato is the
result of the Ionic influence on Attic. With respect to these
markers, however, the dependence of these authors on Ionic, and
especially on Herodotus, has not been analyzed in any detail,14
so that it is interesting to see how these Attic authors integrate
these markers into their language systems. First, then, let us
analyze the texts in which Thucydides uses μέχϱι οὗ and ἕως:
(11) Ἀθηναῖοι δὲ μέχϱι μὲν οὗ πεϱὶ Ἐλευσῖνα καὶ τὸ Θϱιάσιον
πεδίον ὁ στϱατὸς ἦν, καί τινα ἐλπίδα εἶχον ἐς τὸ ἐγγυτέϱω αὐτοὺς μὴ
πϱοϊέναι, “The Athenians, so long as the Lacedaemonians were in the
neighbourhood of Eleusis and the plain of Thria, entertained a hope
that they would come no further” (Th. 2.21.1)
(12) μέχϱι μὲν οὖν οἱ τοξόται εἶχόν τε τὰ βέλη αὐτοῖς καὶ οἷοί τε
ἦσαν χϱῆσθαι, οἱ δὲ ἀντεῖχον…·“So, while their archers had arrows
and were able to use them, the Athenians maintained their ground…”
(Th. 3.98.1)
In common to texts (11) and (12) is the fact that the actions of
the main sentence and of the subordinate one are simultaneous,
of identical duration and, as in Herodotus, the use of μέχϱι
emphasizes the fact that both actions finish at the same time,
‘throughout the time until the end’. Moreover, the action of the
main sentence has its raison d’être in that of the subordinate: in
(11) the Athenians had a similar precedent for action by the
Spartans, as is explained in the following text, to which the
action of Pleistoanax, the king of the Lacedaemonians, is related,
and in (12) the fact that the Athenians faced the Aetolians is a
consequence of the skill of their archers.15
For ἕως + imperfect construction we have the following
examples:
__________
14
It has been said that the use of such markers by Thucydides does not
necessarily depend on Ionic; cf. Hiersche (1970: 215), who follows Rosenkranz (1930: 164‒5).
15
Cf. also Th. 3.10.4.
153
(13) οἱ Ἀθηναῖοι … ἔθεον δϱόμῳ ἐκ τῆς ἐνέδϱας, βουλόμενοι
φθάσαι πϱὶν ξυγκλῃσθῆναι πάλιν τὰς πύλας καὶ ἕως ἔτι ἡ ἅμαξα ἐν
αὐταῖς ἦν, κώλυμα οὖσα πϱοσθεῖναι “The Athenians… ran at the top
of their speed from the ambush in order to reach the gates before they
were shut again, and while the cart was still there to prevent their
being closed” (Th. 4.67.4)
(14) στϱατόπεδον γὰϱ δὴ τοῦτο κάλλιστον Ἑλληνικὸν τῶν μέχϱι
τοῦδε ξυνῆλθεν·ὤφθη δὲ μάλιστα ἕως ἔτι ἦν ἁθϱόον ἐν Νεμέᾳ, …
“Indeed this was by far the finest Hellenic army ever yet brought
together; and it was most to be seen while it was still united at
Nemea…” (Th. 5.60.3)
(15) ἐστϱάτευσαν ἅπαντες οἱ ξύμμαχοι οὗτοι εὐθὺς ἐπὶ Ἐπίδαυϱον,
ἕως οἱ Λακεδαιμόνιοι Κάϱνεια ἦγον, “All these allies marched at once
against Epidaurus while the Lacedaemonians were keeping the
Carnea” (Th. 5.75.5)
In (13) the ἕως construction indicates total simultaneity between the actions of the main clause and the subordinate clause
(“they ran while the doors were open”), at the same time as the
foundation of the action of the main clause is in that of the
subordinate (“if the doors had not been open, they would not
have run”), as in (2), a ἕως construction of Herodotus. In (14)
there is partial simultaneity, since the use of the aorist ὤφθη
points to the actual fact of having seen the army when it was
concentrated in Nemea, but, as in (13), the action of the main
construction has its foundation in that of the subordinate (“if the
soldiers had not been concentrated in Nemea, it would not have
been possible to verify that the Greek army was the best
concentrated so far”); consequently, the use of ἕως here is also
similar to that of Herodotus (2). And the same can be said of
(15), since, as well as there being partial simultaneity between
the actions of the main and subordinate sentences, the temporal
coincidence between them is not merely fortuitous: the allies
made an expedition against Epidaurus because the Spartans
were celebrating the Carnean feast; so the ἕως construction is
ipabo_66.249.78.36
154
Antonio Lillo
also the cause or the condition of the action expressed in the
main clause,16 as in (2).17
Let us now analyze the use of the phrase ἐν ᾧ / ὅσῳ in
Thucydides. The difficulty, however, in determining whether or
not the appearance of this phrase is due to the influence of
Herodotus is that, as it is an analyzable construction and both
constituent elements are, therefore, significant on their own, the
relative may have other values besides the strictly temporal
“while” as a phrase; alternatively, it may have a more or less
implicit or explicit antecedent of the relative with a temporal
value. For that reason, there are few texts of Thucydides in
which the phrase ἐν ᾧ / ὅσῳ works clearly with the temporal
value “while” and not as a relative with preposition. Hence, for
example,18
(16) Κεϱκυϱαῖοι δὲ… τὰς ναῦς πεϱιπλεῦσαι κελεύσαντες ἃς
ἐπλήϱωσαν ἐς τὸν Ὑλλαϊκὸν λιμένα, ἐν ὅσῳ πεϱιεκομίζοντο, τῶν
ἐχθϱῶν εἴ τινα λάβοιεν, ἀπέκτεινον·“The Corcyraeans, … ordered the
fleet which they had manned to sail round into the Hyllaic harbour,
while it was so doing, slew such of their enemies as they laid hands
on” (Th. 3.81.2)
(17) Δεϱκυλίδας τε ἀνὴϱ Σπαϱτιάτης στϱατιὰν ἔχων οὐ πολλὴν
παϱεπέμφθη πεζῇ ἐφ’ Ἑλλησπόντου Ἄβυδον ἀποστήσων…, καὶ οἱ
Χῖοι, ἐν ὅσῳ αὐτοῖς ὁ Ἀστύοχος ἠπόϱει ὅπως βοηθήσοι, ναυμαχῆσαι
πιεζόμενοι τῇ πολιοϱκίᾳ ἠναγκάσθησαν. “Dercyllidas, a Spartan, was
sent with a small force by land to the Hellespont to effect the revolt of
Abydos,… and the Chians, while Astyochus was at a loss how to help
__________
16
Rijksbaron (1976: 144) thinks that the mere coincidence between both
actions seems only to be expressed in this passage, as in Herodotus (3), with
the temporal sentence introduced by ἐν ᾧ. However, we do not agree, since
above, in Th. 5.75.2, the Lacedaemonians are said to have gone to celebrate
the feast of the Carnea, a fact criticized by the allies; as Epidaurus was an ally
of the Spartans, the celebration of the feast by the Lacedaemonians undermined the capacity of the Epidaurians.
17
What has been said of (13) (14) and (15) is valid for Th. 8.78, 7.63.4
and 7.71.4.
18
Classen-Steup (1892: 161 and 1922: 144) proposes for the locution ἐν
ὅσῳ of 3.81.2, 8.61.1 and 8.87.4 the value “während”, equivalent to ἐν ᾧ in
Herodotus.
155
them, were compelled to fight at sea by the pressure of the siege” (Th.
8.61.1)
In (16) the process described in the main clause, Κεϱκυϱαῖοι
δὲ ... τῶν ἐχθϱῶν εἴ τινα λάβοιεν, ἀπέκτεινον, does not depend
on the subordinate, ἐν ὅσῳ πεϱιεκομίζοντο, as occurs in Herodotus (3), where the subordination marker is ἐν ᾧ. Sailing to the
Hyllaic harbour is not the necessary cause of killing the enemies
found in their path; rather, these are merely two actions occurring simultaneously, without being a necessary consequence
one of the other. And the same can be said of (17), since it is not
clearly stated in the text that the siege of the inhabitants of Chios
and being forced into a naval battle are motivated by the
Astyochus’ difficulties in reaching Chios from Rhodes. We have
therefore a situation similar to that presented by Herodotus,
though with a different phraseology: the sentences introduced by
ἐν ᾧ / ὅσῳ in Herodotus are usually constructed with a present
stem, as in (16) and (17), but these are usually at the beginning
of the construction, as in (3). Consequently, it is reasonable to
think that Thucydides has taken the Ionic prepositional phrase
and adapted it to the Attic system, where, unlike Ionian, the only
procedure for indicating the simultaneity of actions, judging by
the lack of the markers μέχϱι and ἐν ᾧ to introduce temporal
sentences in 4th-century BC oratory, is ἕως.
From the above we may assume that Thucydides adapts to
Attic the Ionic procedures for indicating simultaneity in sentence constructions with μέχϱι and ἐν ᾧ / ὁσῳ, because a) he
uses μέχϱι to specify simultaneity, with particular emphasis on
the fact that both actions, that of the main clause and that of the
subordinate, end at the same time, and also to indicate that the
action of the main clause has its raison d’être in the subordinate,
as in Herodotus; b) he also uses ἐν ᾧ / ὅσῳ in the same way as
in Herodotus, that is, to point out the mere temporal coincidence
of the actions of the main clause and the subordinate.
For his part, Xenophon also points out simultaneity by ἕως,
μέχϱι and the phrase ἐν ᾧ, besides ἔστε, which usually appears
in Herodotus followed by the particle ἄν. The phrase ἐν ᾧ is not
ipabo_66.249.78.36
Antonio Lillo
156
uncommon and functions as a mere doublet of ἕως, just like
ἔστε. Let us quote the following examples:
(18) ἐν ᾧ δὲ ταῦτα ἦν ἐπὶ λείαν ἐξῇσαν οἱ Ἕλληνες “While these
things were going on, the Greeks were making forays in quest of
booty” (X. An. 5.1.17)
(19) ὅμως δὲ ὁ Ἰφικϱάτης, ἕως μὲν ἐν τῷ ἰσοπέδῳ τὸ στϱάτευμα
τοῦ Ἀναξιβίου ἦν, οὐκ ἐξανίστατο “Nevertheless, Iphicrates did not
rise from ambush so long as the army of Anaxibius was on the level
ground” (X. HG. 4.8.37)
(20) ὁ δὲ Κῦϱος τὸ μὲν πϱῶτον ἐθαύμασεν· ὡς δ’ ἔγνω τὸ πϱᾶγμα,
19
ἔστε μὲν πάντες ἐναντίοι ἤλαυνον, ἐναντίος καὶ αὐτὸς ἐν τάξει ἦγε
τὴν στϱατιάν “At first Cyrus was surprised; but when he comprehended the situation, he continued, while the enemy were all riding
against him, to lead his army in battle order against them” (X. Cyr.
5.4.7)
In (18), a text which has an identical construction to that
appearing in the Herodotean text (2), the sentence introduced by
ἐν ᾧ refers to the capture of vessels and goods by a certain Polycrates, while a number of things are happening: specifically, the
escape of Dexippus and his subsequent death, the purchase of a
ship of thirty oars and the requisitioning of goods; because of
this, it is clear that the subordinate clause refers to an action
which is simultaneous with the main, with no further link than
mere temporal coincidence. And the same can be said in (19),
where the action of the sentence introduced by ἕως, the fact that
Anaxibius’ army is on the plain, is coincident with the fact that
Iphicrates did not rise and the context offers no link between the
two events beyond their temporal coincidence; therefore, ἕως is
interchangeable with ἐν ᾧ. In (20), as in (18) and (19), ἔστε is
also interchangeable with ἕως and ἐν ᾧ, by introducing a temporal sentence whose action is simply simultaneous with that of
the main one, without limitative or conditional connotations, as
in Herodotus.
__________
19
Other manuscripts present the form ἕως.
157
But Xenophon uses μέχϱι to introduce sentences with a verb
in the present stem expressing simultaneity with the action of
the main sentence, as in Thucydides and Herodotus, with the
specific mark of the coincidence at the end of both actions,20
although the evidence with verbs in the present stem are very
scarce. An example is the following passage:
(21) καὶ ὁ Φαϱνάβαζος παϱεβοήθει, καὶ ἐπεισβαίνων τῷ ἵππῳ εἰς
τὴν θάλατταν μέχϱι δυνατὸν ἦν ἐμάχετο, καὶ τοῖς ἄλλοις τοῖς αὑτοῦ
ἱππεῦσι καὶ πεζοῖς παϱεκελεύετο “And Pharnabazus came along the
shore to their aid, and riding his horse into the sea as far as possible
(while / until / throughout the time until the end when it was possible),
fought and cheered on his followers, cavalry and infantry” (X. HG
1.1.6)
As with μέχϱι, Xenophon sometimes uses ἔστ’ἄν in the same
way as Herodotus, in contexts where there are two simultaneous
actions with the specific mark of the end of both, as in (22):
(22) οὐδεὶς γοῦν οὔτε ἀπολείπει αὐτὸν [τὸν ἵππον] οὔτε
ἀπαγοϱεύει θεώμενος, ἔστ’ ἄν πεϱ ἐπιδεικνύηται τὴν λαμπϱότητα “At
all events no one leaves him or is tired of gazing at him so long as
(throughout the time until the end, at which) he shows off his
21
brilliance” (X. Eq. 11.9)
Xenophon, in short, represents a highly advanced state of the
use of these markers in relation to Herodotus and Thucydides,
since ἕως, ἐν ᾧ and ἔστε are fully interchangeable and it seems
that only μέχϱι and ἔστ’ἄν maintain the value found in Herodotus and Thucydides: to emphasize the fact that two simultaneous
actions, that of the main clause and that of the subordinate, end
at the same time.
The situation in Plato is not exactly the same as that in
Thucydides and Xenophon. He uses neither ἔστε / ἔστ’ἄν nor a
phrase ἐν ᾧ with temporal value and only the marker μέχϱι,
together with ἕως, so that his texts have a “more Attic” air, as
__________
20
21
The same value can be attributed to ἄχϱι οὗ in X. HG 6.4.37.
Cf. also X. Oec. 1.23, 7.33; Lac. 5.3.
ipabo_66.249.78.36
158
Antonio Lillo
opposed to Xenophon, who uses a greater diversity of procedures. We have previously pointed out that on three occasions
we find the expression ἕως ἂν ζῇ and, on two, μέχϱι ἂν ζῇ to
indicate “while alive”, which gives a good idea of the commutability between the author’s use of μέχϱι and ἕως and also of
the extent to which both procedures have been integrated into
the Attic system:
(23) ἐκ παίδων σμικϱῶν ἀϱξάμενοι, μέχϱι οὗπεϱ ἂν ζῶσι, καὶ
διδάσκουσι καὶ νουθετοῦσιν. “They teach and admonish them from
earliest childhood so long as they live (till the last day of their lives)”
22
(Pl. Prot. 325.c)
(24) εἰ γὰϱ τούτου ἐπιθυμεῖτε, θέλω ὑμᾶς συντῆξαι καὶ
συμφυσῆσαι εἰς τὸ αὐτό, ὥστε δύ ὄντας ἕνα γεγονέναι καὶ ἕως τ’ ἂν
ζῆτε, ὡς ἕνα ὄντα, κοινῇ ἀμφοτέϱους ζῆν, “If that is your craving, I
am ready to fuse and weld you together in a single piece, that from
being two you may be made one; that so long as you live, the pair of
you, being as one, may share a single life” (Pl. Smp. 192.e)
Let us now examine the use of these temporal markers of
subordination with verbs in the aorist stem, a syntactic
procedure used to mark the limit of the action of the main
clause, “until”. To do this we will focus on the same authors,
namely, Herodotus, Thucydides, Xenophon and Plato, since,
with the exception of ἕως, these markers are, as we stated above,
not used in oratory.
__________
22
It is clear that the appearance of μέχϱι here, instead of ἕως, is conditioned by the form ἀϱξάμενοι from expressions of the type ἐξ ἀϱχῆς μέχϱι
τέλους (Pl. Lg. 653.c) in combination with sequences of the type τέλος /
τελευτὴ τοῦ βίου. But equally clear is the equivalence of μέχϱι and ἕως from
testimonies as αὗται δὲ <φλέβες>, μέχϱις οὗ ἂν ζώῃ ὁ ἄνθϱωπος, ἀνεῴγασι
καὶ δέχονταί τε καὶ ἀφιᾶσι νέον ὑγϱόν “these <veins>, while the man is alive,
are open, receive and release a new liquid” (Hp. Genit. 4.39.19) and οὔ φημι
εἶναι δυνατὸν ἀνθϱώποις μακαϱίοις τε καὶ εὐδαίμοσιν γενέσθαι πλὴν ὀλίγων
– μέχϱιπεϱ ἂν ζῶμεν τοῦτο διοϱίζομαι “I say it is impossible for men to be
blessed and happy, except a few; that is, so long as we are living: I limit it to
that” (Pl. Epin. 973.c, although Plato’s authorship of this work is a matter of
discussion), where the use of μέχϱι is not conditioned by ἄϱχω type forms.
159
As with the expression of simultaneity, Herodotus uses ἕως,
μέχϱι, ἔστ’ἄν and ἐς ὅ, instead of ἐν ᾧ. Let us analyze the
evidence for the use of each of these markers.
(25) Κατὰ Σάνην πόλιν σχοινοτενὲς ποιησάμενοι, ἐπείτε ἐγίνετο
βαθέα ἡ διῶϱυξ, οἱ μὲν κατώτατα ἑστεῶτες ὤϱυσσον, ἕτεϱοι δὲ
παϱεδίδοσαν τὸν αἰεὶ ἐξοϱυσσόμενον χοῦν ἄλλοισι κατύπεϱθε
ἑστεῶσι ἐπὶ βάθϱων, οἱ δ’ αὖ ἐκδεκόμενοι ἑτέϱοισι, ἕως ἀπίκοντο ἐς
τοὺς ἀνωτάτω·οὗτοι δὲ ἐξεφόϱεόν τε καὶ ἐξέβαλλον. “<The barbarians> made a straight line near the town of Sane; when the channel
had been dug to some depth, some men stood at the bottom of it and
dug, others took the dirt as it was dug out and delivered it to yet others
that stood higher on stages, and they again to others as they received
it, until they came to those that were highest; these carried it out and
threw it away” (Hdt. 7.23.1)
(26) Μετὰ δὲ <Ξέϱξης> ἐποίεε ταῦτα, καὶ διεξελαύνων ἐπὶ
ἅϱματος παϱὰ ἔθνος ἓν ἕκαστον ἐπυνθάνετο… ἕως ἐξ ἐσχάτων ἐς
ἔσχατα ἀπίκετο καὶ τῆς ἵππου καὶ τοῦ πεζοῦ. “Then <Xerxes> did
this; as he rode in a chariot past the men of each nation, he questioned
them… until he had gone from one end to the other of the cavalry and
infantry” (Hdt. 7.100.1)
(27) Οὗτος ὁ τάφος ἦν ἀκίνητος μέχϱις οὗ ἐς Δαϱεῖον πεϱιῆλθε ἡ
βασιληίη. “This tomb remained untouched until the kingship fell to
Darius” (Hdt. 1.187.3)
(28) Καὶ πεϱὶ Πέϱσας μὲν ἦν ταῦτα τὸν πάντα μεταξὺ χϱόνον
γινόμενα, μέχϱι οὗ Ξέϱξης αὐτός σφεα ἀπικόμενος ἔπαυσε. “This was
the plight of the Persians for all the time until the coming of Xerxes
himself ended it” (Hdt. 8.100.1)
(29) Οἱ μέν νυν ἄλλοι παῖδες τὰ ἐπιτασσόμενα ἐπετέλεον, οὗτος δὲ
ἀνηκούστεέ τε καὶ λόγον εἶχε οὐδένα, ἐς ὃ ἔλαβε τὴν δίκην. “The
other boys then did as assigned: but this one was disobedient and
cared nothing for me, until he received his punishment” (Hdt. 1.115.3)
(30) “Ὦναξ, χϱῆσον ἡμῖν ἄμεινόν τι πεϱὶ τῆς πατϱίδος, … ἢ οὔ τοι
ἄπιμεν ἐκ τοῦ ἀδύτου, ἀλλ‛ αὐτοῦ τῇδε μενέομεν ἔστ’ ἂν καὶ
23
τελευτήσωμεν”. “Lord , give us some better answer concerning our
__________
23
The god is Apollo.
ipabo_66.249.78.36
160
Antonio Lillo
country… or otherwise we will not depart from your temple, but
remain here until we die” (Hdt. 7.141.2)
In the texts with the marker ἕως, (25) and (26), we find that
the action indicated in the subordinate clause, besides being
limited to the action of the main clause, is a consequence of
what is indicated in that same main clause and is a part of the
same situation expressed in the whole construction: specifically,
in (25) the excavators passed the earth from the deepest part of
the trench to those who were above them, until reaching those
who were at the highest level; and in (26), in the context of
inspecting the troops from one end to the other, Xerxes asked
each member of the squadrons till reaching the end. Ἕως constructions do not indicate, therefore, an action representing the
end of a situation and the transition to another different from
that indicated in the main clause, but rather the action of the
subordinate remains part of the same situation.
In a different way, the marker μέχϱι οὗ in texts (27) and (28)
is used when the action of the subordinate is the end-point of the
action of the main; not an arbitrary end-point, but rather a
necessary step towards a new situation resulting from the previous one: specifically, in (27) the tomb remained closed until a
king ordered it to be reopened, and in (28) the state of desolation
of the Persian army from the outcome of a naval battle ended
with the arrival of Xerxes. The real Ionic procedure is, therefore,
to mark the limit not only of an action, but also of a situation,
and the step towards a new situation.24
Unlike the two previous markers, ἐς ὅ introduces a subordinate whose action is an arbitrary end-point to the action of the
main sentence, not a logical or necessary consequence, as in the
previous texts: in (29) the fact that he does not pay attention
makes the boy receive a punishment, though the punishment is
not the necessary consequence of his not paying attention.25
__________
24
Cf. also Hdt. 5.28 and 7.60.3.
There are many testimonies of this prepositional phrase, so the only
evidence of the same type cited by way of example is Hdt. 1.130.3, 2.103.1,
4.122.3, etc.
25
161
Finally, the marker ἔστ’ἄν introduces a subordinate whose
action is also an arbitrary end-point to the action of the main
sentence, as indicated in (30), where the act of dying is not the
necessary consequence of the act of remaining in the sanctuary.26 This marker ἐστ’ἄν is interchangeable with ἐς ὃ ἄν, as
can be seen if (30) is compared with (31):
(31) Τοῦ γὰϱ δὴ Λυδῶν δήμου αἱ θυγατέϱες ποϱνεύονται πᾶσαι,
συλλέγουσαι σφίσι φεϱνάς, ἐς ὃ ἂν συνοικήσωσι τοῦτο ποιεῦσαι·
ἐκδιδοῦσι δὲ αὐταὶ ἑωυτάς. “All the daughters of the people of Lydia
ply the trade of prostitutes, to collect dowries, until they can get themselves husbands; and they themselves offer themselves in marriage”
(Hdt. 1.93.4)
In (31) the act of marrying is the arbitrary limit to that of
prostitution, not a necessary reason,27 as in (30), where the act of
dying is an arbitrary limit to that of remaining in the temple. The
commutability of procedures is, therefore, evident.
Consequently, as was the case with these same markers when
introducing subordinate clauses with the verb in the present
stem, a parallel can be drawn between ἕως and μέχϱι in relation
to ἐς ὅ and ἔστ’ἄν, because the actions expressed by the subordinates introduced by ἕως and μέχϱι are the consequence of
what has been indicated in the respective main sentences, as
opposed to those introduced by ἐς ὅ and ἔστ’ἄν, where the
actions of the sentences introduced by these markers are an
arbitrary end-point with regard to the actions of the main
clauses. And in a parallel way to what happens with these
markers when introducing sentences with a verb in the present
stem, the use of ἕως here also differs from that of μέχϱι, but now
the constructions introduced by ἕως indicate a limit to the action
of the main clause, not the end-point of the situation in which
both are framed, though both continue to be part of that
situation; on the other hand, μέχϱι introduces a sentence which,
besides being a limit to the action of the main sentence, presents
__________
26
27
Cf. also Hdt. 4.12.3, 7.158.4 and 8.4.2.
Cf. also Hdt. 2.115.6, 3.82.4, 4.30.2, 4.196.2, 4.43.3 and 8.108.4.
ipabo_66.249.78.36
162
Antonio Lillo
an action which is a step towards a different situation from that
of the main clause.
Let us see now how these procedures work in Attic.
With regard to Thucydides, the situation is not the same as in
Herodotus, because the Attic historian uses only two markers to
make sentences with the verb in the aorist stem: ἕως and μέχϱι.
He does not use ἔστε or ἔστ’ἄν, as was the case with constructions expressing simultaneity, but nor does he use a prepositional locution of the type ἐς ὅ or εἰς ὅ. Let us see some passages
with both constructions.
(32) καὶ πϱοσβαλοῦσα ἡ πᾶσα παϱασκευὴ πϱός τε ἄκϱαν Ἰαπυγίαν
καὶ πϱὸς Τάϱαντα καὶ ὡς ἕκαστοι ηὐπόϱησαν, παϱεκομίζοντο τὴν
Ἰταλίαν…, ἕως ἀφίκοντο ἐς Ῥήγιον τῆς Ἰταλίας ἀκϱωτήϱιον. καὶ
ἐνταῦθα ἤδη ἡθϱοίζοντο, καὶ ἔξω τῆς πόλεως… στϱατόπεδόν τε
κατεσκευάσαντο ἐν τῷ τῆς Ἀϱτέμιδος ἱεϱῷ, “The whole force arriving
at the Iapygian promontory and Tarentum with more or less good
fortune, coasted along Italy… until they arrived at Rhegium, the
extreme point of Italy. Here at length they reunited, and outside the
city… encamped at the temple of Artemis” (Th. 6.44.2)
(33) οἱ δὲ Πελοποννήσιοι καὶ οἱ ξύμμαχοι τϱία τέλη ποιήσαντες
σφῶν αὐτῶν ἐχώϱουν πϱὸς τὴν τῶν Στϱατίων πόλιν, ὅπως… ἔϱγῳ
πειϱῷντο τοῦ τείχους. καὶ μέσον μὲν ἔχοντες πϱοσῇσαν Χάονες καὶ οἱ
ἄλλοι βάϱβαϱοι... καὶ οἱ μὲν Ἕλληνες τεταγμένοι τε πϱοσῇσαν καὶ
διὰ φυλακῆς ἔχοντες, ἕως ἐστϱατοπεδεύσαντο ἐν ἐπιτηδείῳ· “The
Peloponnesians and their allies advanced in three divisions towards
Stratus, with the intention of… attempting the wall by force. They
advanced occupyng the Chaonians and the rest of the barbarians the
center … and the Hellenes advanced in good order, keeping a lookout
till they encamped in a suitable place” (Th. 2.81.4)
(34) Γέγϱαφε δὲ καὶ ταῦτα ὁ αὐτὸς Θουκυδίδης Ἀθηναῖος ἑξῆς, ὡς
ἕκαστα ἐγένετο, κατὰ θέϱη καὶ χειμῶνας, μέχϱι οὗ τήν τε ἀϱχὴν
κατέπαυσαν τῶν Ἀθηναίων Λακεδαιμόνιοι καὶ οἱ ξύμμαχοι, καὶ τὰ
μακϱὰ τείχη καὶ τὸν Πειϱαιᾶ κατέλαβον. “The same Thucydides of
Athens has also written the history of this period, following the order
of events by summers and winters, until the Lacedaemonians and their
allies put an end to the Athenian empire and took the Long Walls and
Piraeus” (Th. 5.26.1)
163
(35) καὶ ἅμα τῇ ἕῳ διεσπασμένου τοῦ ναυτικοῦ… ἐπανάγονται
κατὰ τάχος ὁ Χαϱμῖνος καὶ οἱ Ἀθηναῖοι ἐλάσσοσιν ἢ ταῖς εἴκοσι
ναυσί… καὶ πϱοσπεσόντες εὐθὺς κατέδυσάν τε τϱεῖς καὶ κατετϱαυμάτισαν ἄλλας, καὶ ἐν τῷ ἔϱγῳ ἐπεκϱάτουν, μέχϱι οὗ ἐπεφάνησαν
αὐτοῖς παϱὰ δόξαν αἱ πλείους τῶν νεῶν “In the morning the fleet was
dispersed … Charminus and the Athenians put out to sea with part of
their twenty ships… and at once attacking them sank three ships and
disabled others, and had the advantage in the action until the main
body of the fleet to their surprise appeared there” (Th. 8.42.3)
We see in (32) that what is indicated by the ἕως sentence, the
act of arriving in Rhegium, is not a mere end-point to the action
of coasting along Italy, but that the action of the main clause
(coasting along Italy) has its raison d’être, its purpose, in
reaching Rhegium and building a camp there. And something
similar can be said of (33): the act of camping in an appropriate
place, indicated by the ἕως sentence, is the reason for the statement in the main clause, the advance of the Greeks in formation.
But, unlike (32) and (33), what is indicated in (34) with the
μέχϱι sentence is an arbitrary limit to the action of the main
sentence, so that what is stated in the main clause does not have
its raison d’être in what is stated in the subordinate; and the
same can be said of (35), since what is indicated in the μέχϱι
sentence, the appearance of the main fleet, is an arbitrary limit to
what is expressed in the main clause, the act of dominating the
battle, without one action being motivated by the other, as in the
constructions introduced by ἕως. In sum, we can say that the
distribution of markers introducing limitative temporal sentences in Thucydides, ἕως and μέχϱι, is not the same as that of
Herodotus, and yet is somehow reminiscent of it, since the use
of ἕως is the same as in Herodotus (25) and (26), unlike μέχϱι,
which in Thucydides takes the functions of ἐς ὅ and ἔστ’ἄν.
Thucydides thus maintains the differences for the expression of
a time-limit presented by Herodotus, but with a new distribution
of markers: ἕως to indicate that the action expressed in the main
sentence has its cause or raison d’être in what is indicated in the
ipabo_66.249.78.36
164
Antonio Lillo
limitative subordinate,28 unlike μέχϱι, which is used to indicate
that the action of the subordinate clause is an arbitrary limit to
that of the main clause.29
Let us now examine the case of Xenophon. An example is the
following pair:
(36) οὐδενὸς δὲ οὐδὲν ἐπαιτιωμένου, δεομένων ἔμειναν ἕως
ἀφίκοντο οἱ ἀντ’ ἐκείνων στϱατηγοί, “But when no one brought any
charge against them, at the request of the troops they remained until
their successors arrived” (X. HG. 1.1.29)
(37) ὁ δὲ <Λύσανδϱος> ἀφικόμενος εἰς Ῥόδον καὶ ναῦς ἐκεῖθεν
λαβών, εἰς Κῶ καὶ Μίλητον ἔπλευσεν, ἐκεῖθεν δ’ εἰς Ἔφεσον, καὶ
ἐκεῖ ἔμεινε ναῦς ἔχων ἑβδομήκοντα μέχϱι οὗ Κῦϱος εἰς Σάϱδεις
ἀφίκετο. “And after Lysander had arrived at Rhodes and secured some
ships there, he sailed to Cos and Miletus, and from there to Ephesus,
where he remained with seventy ships until Cyrus arrived at Sardis”
(X. HG. 1.5.1)
We see that in (36) and (37) both procedures are interchangeable and it is clear that there is no difference of use
between the two markers. And the same can be said with regard
to ἔστε + aorist stem. Consider the pair
(38) καὶ πϱοπέμψαι <τοὺς πολεμίους> ἐπέταξε τοὺς ἀμφ’ αὐτὸν
ἱππεῖς ἔστε ἐν τῷ ἀσφαλεῖ ἐγένοντο. “And he ordered the horsemen
who were with him to escort <the enemies> until they were in a safe
30
place” (X. Ages. 2.13)
(39) καὶ ταῦτα ἐποίουν μέχϱι σκότος ἐγένετο· “and did these things
31
until darkness came on” (X. An. 4.2.4)
Thus we see that ἕως is interchangeable with μέχϱι and that,
in its turn, μέχϱι is interchangeable with ἔστε. The situation,
then, has clearly evolved with regard to Thucydides’ stage of the
language, with a simplification of the system, in which all these
__________
28
Cf. also, for example, Th. 2.19.2, 6.62.3, 7.26.3 and 7.35.2.
Cf. also, for example, Th. 1.109.4, 2.22.2 and 5.26.4.
30
Cf. also X. An. 2.5.30, 3.1.28 and 3.4.49.
31
Cf. also X. An. 7.1.1.
29
165
markers are now interchangeable, and the incorporation of a
marker ἔστε, not used by Thucydides.
And the same can be said of Plato, as is suggested by the
following texts:
(38) νομοθετούντων δ’ αὐτὰ οἱ νέοι πϱὸς τὰ τῶν πϱόσθεν
νομοθετήματα ἀπομιμούμενοι…, μέχϱιπεϱ ἂν πάντα ἱκανῶς δόξῃ
κεῖσθαι· “So the young lawgivers shall make laws for these cases,
modelling… until it seems that everything has been adequately
established” (Pl. Lg. 846.c)
(39) τὰ δὲ δημόσια < δικαστήϱια> καὶ κοινὰ … ἔστ’ ἐν πολλαῖς
πόλεσιν οὐκ ἀσχήμονα ἐπιεικῶν ἀνδϱῶν οὐκ ὀλίγα νομοθετήματα,
ὅθεν νομοφύλακας χϱὴ τὰ πϱέποντα τῇ νῦν γεννωμένῃ πολιτείᾳ
κατασκευάζειν…, ἕως ἂν ἱκανῶς αὐτῶν ἕκαστα δόξῃ κεῖσθαι, “In
dealing with the public <courts> of the State… there exist in many
States quite a number of admirable ordinances of worthy men, from
which the Law-wardens must construct a code suitable to the polity
now …, until each such ordinance be deemed satisfactory” (Pl. Lg.
957.b)
The marker ἔστε does not appear in Plato and, on the other
hand, ἕως and μέχϱι are completely equivalent. There is also the
fact that, as in Xenophon, μέχϱι, is used mostly as a preposition
and relatively little as a conjunction. We have further proof of
this equivalence in Plato in the pair
(40) μέχϱι τούτου αὐτοῦ ἕως ἂν εἰδῶμεν εἴτε… “to this same, just
32
until we can see whether…” (Pl. Tht. 169.a)
(41) μέχϱι τούτου ποιήσονται τὴν διαφοϱάν, μέχϱι οὗ ἂν οἱ αἴτιοι
ἀναγκασθῶσιν… δοῦναι δίκην “but they will carry the conflict to this
point, just until the guilty will compel to do justice…” (Pl. R. 471.b)
What we find in Attic oratory is always μέχϱι τούτου ... ἕως
... .
It is time to draw some conclusions from what we have said.
We have, in sum, been dealing with some sentence procedures
__________
32
This is the normal procedure in Attic. It appears also, for example, in
Andocides 3.15 or in Demosthenes 18.48.
ipabo_66.249.78.36
Antonio Lillo
166
indicating simultaneity and temporal limit, ἕως, μέχϱι, ἄχϱι, ἐν
ᾧ / ὅσῳ, ἐς ὅ and ἔστ’ἄν, which appear in the Ionic of Herodotus
and which, with certain variations of use, are found again in the
Attic prose of Thucydides, Xenophon and Plato; they are not,
however, used in Attic oratory, a level of Attic which is
considered more “pure” and which only uses the marker ἕως.
This suggests that these procedures are taken from Ionic and
adapted to Attic prose by Thucydides, Xenophon and Plato.
Herodotus presents the most complex system, in which the use
of each marker implies a specific notion which is distinct from
the rest: the use of ἕως and μέχϱι implies that the action
expressed by the main clause has its cause in the action of the
subordinate or is a consequence thereof, while the use of ἐν ᾧ,
ἐς ὅ and ἔστ’ἄν implies that there are two actions, that of the
main sentence and that of the subordinate, which have no relation of consequence, but are simply two simultaneous actions.
This situation persists in Thucydides, though with certain
variants, since this author uses neither ἐς ὅ nor ἔστ’ἄν, but
follows the same general pattern of Herodotus. The situation in
Xenophon represents an evolution, since all the markers, ἕως,
μέχϱι, ἐν ᾧ, ἔστε, are interchangeable and only μέχϱι is used
occasionally with a marked sense of limit, while Plato, whose
text has a “more Attic” air, uses only ἕως and μέχϱι, which are
fully equivalent. We are, then, dealing with texts, that of Thucydides, but especially those of Plato and Xenophon, which, albeit
with with variants, reflect the conversational tone of the Athenian educated classes and thus show how a number of Ionic
features became part of the body of the Attic dialect, the new
Attic, which is at the origin of the Koiné.33
The equivalence of ἕως and μέχϱι in this new Attic can better
explain the fact that ἕως is also used as a preposition with the
meaning “till” precisely from the prepositional value of μέχϱι.
This new use of ἕως appears in Hellenistic Greek, but may
already be present in “recent Attic” (Chantraine, 1968–1980:
__________
33
Cf. López Eire (1981: 392 = 1986: 352).
167
395, s.v. ἕως; Beekes, 2010:493, s.v. ἕως). In fact this prepositional use of ἕως + genitive appears in technical language.
Examples include:
(42) φυλακὴ δὲ ἔστω ἕως ἀπυϱέτου καὶ ἑβδόμης· “to monitor <the
patient> until he is without fever and seven days have passed” (Hp.
34
Acut. (Sp.) 11.19 Littré)
(43) εἶτ’ ἐφεξῆς ἕως τῶν ἀτόμων “and then proceed in order down
to those that are not further divisible” (Arist. Top. 109b.16)
(44) Λιβύη δὲ τὸ ἀπὸ τοῦ Ἀϱαβικοῦ ἰσθμοῦ ἕως Ἡϱακλέους
στηλῶν. “And Lybia extends from the Arabian isthmus to the Pillars
of Heracles” (Arist. Mu. 393b.32)
(45) καὶ οὕτως εἰς τὸ πϱόσω ἕως τῶν ἀτόμων “and so on to those
that are not further divisible” (Arist. PA 643a.22)
(46) ἕως τῆς ἰνός “until the tendon” (Thphr. HP 3.12.7)
Text (42) belongs to an appendix to a work by Hippocrates,
whose author was probably not the latter, so that it is possible to
date the text in the 4th century BC or even later. But the chronology of texts (43), (44) and (45) is clear and shows unequivocally
that the use of ἕως as a preposition is typical of 4th century BC
technical language, though not of literary language, such as that
of Demosthenes, whose birth and death coincide with those of
the philosopher. A later example is (46), a text by Theophrastus,
which also uses technical language.
But the use of ἕως as a preposition followed by the genitive
in the 4th century BC with the meaning of “till” is now extended
to the language of the inscriptions, as, for example, in
(47) ἕως της λατομίας (SEG 28.697 A 19, Asclepieion of Cos, end
of the 4th century BC = IG XII 4.1.264)
__________
34
For further exemples cf. Kühn/Fleischer (1989:356), s.v. ἕως.
ipabo_66.249.78.36
168
Antonio Lillo
But ἕως also appears associated with the preposition εἰς in the
phrase ἕως εἰς also with the meaning “till”. In fact, this phrase
appears in the same inscription of text (47):
(48) ἀπὸ δὲ τοῦ το ὁμαλοῦ κατὰ κϱάναν ἕως εἰς τ[ὸ] (SEG 28.697
A 12 = IG XII 4.1.264)
(49) ἕως ἐς τὸν Ἀϱνικέφαλον (SEG 28.697 A 13 = IG XII 4.1.264)
Ἕως εἰς is also found in Aristotle, in
(50) ὃ καλοῦσιν οἰσοφάγον… ἕως εἰς τὴν κοιλίαν “the so-called
oesophagus… till the stomach” (Arist. PA 650a.17)
As proposed with the use of ἕως as a conjunction, the phrase
ἕως εἰς is to be explained by parallel with μέχϱι in the phrase
μέχϱι ἐς:
(51) <τὸ ὕδωϱ> ψύχεται μέχϱι ἐς ἠῶ “the water grows cool until
the sunset” (Hdt. 4.181.4)
But as already noted with regard to texts (25) to (28), this ἕως
is not yet equivalent to μέχϱι in Herodotus; accordingly, the
appearance of the phrase ἕως εἰς can only take place when ἕως
and μέχϱι are equivalent, i.e., at a later stage to Herodotus.
Therefore, the text presented by Hude and Rosén, editors of
Herodotus, does not seem acceptable,
(52) ἐντειλάμενος ἐς τὸ ὀπίσω δι’ Ἡϱακλέων στηλέων διεκπλέειν
ἕως ἐς τὴν βοϱηίην θάλασσαν καὶ οὕτω ἐς Αἴγυπτον ἀπικνέεσθαι
“instructing them to sail on their return voyage past the Pillars of
Heracles until the northern sea and so to arrive to Egypt” (Hdt. 4.42.2)
More acceptable to us seems the text of the manuscripts R
and V, διεκπλέειν ἐς τὴν βοϱηίην θάλασσαν, texts that are
tacitly supported by Powell (1938:157, s.v. ἕως), who does not
include the ἕως of this passage in his lexicon.
169
Bibliography
Adrados, F. R. (1999): Historia de la lengua griega: de los orígenes a
nuestros días, Madrid.
Beekes, R. (2010): Etymological Dictionary of Greek, Leiden.
Chantraine, P. (1968–1980): Dictionnaire étymologique de la langue grecque. Histoire des mots, Paris.
Classen, J. - Steup, J. (1892): Thukydides III, erklärt von --, Berlin.
– (1922): Thukydides VIII, erklärt von --, Berlin.
Fuchs, A. (1902): Die Temporalsätze mit den Konjunktionen „bis“ und „so
lange als“, Würzburg.
Gautier, L. (1911): La langue de Xénophon, Genève.
Hiersche, R. (1970): Grundzüge der griechischen Sprachgeschichte bis zur
klassischen Zeit, Wiesbaden.
Hoffmann, O., Debrunner, A. u. Scherer, A. (1969): Geschichte der griechischen Sprache I, Berlin 1969.
Hude, C. (1927): Herodoti Historiae, Oxford.
Kühn, J.-H. - Fleischer, U. (1989): Index Hippocraticus, Göttingen.
López Eire, A. (1981): «Del ático a la koiné», Emerita 49, 377–392 = López
Eire (1986), Estudios de lingüística, dialectología e historia de la lengua
griega, Salamanca, 337–352.
Meillet, A. (1971): Aperçu d’une histoire de la langue grecque (8ª éd.), Paris.
Meisterhans, F. u. Schwyzer, E. (1900): Grammatik der attischen Inschriften
(3ª ed.), Berlin3.
Monteil, P. (1963): La phrase relative en grec ancien, Paris.
Palmer, L. R. (1980): The Greek Language, London.
Powell, J. E. (1938): A Lexicon to Herodotus, Cambridge (rpr. Hildesheim
1979)
Rijksbaron, A. (1976): Temporal and Causal Conjunctions in Ancient Greek,
Amsterdam.
Rosén, H. B. (1987): Herodotus Historiae, Leipzig.
Rosenkranz, B. (1930): „Der lokale Grundton und die persönliche Eigenart in
der Sprache des Thukydides und der älteren attischen Redner“, IF 48,
127–178.
Threatte, L. (1996): The Grammar of Attic Inscriptions II. Morphology,
Berlin.
ipabo_66.249.78.36
Etimo, uso e significato di v
Di ANIKA NICOLOSI, Parma
Il composto, pur non comune, è ben documentato nei poemi
omerici e nell’epos più in generale, ma è attestato anche in altri
autori. Implicando un’idea di ‘gratificazione’, suscettibile di
slittamento semantico verso la nozione di ‘abbondanza’,
l’epiteto viene di norma riferito a doni: Hom. Il. XIX 143s.
  /  , Ap. Rh. I 657, Orph. Arg. 1202
 ,  a pasti e, più in generale, a cibi e bevande:1
Hom. Il. IX 90  , 227s.   /
, XXIII 29   , Od. V 165s.
      /… , 267 
...  v, VI 76s. ’  / ,2
XX 390s.  ... /    v, Ap. Rh. II 495,
Opp. Hal. II 567  . Nonostante la risorsa di un
discreto manipolo di occorrenze, la comprensione del significato
e di conseguenza la traduzione proposta non è sempre univoca,
variando a seconda del contesto in cui il termine viene, di volta
in volta, utilizzato. Spesso il composto indica ‘grande quantità’
o ‘adeguata misura’, concordato con i referenti più diversi, come
in Hom. Il. XXIII 139  ... , 650 
, Od. IX 158  , XII 382 v’
 (cf., con diverso valore, Ap. Rh. IV 1353 
 ... ), XIII 273   (così anche
Hom. H. Merc. 330, Opp. Hal. V 374), XIV 232 
_____________
1
In Hom. Od. XIII 409   è cibo di scrofe (cf. Ael. NA
V 45,19         
’  v); l’aggettivo è riferito a bevande in Nic. Th. 507
 , Al. 359   , 515   

2
Cf. Suda  362 A. s.v. , rr. 5s. ”   v :
 ’  v    /     
 i’, dove il primo esametro è versione variata di Od. VI 76,
il secondo ripete esattamente III 480.
Glotta 89, 170–177, ISSN 0017–1298
171
  e XVI 429  …  ,
due luoghi, questi ultimi, nei quali il vocabolo denota
l’abbondanza del bottino di guerra.
Oltre al composito spettro semantico, è da considerare
un’incertezza già antica circa almeno uno dei due componenti
del termine: se indubbia risulta, infatti, la presenza di  nel
primo membro, meno chiara appare la derivazione del secondo
elemento del composto, che potrebbe muovere secondo alcuni
da , ‘cedere’, ‘concedere’, secondo altri da , ‘essere
opportuno’ o ‘conveniente’.3 Ambigui si dimostrano talora in
_____________
3
I dizionari moderni si dividono in egual misura tra le due scelte:
propendono per  LSJ9 (A Greek English Lexicon, compiled by H. G.
Liddell and R. Scott, rev. and augm. by H. S. Jones, with the assist. of R.
McKenzie, al., Oxford 19409) 1103 s.v. “suited to the desires, satisfying,
agreeable, to one’s taste” (vd. anche Revised Supplement, ed. by P. G. W.
Glare, with the assist. of A. A. Thompson, Oxford 1996, 206; R. Renehan,
Greek Lexicographical Notes. A Critical Supplement to the Greek-English
Lexicon of Liddell-Scott-Jones, Second Series, Göttingen 1982, 99 s.v.);
Gemoll (G. G., Vocabolario greco-italiano, traduzione e aggiunte di D. Bassi
e E. Martini, Firenze 196128 [19221; Wien 1908]) 666s. s.v. («che assente al
desiderio, che rallegra il cuore; ricco, eccellente»); Passow (F. P., Handwörterbuch der griechischen Sprache, Darmstadt 1841–18575) II/1, 182 s.v.
(“dem Verlangen, der Neigung, Begier od. Leidenschaft angemessen,
entsprechend, genügend, das Verlangen stillend, sättigend od. befriedigend,
also genügend, hinlänglich, reichlich”). Di contro, prediligono  Rocci
(L. R., Vocabolario greco-italiano, Roma-Napoli-Città di Castello 19433
[19391]) 1207 s.v. («che soddisfa l’animo, sufficiente, piacevole, gradito,
eccellente, lauto, copioso»); Bailly (A. B., Dictionnaire Grec-Français, Paris
196326 [195016, 18941]) 1252 s.v. («qui plaît à l’esprit, qui réjouit le coeur,
agréable […] p. suite convenable»); mentre Montanari (F. M., GI. Vocabolario della lingua greca, Torino 20042 [19951]), fornendo la plausibile spiegazione del lemma («conforme al desiderio quindi gradevole, eccellente, lauto,
copioso»), nella prima edizione (p. 1256 s.v.) accredita una derivazione da
, nella seconda (p. 1330 s.v.) da . Ugualmente i moderni etimologici oscillano tra il referente , preferito da Chantraine (P. C., Dictionnaire étymologique de la langue grecque, Paris 1968–1980) 685 s.v. 
(«conforme au désir, qui satisfait le désir»), ed  (< ), avallato da
Boisacq (E. B., Dictionnaire Étymologique de la Langue Grecque, Heidelberg-Paris 1923) 627 s.v.  («qui répond au désir, qui l’apaise >
abondant») e da Hofmann (J. B. H., Etymologisches Wörterbuch des Griechischen, München 1949) 197 s.v.  (“dem Wunsche entsprechend, überfließend”); la derivazione non è esplicitata da Frisk (H. F., Griechisches
etymologisches Wörterbuch, I–II, Heidelberg 1954–1973) II 208 s.v. 
(“dem Geist angemessen, herzerfreuend, reichlich”), né da Beekes (R. B.,
Etymological Dictionary of Greek, Leiden-Boston 2010) II 931 s.v. 
ipabo_66.249.78.36
172
Anika Nicolosi
proposito i lessici antichi. La Suda  614 A. s.v. v
chiosa il termine con  , spiegazione che pare insistere su un’idea di sazietà nella misura. Più interessante Esichio,
nel quale si registrano due glosse pertinenti:  853 L. s.v
    (locus classicus il succitato Hom. Od. XVI 429), dove i glossemata propongono un’idea
sia di piacevolezza che di quantità, e  854 L. s.v. 
0     , interpretamentum
in cui, a sanare la corruttela,4 proporrei di leggere  –
ravvisando in   (cf. Hom. Il. XXIII 29) un caso
riconducibile al meccanismo della ‘coppia contigua’5 –, cf.
anche schol. Hom. Od. XIII 409  
    (scil.  ). Una propensione per la derivazione da  mostra Et. M. 580 s.v.
          
(cf. già Orion  98 s.v      ),
mentre in Ps.-Zon.  1348 Tittm    
_____________
(«suitable for the mind, delightful, plentiful»). Per quanto riguarda i lessici
omerici, Ebeling (H. E., Lexicon Homericum, I, Lipsiae 1880) 1064 s.v.
apoditticamente afferma: «a  et  omnes ducunt»; anche Nazari (O.
N., Dialetto omerico, Torino 19042; 18931) 227 s.v. accredita la derivazione
da  e spiega: «che basta alla brama, sufficiente, copioso, lauto, gradito,
caro»; non si pronuncia Monro (D. B. M., A Grammar of the Homeric
Dialects, Oxford 1882, 82 (§ 124), limitandosi a tradurre: “pleasing to the
spirit”.
4
Tràdito è :   (Marc. Gr. 622), corretto dal Musuro,
che nel glossema proponeva  («an ?» Latte). Locus
classicus di riferimento sarà non Hom. Od. V 166, come indicato dal Latte,
ma piuttosto Hom. Il. XXIII 29 (vd. A. Nicolosi, Hesychianum, Eikasmos
XXIII, 2012, 315s.).
5
In particolare il nostro esempio rientrerebbe nella casistica delle
cosiddette «glosse sintattico-contestuali», costituite da un aggettivo chiosato
da un sostantivo a cui si riferisce in un determinato contesto, vd. E. Degani,
Problemi di lessicografia, BIFG IV (1977–1978) 135–146: 142–146 (rist. in
Filologia e storia. Scritti di Enzo Degani, Hildesheim-Zürich-New York
2004, 730–741: 737–741) e F. Bossi - R. Tosi, Strutture lessicografiche greche, BIFG V (1979–1980) 7–20: 19s.; più in generale sul fenomeno della
‘coppia contigua’, vd. B. Marzullo, La «coppia contigua» in Esichio, QIFG
III (1968) 70–87: 74s. (= La «coppia contigua» nella glossografia di Esichio,
in AA.VV., Studia classica et orientalia Antonino Pagliaro oblata, III, Roma
1969, 85–105: 90s.) e R. Tosi, Studi sulla tradizione indiretta dei classici
greci, Bologna 1988, 127–130.
173
        
   ) affiora l’idea di un’afferenza del
primo membro del composto a . Meno orientata la
spiegazione fornita da Ap. Soph. 111,7 Bekk. 
        
‘ ’.
In favore di una derivazione da  si pronunciano Arist.
Gramm. De signis, ad Il. XXIII 139 ( schol. ad l.) 
         
     v  ;    
  , nonché schol. Hom. Od. XIV 232
     . Altri materiali scoliastici
riportano diverse spiegazioni, per lo più alquanto vaghe: oltre al
già menzionato schol. Hom. Od. XIII 409, si possono segnalare
schol. Hom. Od. V 166    
    , schol. Nic. Al. 515f 
  , schol. Opp. Hal. II 567,2 
       
    .
Come si è detto, in molti luoghi in cui il termine è attestato,
esso indica ‘grande quantità’ intesa non di rado come ‘eccesso’.6
Lo stesso schol. Hom. Od. XVI 429  
     i, ad esempio, evocando l’area semantica di termini suscettibili di valenza rilevata,
come  e , suggerisce di fatto un’idea di dismisura: in
tal senso, il significato del composto parrebbe non facilmente
conciliabile con una derivazione da , che diverrebbe quasi
ossimorica.7 L’idea di una misura ‘conveniente’ e ‘appropriata’,
_____________
6
In esempi come Ap. Rh. III 984   e IV 1353
 …  il significato del termine pare aver perduto qualcosa
della connotazione originaria; nel primo caso, infatti, l’impiego dell’aggettivo
è funzionale ad esprimere, più che il ‘gradimento’, il ‘bisogno’ dei ,
nel secondo passo il composto definisce ‘giusta’, ‘equa’, una ricompensa (in
Hom. Od. XII 382 il senso di   è invece quello di ‘pena
adeguata’).
7
Tale esegesi può valere anche per   in Hom. Il. XXIII
29, in relazione all’iperbolica mattanza di vittime sacrificate da Achille in
occasione dei funerali di Patroclo.
ipabo_66.249.78.36
174
Anika Nicolosi
in effetti, si addice maggiormente al composto ,8 come
è esplicitato in Eustath. ad Hom. Il. XXIII 246, 1298,6 (IV
719,12s. Valk), dove, in riferimento alla provvisoria sepoltura di
Patroclo (vv. 245s      
, /   ) e con esplicito richiamo al
contesto luttuoso che caratterizza il canto, il Tessalonicese –
chiamando in causa  – spiega:     
      e prosegue (1298,
6–9 [IV 719,13–17 Valk:      
         
        ‛
’ (Il. XXIII 139)      
  ‛’        
        
        .9
Il significato di , peraltro, ben si coniuga con un’idea di
eccesso, essendo il verbo spesso impiegato col valore metaforico
_____________
8
Derivato da  +  (vd. Chantraine, DELG 355,3 s.v.  e 358
s.v. ; Frisk, o. c., I 536 s.v.; Beekes, o. c., I 435 s.v.  e 440 s.v.
), di contro ad , che muove invece da  + 
(Chantraine, DELG 318 s.v. ; Boisacq, o. c., 265 s.v. e 222 n.1 s.v. ;
Hofmann, o. c., 87 s.v.); quest’ultimo composto, piuttosto raro e sempre preceduto da negazione col significato primario di ‘non cedevole’, quindi ‘insopportabile’, ‘indomabile’, è detto di forza (Hom. Il. VIII 32), dolore (Hom.
Il. XVI 549), azioni (Hom. Od. VIII 307) e attributo di  in Hom. Il. V
892 e Od. XIX 493; Frisk, o. c., I 536 e Beekes, o. c., I 440 s.v. 
(sic) ammettono anche un’eccezionale sovrapposizione di significato con
, peraltro solo in relazione all’ambigua valenza di Hom. Od. VIII
307. Una certa confusione semantica testimoniano Hesych. o 1694 L. s.v. 
  , Add. in Et. Gud. e 503 s.v.   
   , schol. D Hom. Il. XVI 549   
  
9
L’equipollenza dei due termini è riaffermata in un altro luogo dello
stesso Eustazio ad Hom. Il. XIX 147s. (1176,50s. [IV 302,21–23 Valk])
   ‛ ’     ‛’  
          
           
 ‛ ’ Analogo procedimento assimilativo si riscontra anche
nello schol. Hom. Il. XXIII 139b      
‛    ’ (Hom. Il. XXIII 50    
            

175
di ‘cedere’ alle passioni e agli impulsi dell’animo (cf. ad es.
Hom. Il. IX 109s., XXIV 42s.  / );10 analoga impronta contraddistingue, pur in modo diverso, l’espressione formulare di Hom. Il. XXII 459 (= Od. XI 515)    
 Il ThGL V 784 s.v.  offre come traducenti
«Animo gratus, Animum permulcens,    s. 
   : aut    ,
a , quo significatur » e, con riferimento ad Hom.
Il. XIX 144, aggiunge: «verba Agamemnonis ad Achillem,
quem muneribus placare vult», attingendo quindi alla spiegazione fornita da Eustazio nel commento al passo iliadico in
questione (1176,29s. [IV 301,23s. Valk])   ‛
’        
    e in altri due luoghi: ad Hom. Il. IX
227 (750,13 [II 709,31s. Valk])    
          
          e ad Hom.
Od. V 166 (1529, 5–7 [I 207,16s. Stallbaum   ‛ 
       
        11 Da segnalare pure Dio Chrys. Or. 2,48,4, dove si legge 
      .
L’ambigua derivazione del secondo membro del composto
già per gli antichi è evidenziata da due luoghi classici che, in
palese contraddizione tra loro, paiono ricondurre il vocabolo ai
due diversi semantemi implicati. Si tratta di Ap. Rh. I 657s.
  
       
dov’è esplicitata la parentela con , e di Plut. Phoc. 2,3–4
_____________
10
Cf. ThGL III 233D s.v. 
Cf. anche schol. D Hom. Il. IX 90    
 schol. D Hom. Il. IX 227     
schol. Hom. Il. XIX 144b1     
 schol. Nic. Al. 359g    
11
ipabo_66.249.78.36
176
Anika Nicolosi
        
       
         
         
  12
dove si accredita invece la dipendenza del termine da  per
il tramite del richiamo al composto con preverbio 13
Tuttavia un’analisi della catena associativa condotta sulla scorta
degli indici a rovescio14 porta a constatare che i pur non molto
numerosi composti terminanti in -15 vale a dire  vel
 e   e  
e , muovono tutti da /16 mentre sono
riconducibili ad  solo alcune forme nominali  e
), il già menzionato , e derivati in  come
 e 17 Ambigua solo per l’erudizione antica è
la derivazione del semplice , ricondotto in un caso ad
 (cf. Theognost. Can. 15,32s. [= 90,3s.] Cramer  [scil.
]  ), in un altro a  (cf. Al. Aphrod. Quaest.
62,4           
_____________
12
Cf. anche Plut. De tuenda sanitate praecepta 133e Klaerr  
        “”  
   
13
Jucquois-Devlamminck (G. J. - B. D., Compléments aux dictionnaires
étymologiques du grec ancien, I, Louvain 1977) p. 72 s.v. sottolineano la
subordinazione semantica del verbo semplice al composto: «le sense ‘céder’
est dû à l’ ‘irradiation’ de certains pré-verbes, p. ex. -».
14
Cf. C. D. Buck - W. Petersen, A Reverse Index of Greek Nouns and
Adjectives, Chicago 1945, 718 e P. Kretschmer - E. Locker, Rückläufiges
Wörterbuch der griechischen Sprache, Göttingen 1944, 234.
15
Isolato risulta Hesych. k 1580 L. s.v. , chiosato dal lessicografo (e quindi da LSJ9 923 s.v. e ThGL IV 1318C s.v.) con .
16
Concordi gli etimologici moderni, vd. Chantraine, DELG 355,3 s.v.
, 358 s.v. ; Frisk, o. c., I 38 s.v. , 536 s.v. ;
Beekes, o. c., I 38 s.v. , 440 s.v. ; vd. anche ThGL I 741 s.v.
. Probabilmente un errore, data la contraddizione, il rinvio ad  di
Chantraine, DELG 33 s.v. .
17
Vd. Chantraine, DELG 318 s.v. ; Frisk, o. c., I 454 s.v. ;
Beekes, o. c., I 382 s.v. , 440 s.v. 
177
  ).18 In conclusione, se da un punto
di vista linguistico è giocoforza riconoscere la derivazione di
 da a, d’altro canto ben si comprende la facile
confusione da parte di antichi e moderni, favorita dall’assonanza
con , dalla peculiare accezione del verbo e dai contesti in
cui spesso il composto viene impiegato.
_____________
18
Cf. LSJ9 485 s.v. Oscillanti e farraginose le spiegazioni fornite in Et. M.
297 s.v.         
              
            
           
         
     e in Ps.-Zon. 634 Tittm. s.v.  
             
ipabo_66.249.78.36
Ein linguistisches Problem mit Konkubinen*
Von RAFAŁ ROSÓŁ, Poznań
1. Einführung
Das im Titel angekündigte Problem stellt sich sowohl für
Forscher der klassischen Sprachen als auch für Semitologen. Es
handelt sich um folgende Wörter: griech.  bzw.
, lat. paelex und hebr. pileḡeš / pîleḡeš. All diese
Wörter haben die Bedeutung ‘Konkubine, Kebsweib’1 und sind
ohne Zweifel miteinander verwandt, aber ihr Verhältnis zueinander ist in der Forschung seit langem sehr umstritten.
Griech. , - ist seit Homer belegt und wird
allgemein für eine mithilfe des Suffixes -- gebildete Ableitung
aus der Form  f. gehalten,2 die in den griechischen
Quellen erst in klassischer Zeit erscheint (zum ersten Mal bei
Herodot).3 Darüber hinaus notiert Etymologicum Magnum
__________
*
Für die sprachliche Korrektur des Textes möchte ich Frau Cecilie Koch
meinen aufrichtigen Dank aussprechen.
1
Für griech.  /  nimmt man auch die Bedeutung
‘Mädchen’ an; vgl. LSJ: 1293. Diese Ansicht basiert auf den Belegen der
Grammatiker und zwar: Eust. 763.22–24:     
         
          
          
 (= Ael. Dion.  6 Erbse (dubitans), in: H. Erbse, Untersuchungen zu den
attizistischen Lexika, Berlin 1950, S. 95–151); EM 649.51: ,
     ; Phot., Lex. (Glossae Homericae):
 ; vgl. Ael. Dion.  8 Erbse (dubitans). Dazu vgl. aber W. J.
Slater 1986: 30, der schreibt: „ is not attested as anything but
concubine; any attempts to make it mean ‘young girl’ are probably derived
from an etymology of “.
2
Dazu s. vor allem M. Meier-Brügger 1975: 21 u. 37; er stellt dabei fest:
„Die Wahl von  ist bei Homer wohl metrisch bedingt“ (S. 37).
3
Zu griech.  und  s. z.B. K. Erdmann 1949; LfgrE:
III, 947.
Glotta 89, 178–194, ISSN 0017–1298
Ein linguistisches Problem mit Konkubinen
179
(649.51) die Form  f. und betrachtet sie als dorisch.4
Vom Stamm - wurden auch die maskuline Form
 m. (Inschr. aus Samos, 3./2. Jh. v.Chr.; Aristoph. Byz.,
Fr. 42 Slater u.a.) bzw.  m. (Eust.; viell. schon Ael.
Dion.)5 mit der Bedeutung ‘Junge’ und eine Diminutivform
 ‘ds.’ (Plat. Com.; Ael. Dion.;6 Hsch.) gebildet; überdies bucht Hesychios die Glosse  7 In den
antiken Quellen kommen einige weitere Ableitungen aus
- oder - vor, nämlich  n. ‘Konkubine’ (Plut.),  bzw.  f. ‘Konkubinat’
(seit Is.),  m. ‘Sohn. einer Konkubine’ (Sophr.),
 oder  ‘eine Konkubine sein (auch
in sakraler Prostitution)’ (Str.; Plut. u.a.) oder ‘als eine Konkubine haben’ (Hdt.).8 Die aufgelisteten Belege weisen darauf hin,
dass die ganze Wortgruppe auf  /  ‘Konkubine, Kebsweib’ zurückgeht. Erst sekundär sind die Formen mit
der Bedeutung ‘Junge’ (wohl zuerst ‘Geliebter’) entstanden. Aus
‘Konkubine, Kebsweib’ hat sich auch die – nur marginal
bezeugte – Bedeutung ‘Mädchen’ entwickelt.9
Bereits Aulus Gellius10 brachte das griechische Wort mit dem
lateinischen paelex, -icis f. in Verbindung. Das Wort tritt zum
ersten Mal in einem Fragment der Leges Regiae auf. Bekannt sind
__________
4
S. Anm. 1. Vgl. Aul. Gell. 4.3.3: ‘Paelex’ autem quasi , id est
quasi . Ut pleraque alia, ita hoc quoque vocabulum de Graeco
flexum est.
5
Eust. 1419.50 (= Ael. Dion.  8 Erbse (dubitans)); vgl. Eust. 763.21 u.
1742.37. Des Weiteren vgl. ᾱ  m. ‘Jungling’ (Philistid.) (nach
ᾱ?).
6
Ael. Dion.  7 Erbse:      (= Eust.
1419.51).
7
Vgl. Phot., Lex.:   .
8
Überdies vgl.  n. ‘Konkubinat’ (Gloss.) und 
(mit dem Itazismus statt ) n. ‘Page’ (Pap.).
9
Manche Forscher plädieren für die umgekehrte Bedeutungsentwicklung,
d.h. ‘Mädchen’ → ‘Konkubine, Kebsweib’; so z.B. A. Walde 1921: 85f.; B.
Čop 1952: 222f.; ähnlich A. Fick 1897: 125.
10
S. Anm. 4. Vgl. auch Non. Marc. 6 Lindsay: Pelicis a graeco voacabulo
significantiam sapientes inflexam putant, quasi pallex, hoc est ut ,
quodsi hoc non est, vana compositio hominis videri potest.
ipabo_66.249.78.36
180
Rafał Rosół
auch zwei Nebenformen, d.i. pēlex (mit vereinfachter Schreibung des Diphthongs) und pellex (mit volksetymologischer
Assoziation mit dem Verb pellicere ‘anlocken, verlocken’).11
Seit Cicero kommt auch die Ableitung paelicātus, -ūs m.
‘Verhältnis mit einem Kebsweib’ häufig vor.
Im Bibelhebräischen ist das Wort pileḡeš / pîleḡeš f. 36 Mal
bezeugt, und zwar in der Genesis (4mal), im Buch der Richter
(12 Mal), im zweiten Buch Samuels (8 Mal) usw.12 Seit dem
Mittelhebräischen ist auch die Ableitung pîlegšûṯ f. ‘Konkubinat’ bekannt. Eine ähnliche Bezeichnung für ‘Konkubine, Kebsweib’ findet sich in der jüdisch-aramäischen Sprache, d.i. plqh,
plqth, paləqtā, pîllaqtā, pîlaqtā, pillaqtā,13 und im Syrischen,
d.i. plqD,14 diese Formen können jedoch einerseits nur eine späte
Entwicklung von hebr. pîleḡeš / pileḡeš darstellen, andererseits
unter dem Einfluss von griech.  bleiben. Weitere Anknüpfungen sind im semitischen Sprachbereich nicht bekannt.15
In vielen, besonders älteren Arbeiten wurde versucht, eine
Verbindung von griech.  / , lat. paelex und
hebr. pileḡeš mit den Angaben aus anderen Sprachen herzustellen.16 Am häufigsten wurde avest. pairikā f. ‘böser weiblicher Dämon in tierischer oder menschlicher Gestalt (auch als
schöne Frau personifiziert)’, herangezogen.17 Dieses Wort
wurde im Mittelpersischen zu parik und im Neupersischen zu
perī, d.i. ‘Peri (ein guter Genius)’ weiterentwickelt (vgl. auch
__________
11
Zu lat. paelex s. ThLL: X.1, 37–40; K. Erdmann 1942.
Zu hebr. pileḡeš s. in erster Linie HALAT: 878; K. Engelken 1989; T.
Davidovich 2007; vgl. auch M. Jastrow 1903: 1177. Im samaritanischen
Pentateuch-Text kommt die Form filgåš vor; s. HALAT: 878.
13
DJPA: 437 gibt plqh / plqth an, J. Levy 1881: II, 271 – pîllaqtā und
paləqtā, M. Jastrow 1903: 1185 – pillaqtā und pîlaqtā, G.H. Dalman 1922:
337 und HALAT: 878 – paləqtā.
14
LS: 576.
15
Manchmal zieht man arab. Bilqīs, der Name der Königin von Saba,
heran; s. HALAT: 878; M. Ellenbogen 1962: 134. Vgl. aber R.G. Stiegner
1979: 128–136, die sehr skeptisch gegenüber dieser Gleichung ist.
16
Vgl. z.B. LEW: 233f.; J.B. Hofmann 1950: 251; DELL: 474; GEW: II,
468.
17
Über die parikās s. z.B. M. Boyce 1989: 86. Vgl. Ch. Bartholomae
863f. („‘Zauberin, Hexe’ (...) Die Parikas schaden besonders dadurch, dass
sie Masdayasner durch Liebeskünste vom Glauben abspenstig machen“).
12
181
die aus dem Iranischen entlehnten Wörter parik im Armenischen
und parī im Pashtu).18 Außer den iranischen Formen wurde das
mittelirische airech, Gen. airige f. ‘Konkubine’ erwähnt, bei
dem wir es mit einem Schwund des anlautenden Labials zu tun
hätten.19 Heutzutage wird zumeist vermutet, dass avest. pairikā
und mir. airech – trotz der Abweichung in semantischer Hinsicht – urverwandt sind.20 Aus phonetischen Gründen werden sie
jedoch mit der griechisch-lateinisch-hebräischen Isoglosse nicht
mehr in Verbindung gebracht.21
2. Bisherige Erklärungsversuche
In erster Linie stellt sich die Frage, in welcher Relation lat.
paelex zu den griechischen Formen  und 
steht. In der Forschung herrscht die Überzeugung vor, dass das
lateinische Wort aus dem Griechischen entlehnt ist; diese Hypothese bedarf aber wegen der ziemlich großen Abweichungen
zwischen der lateinischen und der griechischen Form weiterer
Begründung.22
__________
18
Zu den iranischen Formen s. z.B. P. Horn 1893: 310.
Zu mir. airech s. R. Thurneysen 1924: 146f.; R. Matasović 2009: 127f.
20
So z.B. IEW: 789 (uridg. *parīkā f. ‘Buhlerin, Konkubine’); R.
Matasović 2009: 127f. (aufgrund von diesen Wörtern (und dem – mit Fragezeichen versehenen – lat. Parcae erschließt er urkelt. *ferikā f. ‘Konkubine’
und uridg. *perikeh2 f. ‘Dienerin?’; er gibt aber zu: „A rather speculative
etymology“). Ähnlich J. P. Mallory – D. Q. Adams 1997: 123 mit der Bemerkung: „Phonologically, the two forms are comparable, although the semantic
relationship is somewhat distant“; J. P. Mallory - D. Q. Adams 2006: 208. G.
Dumézil (1940: 68) setzt diese Formen auch in Beziehung zu armen. harč
‘Konkubine, Kebsweib’ (aus *pargyā- < *parikyā- ); Zustimmung bei P.
Friedrich 1979: 228; B. A. Olsen 1999: 82f. (mit weiteren Argumenten).
Ganz verfehlt ist die von G. Dumézil 1940: 68 vorgeschlagene Anknüpfung
an griech.  ‘Ehefrau’.
21
Vgl. DELG: 853f.; M. de Vaan 2008: 439; R. Beekes 2010: 1147.
22
Vgl. z.B. die Aussage in ThLL: X.1, 37 und zwar: „coniungitur (scil.
paelex) nimirum cum gr.   fere eiusdem sensus, sed quomodo
forma explicanda sit non liquet“.
19
ipabo_66.249.78.36
182
Rafał Rosół
Aulus Gellius23 suggeriert, dass paelex aus  f.
stammt, und seine Ansicht findet mehrere Anhänger in der
modernen Forschung.24 Abgesehen davon, dass griech. 
nur bei den Grammatikern vorkommt, wäre es aus phonetischen
Gründen unhaltbar, diese Form als Quelle für paelex zu
betrachten.25
Eine Erklärung der Phonetik hat A. Walde (1921: 85–89)26
vorgeschlagen. Er hält paelex für Entlehnung der dialektalen
Form *. Demzufolge rekonstruiert er die ursprüngliche
Form *¤, in der sich die Gruppe -¤- zu der Geminate -oder – wie es im Kyprischen und vielleicht im Elischen der Fall
ist – durch Metathese zu -- entwickelt hätte. Als Argument
führt er avest. pairikā. Er weist auch auf griech.  ‘Knabe,
Mädchen’ als eventuelle Parallele hin. Da die postulierte
Metathese jedoch nur auf einem sehr begrenzten Gebiet der
griechischen Dialekte vorkam, hat diese Interpretation keine
Überzeugungskraft.27
Eine andere Erklärung stammt von A. Ernout (1930: 122),
der behauptet, dass das lateinische Wort nicht direkt, sondern
durch etruskische Vermittlung aus dem Griechischen entlehnt
ist. Diese Ansicht wird von einigen Forschern geteilt, aber
bisher wurde kein etruskisches Pendant gefunden.28
Wegen der phonetischen Schwierigkeiten bei der Ableitung
des lateinischen Wortes aus dem griechischen hält M. de Vaan
(2008: 439) die beiden Wörter für nicht verwandt und schlägt
für paelex eine einheimische Etymologie vor. Er sieht in paelex
dieselbe Wurzel wie in paedor, -oris m. ‘Schmutz, Dreck’ und
erschließt die Form *paed-Vk-s, obwohl er selbst feststellt: „this
is only a guess which cannot be substantiated“. Einerseits ist die
__________
23
S. Anm. 4.
So z.B. O. Weise 1882: 484; G. A. E. A. Saalfeld 1884: 798.
25
Vgl. L. Ceci 1892: 96, Anm. 1.
26
S. auch A. Walde 1930–32: II, 7. Zustimmung bei M. Leumann 1977:
69; vgl. LfgrE: III, 947 s.v. .
27
Waldes Hypothese weist schon LEW: 233 zurück.
28
Zu dieser Erklärung neigen z.B. LEW: 233; DELL: 474; GEW: II, 468;
DELG: 853f.; G. Bai 2009: 87.
24
183
Trennung des lateinischen paelex von den griechischen Wörtern
nicht gerechtfertigt, andererseits muss de Vaans Etymologie aus
semantischen Gründen abgelehnt werden.
In der älteren Forschung gab es kaum Zweifel daran, dass
paelex ein griechisches Lehnwort ist, weshalb vor allem nach
der Etymologie von  gesucht wurde. Bisher wurde eine
Reihe indogermanischer Etymologien vorgeschlagen, die aber
mit Recht keine allgemeine Anerkennung gefunden haben.29
Die Suche nach einer einheimischen Etymologie für 
ist auch mit der weit verbreiteten Ansicht verbunden, dass das
hebräische Wort aus dem Griechischen übernommen worden
sei.30 Historisch gesehen ist diese Auffassung wenig überzeugend, weil hebr. pileḡeš bereits in den ältesten Büchern des
Alten Testaments vorkommt.31
Nach einigen Forschern handelt es sich hierbei nicht um eine
einfache Entlehnung, sondern um ein Wanderwort unbekannter
Herkunft.32 Diese Hypothese hat jedoch an Attraktivität verloren, seit die oben genannten iranischen und keltischen Angaben
nicht mehr in Betracht gezogen werden. Auch die gelegentlich
postulierten Verbindungen mit Wörtern aus anderen Sprachen
__________
29
Zu griech.  wurden – außer den oben erwähnten iranischen
und keltischen Formen – folgende Wörter herangezogen: 1). skr. bandhu
‘Verwandter, Freund’, bandhu-ka ‘Bastard’, bandha-kî ‘liederliches Weib’; s.
A. Bezzenberger bei A. Müller 1877: 295f.; 2). lat. pellis ‘Fell’; s. O.
Wiedemann 1904: 26f.; vgl. LEW: II, 234; 3). griech.  ‘Fohlen’, dt.
Fohlen usw.; s. A. Fick 1897; F. Bechtel 1914: 268f.; vgl. A. Walde 1930–
32: II, 7; DELG: 854; 4). aksl. človĕkъ ‘Mensch’ (auch mit got. kalkjō
‘Hure’); s. A. Bezzenberger 1890: 249; vgl. F. Fröhde 1891: 308; C.C.
Uhlenback 1902: 124; W. Prellwitz 1905: 350; 5). engl. fellow ‘Gefährte’; s.
W. Prellwitz 1905: 350; 6) skr. paºate ‘kaufen’, paºastrî ‘Hure’, lit. pelnas
‘Verdienst, Erwerb’; s. F. Fröhde 1891: 308; 7) toch. B klīye ‘Frau’, toch. A
kuli ‘ds.’, ir. caile ‘Dörflerin, Fräulein, Mädchen’ u.a.; so B. Čop 1952: 222f.
30
So z.B. F. C. Movers 1856: 80–82; E. B. Pusey 1885: 517; A. von
Gutschmid 1890: 5; O. Wiedemann 1904: 26f.; A. Walde 1921: 86; LEW:
234; DELL: 474; Lipiński 1997: 562; vgl. W. Gesenius 1815: 65; E. Renan
1878: 209f.; LfgrE: III, 947f. s.v.  u. .
31
Vgl. T. Davidovich 2007: 25.
32
S. LEW: II, 233f.; GEW: II, 468; R. G. Stiegner 1979: 128–131; vgl.
auch V. Pisani 1968: 125f.; Mayser 1970: 31; E. J. Furnée 1972: 336 („wohl
vorderasiat. Wanderwort“); E. Klein 1987: 505.
ipabo_66.249.78.36
184
Rafał Rosół
sind wenig plausibel.33 Ohne Überzeugungskraft bleibt ebenfalls
die Hypothese, dass griech.  und hebr. pileḡeš ägäischen bzw. vorgriechischen Ursprungs sind.34
In der Forschung ist auch die Ansicht verbreitet, dass der
Konkubinenname im Hebräischen unter dem Einfluß von den
Philistern erschien, die am Ende des zweiten Jahrtausends aus
dem Westen gekommen sind und sich in Palästina angesiedelt
haben.35 Da wir aber von ihrer Sprache nach wie vor nichts
Sicheres sagen können, hat diese Hypothese keine festen
Anhaltspunkte.
Eine ähnliche Erklärung schlägt J. P. Brown (1995–2001: I,
70)36 vor, der vermutet, dass der griechisch-lateinisch-hebräische
__________
33
Es handelt sich hier um die folgenden Formen: 1). armen. alič, alaxin
‘Dienerin’; s. V. Pisani 1938: 45, Anm. 35 u. 1968: 126 vgl. LEW: 234; 2).
armen. harč ‘Konkubine’; s. H. von Ewald 1859: 344, Anm. 1; C. C.
Uhlenbeck 1902: 124; vgl. Ch. Rabin 1974: 356 u. 360; s. auch oben Anm.
11; 3). got. kalkjō f. ‘Hure’; s. R. Thurneysen 1924: 146f.; vgl. C. C.
Uhlenbeck 1902: 124; LEW: 234; 4) kanaresisch palagu ‘Tänzerin’; s. F.
Focke 1953: 406; 5). aind. pālāgalÏ- f. ‘die vierte (geringste) Frau eines
Fürsten’; s. V. Pisani 1968: 124–127 (pālāgalÏ- aus *pālāga- < *pālāg/ka-);
vgl. R.G. Stiegner 1979: 129f.; EWAia: II, 124.
34
So z.B. A. Cuny 1910: 162; R. Beekes 2010: 1147 (er weist auf das
angebliche vorgriechische Suffix -- hin). Des Weiteren vgl. V. Georgiev
1936: 97f.; 1941–45: 95, der eine „pelasgische“ Etymologie zu aind. (nachved.) bālá- ‘jung, kindlich; Kind, Knabe’, bālá ‘Mädchen, junge Frau’, aind.
bála- ‘Kraft, Stärke’ u.a. postuliert; Zustimmung bei A. J. van Windekens
1954; vgl. D. A. Hester 1965: 362. Darüber hinaus vgl. die Ansicht von F.
Focke 1953: 406f.: „ und seine Sippe (scil.  
 etc.) gehört danach wahrscheinlich einer sowohl vorindogermanischen wie vorsemitischen Sprachschicht an. Nennen wir sie „indoatlantisch“,
weil sie von Indus bis zum Atlantik gereicht hat und – obwohl sicher
vielfältig aufgespalten – zu einer im ganzen einheitlichen Gesittungsstufe –
man spricht mit Recht von einer Hochkultur – gehört zu haben scheint“.
35
So R. A. S. Macalister 1914: 80; F. Bork 1939–41: 228; G. Garbini
1997: 221–226 (bes. 223f.) u. 242; E. Lipiński 2006: 52. Vgl. G. W.
Ahlström 1993: 326; D. I. Block 2003: 78, Anm. 197 („the quadraliteral form
may point to non-Semitic, perhaps Philistine, origin“). An philistinische
Herkunft denkt auch M. Görg 1979: 10f., der aber griech.  /
 nicht berücksichtigt und behauptet, dass die Philister das Wort aus
dem Ägyptischen entlehnt haben; als Quelle dieses Wortes betrachtet er
ägypt. jrj-gs ‘neben jemandem befindlich’ mit dem (maskulinen !) Artikel p3
(die Grundbedeutung wäre ‘der neben jemandem Befindliche’); Kritik bei T.
Davidovich 2007: 26.
36
Anders J. P. Brown 1968: 169.
185
Konkubinenname (zusammen mit dem Konkubinat selbst) auf
eine anatolische Sprache, vielleicht auf das Luwische oder
Hethitische, zurückgeht37 und im Mittelmeerraum unter Vermittlung der nur wenig bekannten Pelasgo-Philistiner verbreitet
wurde. Gegen diese Hypothese spricht vor allem die Tatsache,
dass die anatolischen Quellen keine entsprechende Wortform
liefern.38
Die Ansicht, dass ein ähnliches Wort in der pelasgischen oder
philistinischen Sprache existierte, wurde auch von Ch. Rabin
(1974) gebilligt, der für die griechisch-lateinisch-hebräische Isoglosse eine gemeinsame indogermanische Etymologie postuliert.
Er erschließt eine Zusammensetzung *epi-legh-s, die aus einer
Präposition *epi- ‘zu, auf, bei’ und aus der Wurzel *legh‘liegen’ besteht. Rabin ist sich dabei im Klaren, dass sich diese
Form in keiner der bekannten indogermanischen Sprachen zu
einer Form wie , paelex oder pileḡeš entwickeln
konnte.39 Darum setzt er voraus, dass es sich hier um eine sehr
schwach belegte indogermanische Sprache handelt, die im
Altertum im Mittelmeerraum verwendet wurde und zwar Pelasgisch oder Philistinisch.
Diese Theorie nimmt S. Levin (1983; 1995: 233–237) an und
versucht, sie mit weiteren Argumenten zu stützen. In seiner
neueren Arbeit postuliert er eine Entlehnung des hebräischen
Wortes aus der messapischen oder einer anderen mit ihr verwandten Sprache. Aus geographischen Gründen ist diese Hypothese jedoch noch weniger plausibel als die von Rabin.
In der Forschungsgeschichte wurde auch vermutet, dass das
griechische Wort semitischen Ursprungs sei.40 Diese Hypothese
__________
37
Zur Annahme einer kleinasiatischen Herkunft der Bezeichnung für
Konkubinen neigen auch andere Forscher, und zwar A. Müller 1877: 296
(„Stammt es aus Kleinasien?“); J. B. Hofmann 1950: 251 (griech. 
„samt lat. paelex (...) entl. aus einer kleinasiat. Sprache“); M. Ellenbogen
1962: 134.
38
Vgl. G. Bai 2009: 86.
39
Zur Kritik s. T. Davidovich 2007: 25.
40
So z.B. P. de Lagarde 1856: XXVI; C. Lottner 1858: 165; O. Schrader
1890: 207; L. Ceci 1892: 96; L. M. Epstein 1934‒35: 153; A. Wadler 1935:
ipabo_66.249.78.36
186
Rafał Rosół
stößt jedoch auf ein morphologisches Problem: Hebr. pileḡeš
kann nur als Quelle für die Form  betrachtet werden,
die aber – morphologisch gesehen – eine Ableitung aus 
sein soll.
Eine Kompromisslösung schlägt W. Muss-Arnolt (1892:
65f.) vor. Er nimmt an, dass  auf das von ihm erschlossene phönizische Substantiv *pallāgā zurückgeht, das auf der
semitischen Wurzel p-l-g ‘trennen, teilen’ basieren würde.
Demzufolge bedeute das phönizische Wort ursprünglich ‘eine
Frau, die von ihrer Familie getrennt wurde’ oder – wie G. Bai
(2009: 87) postuliert – ‘eine Frau, die das Bett mit einem Mann
teilt’. Diese Entlehnung hänge mit dem Sklavenhandel zusammen, d.h. die Konkubinen seien ursprünglich Sklavinnen gewesen, die den Griechen von den Phöniziern verkauft worden
seien. Die Form *pallāgā sei im Griechischen als 
wiedergegeben worden, und daraus sei dann im griechischen
Sprachgebiet die Form  entstanden, die ihrerseits als
pileḡeš ins Hebräische entlehnt worden sei.41 Es sei hier hervorgehoben, dass die Rekonstruktion der ganz isolierten Form
*pallāgā eigentlich keine Grundlage hat. In semantischer Hinsicht ist zudem die Verbindung mit der Wurzel p-l-g ‘trennen,
teilen’ wenig überzeugend.
Eine ähnliche, aber noch weniger plausible Hypothese stellt
H. Lewy (1895: 66f.) auf. Er zieht die semitische Wurzel p-l-¡
bzw. p-l-q ‘spalten, furchen’ heran und glaubt, dass sie dem
griechischen Verb  /  /  ‘Geschlechtsverkehr haben’ zugrunde liegt. Erst von diesem Verb sei – Lewy
zufolge – das Substantiv  abgeleitet. Hebr. pileḡeš
gehe wiederum auf die sekundäre Form  zurück.
__________
111f.; W. Plautz 1963: 9, Anm. 31.; W. Krause 1970: 94; Vgl. W. Prellwitz
1905: 350; LfgrE: III, 947. Darüber hinaus vgl. W. Burkert 1984: 42.
41
Außer W. Muss-Arnolt und G. Bai gehört auch M. Bernal 2006: 396f.
zu den Anhängern dieser Hypothese; vgl. auch M. Bernal 2001: 414, Anm.
59.
187
3. Ein neuer Erklärungsvorschlag
Obwohl die oben besprochenen Erklärungen verfehlt zu sein
scheinen, ist es aus kulturgeschichtlichen Gründen sehr wahrscheinlich, dass eine Art der Polygynie samt der Bezeichnung
für die Konkubine in der griechischen Welt semitischen Ursprungs ist, weil die Institution des Konkubinats bei den semitischen Völkern sehr verbreitet war. Beispielsweise kann man die
Belege aus der Bibel heranziehen, aus denen sich ergibt, dass
die frühesten Könige Israels wie Saul, David und Salomon
zahlreiche Konkubinen hatten. Bekanntlich findet sich die Polygynie heutzutage in der arabischen Welt immer noch.
Trifft diese Vermutung zu, ist das hebräische Wort für
Konkubine als ursprünglich zu betrachten.42 Man kann auch vermuten, dass es in der phönizisch-punischen Sprache, die mit der
hebräischen sehr nahe verwandt ist, eine ähnliche Form gab.
Meines Erachtens ist es möglich, eine Verknüpfung zwischen
dieser Form und den griechischen Formen  und
 herzustellen. Da die Form  viel früher
belegt ist als , ist es nicht sicher, dass sie wirklich
sekundär mithilfe des Suffixes -- gebildet worden ist. Es liegt
– im Gegenteil – nahe, dass hebr./phön. pileḡeš als 
übernommen wurde und die Form  erst sekundär entstand. Es handelt sich hier um eine Rückbildung durch Analogie
zu Paaren wie  und   f. ‘Hetäre’ u.ä.
Für die Anpassung eines semitischen Fremdwortes an die
Substantiva mit dem Suffix -- im Griechischen bietet der
Name des Jaspis eine Parallele, und zwar   f. (seit
Plat.) zu akkad. (j)ašpu ‘Jaspis’, hebr. jāšp±ē ‘ds.’, syr. jašfē /
jašfā ‘ds.’.43 Im Fall von  ist eine solche Anpassung
umso mehr begründet, als im Auslaut des semitischen Wortes
der Konsonant š steht, der im Griechischen fast regelmäßig
__________
42
Vgl. W. Gesenius 1815: 65 („Allein pileḡeš scheint wirklich vielmehr
aus dem polygamischen Morgenlande zu den Griechen gekommen zu sein
(...)“).
43
Zu griech.  s. H. Lewy 1895: 56; É. Masson 1967: 65f.
ipabo_66.249.78.36
188
Rafał Rosół
durch  wiedergegeben wird. Beachtenswert in diesem Kontext
ist der Tiername  (auch ) m., Akk. Sing.  /  ‘Löwe’
(seit Hom.), der auf hebr. laîš ‘Löwe’, aram. (Deir Alla) ljš ‘ds.’,
jüd.-aram. lajtā ‘ds.’, arab. laiṯ ‘Löwe’ zurückgeht.44
Der semitische Konsonant g (bzw. ḡ) kann im Griechischen
nicht nur durch , sondern auch durch  repräsentiert werden,
worauf in erster Linie der Tiername  m./f. ‘Kamel’
(Aesch., Hdt. usw.) aus hebr. gāmāl ‘ds.’, offiz.-aram. gml ‘ds.’,
jüd.-aram. gamlā ‘ds.’, asüdarab. gml ‘ds.’, arab. ğamal ‘ds.’
usw. hinweist.45 Die semitische Liquida l in intervokalischer
Position konnte entweder durch griech.  oder  wiedergegeben werden, worauf der Gefäßname  m. ‘Art Gefäß,
Melkeimer, Schöpfeimer, Krug’ (seit Hom.) (akkad. gullu(m)
‘Schale, Becken’, ugar. gl ‘Schale’, hebr. gulā ‘Becken, Schale,
Wasserbecken’ usw.)46 und vermutlich der seit Homer belegte
Beiname der Athene   f. (ugar. bElt, phön. bElt,
d.i. BaEalat ‘Herrin’)47 hindeuten.
Was die Vokale betrifft, lassen sich einige semitische Lehnwörter anführen, in denen griech.  gegenüber hebr. i oder e
steht. Ein gutes Beispiel dafür ist  f. ‘Galbanharz (aus
Doldenpflanzen gewonnen), Galbanum (Ferula galbaniflua)’
(seit Thphr.) gegenüber hebr. ¡elbinā ‘Galbanharz (vgl. auch
aram. ¡elbənā, syr. ¡elbānītā ‘ds.’ usw.)’.48
Hinsichtlich des lateinischen Wortes ist der Ansicht von J. P.
Brown (1968: 166; 1995–2001: I, 65) zuzustimmen, dass paelex
__________
44
Dazu s. in erster Linie V. Blažek 2005: 85f., der schreibt: „Greek 
can be derived from Canaanite dialects of the Hebrew type where Semitic *ṯ
changed to š or it can be derived from the Ugaritic like source where *ṯ
remained unchanged or from a source of the later Aramaic type where *ṯ
became t. In the two latter cases the development, within Greek, of *lit-s to
*lis would be expected“. Vgl. H. Lewy 1895: 6f.; É. Masson 1967: 85–87;
GEW: II, 113; DELG: 643.
45
Dazu s. H. Lewy 1895: 1; É. Masson 1967: 66f. Vgl. Hsch.
   .
46
Dazu s. vor allem H. Lewy 1895: 150f.; É. Masson 1967: 39–42.
47
Diese Etymologie akzeptieren z.B. H. Lewy 1895: 251; O. Szemerényi
1974: 155, Anm. 72; V. Blažek 2007; R. Rosół 2010: 99; vgl. auch L. R.
Palmer 1983: 362; LfgrE: III, 247f..
48
Zu griech.  s. z.B. H. Lewy 1895: 45; É. Masson 1967: 60.
189
dem hebräischen Wort phonetisch näher steht als dem griechischen.49 Dies gilt insbesondere für den Diphthong ae < ai,
dessen zweites Element dem hebräischen i entspricht. Des Weiteren hat die lateinische Form eine Liquida, während die griechischen Wörter eine Geminate aufweisen. Daraus ergibt sich, dass
paelex kein griechisches Lehnwort sein kann, sondern vielmehr
als aus einer westsemitischen Sprache entlehnt anzusehen ist.50
Aus historischen Gründen können wir vermuten, dass wir es
hier mit einem punischen Lehnwort zu tun haben, das schon in
der Königszeit übernommen wurde. Aller Wahrscheinlichkeit
nach stand das Erscheinen dieses Wortes im Zusammenhang mit
der Verbreitung einer Art der Polygynie oder des Konkubinats.
Was die Phonetik angeht, weist die lateinische Form eine Synkope in der letzten Silbe des Nominativs auf. Dieses Phänomen
kommt im Lateinischen häufig vor und ist in einem dreisilbigen
Wort – wie in unserem Fall – sehr plausibel.51
Zusammenfassend lässt sich feststellen, dass die Bezeichnung
für die Konkubine bzw. das Kebsweib im Griechischen und im
Lateinischen aus dem Semitischen entlehnt worden ist. Daraus
ergibt sich auch die Schlussfolgerung, dass sowohl die Griechen
als auch die Römer eine Art der Polygynie oder des Konkubinats von den Phöniziern bzw. den Puniern übernommen haben
müssen. Hinsichtlich des Griechischen und Lateinischen wäre
das Problem der Konkubinen damit gelöst, offen bleibt dagegen
weiterhin die Frage nach der Etymologie des semitischen
Wortes.52
__________
49
Vgl. L. Ceci 1892: 96, Anm. 1: „La fonologia non ci permette di
derivare paelex direttamente da . Bisognerebbe quindi rivolgere gli
occhi ai Fenicj; ma comme e quando i Fenicj abbiano importato in Italia la
Maitressenwirthschaft io no vedo“.
50
Vgl. C. Lottner 1858: 165; O. Schrader 1890: 207; K. Erdmann 1942;
W. Plautz 1963: 9, Anm. 31.
51
Zur Synkope im Lateinischen s. M. Leumann 1977: 95–99.
52
Auf jeden Fall scheint hebr. pileḡeš / pîleḡeš nicht echt semitisch zu
sein; vgl. z.B. die Aussage von T. Davidovich 2007: 26: „Another theory (...)
that the origin of pîleḡeš is Semitic, does not have any foundation in the
conventional opinion that the Semitic languages had in the beginning nouns
containing a root of two consonants that later on developed into three. This type
ipabo_66.249.78.36
190
Rafał Rosół
Bibliographie
Ahlström, G. W. (1993): The History of Ancient Palestine from the
Palaeolithic Period to Alexander’s Conquest, Sheffield.
Bai, G. (2009): Semitische Lehnwörter im Altgriechischen, Hamburg.
Bartholomae, Ch. (1904): Altiranisches Wörterbuch, Berlin.
Bechtel, F. (1914): Lexilogus zu Homer, Halle.
Beekes, R. (2010): Etymological Dictionary of Greek, with the assistance of
L. van Beek, vol. I–II, Leiden-Boston.
Bernal, M. (2001): Black Athena Writes back. Martin Bernal Responds to his
Critics, Durham-London.
– (2006): Black Athena. The Afroasiatic Roots of Classical Civilisation, vol.
III: The Linguistic Evidence, New Brunswick (New Jersey).
Bezzenberger, A. (1890): Die indogermanischen Gutturalreihen, BB 16, S.
234–250.
Blažek, V. (2005): Hic erant leones. Indo-European “lion” et alii, JIES 33,
S. 63–101.
– (2007): Theonymica Helleno-Semitica II. Pallas Athânâ / Athânâ Potnia.
‘Virgin’ or ‘Lady’, DO-SO-MO. Fascicula Mycenologica Polona 7, S.
161–176.
Block, D. I. (2003): Marriage and Family in Ancient Israel, in: K. M.
Campbell (ed.), Marriage and Family in the Biblical World, Downers
Grove, S. 33–102.
Bork, F. (1939–41): Philistäische Namen und Vokabeln, AfO 13, S. 226–230.
Boyce, M. (1989): A History of Zoroastrianism, vol. I: The Early Period, 2nd
ed., Leiden etc.
Brockelmann, C. (1928): Lexicon Syriacum, 2. Aufl., Halle, (= LS).
Brown, J. P. (1968): Literary Contexts of the Common Hebrew-Greek
Vocabulary, JSS 13, S. 163–191.
– (1995–2001): Israel and Hellas, vol. I–III, Berlin-New York.
Burkert, W. (1984): Die orientalisierende Epoche in der griechischen
Religion und Literatur, Heidelberg.
Ceci, L. (1892): Le etimologie dei giureconsulti Romani, Torino.
Chantraine, P. (1999): Dictionnaire étymologique de la langue grecque.
Histoire des mots, Avec un Supplément sous la direction de A. Blanc, C.
de Lamberterie, J.-L. Perpillou, Paris, (= DELG).
Čop, B. (1952): Gr.  , Živa Antika 2, S. 221–223.
__________
of three-radical root has become the most common in the Semitic languages.
The asssumption that a rather old word like pîleḡeš with a root of four
consonants, existed in Semitic languages is quite doubtful“.
191
Cuny, A. (1910): Les mots du fonds préhellenique en grec, latin et sémitique
occidental, REA 12, S. 154–164.
Dalman, G. H. (1922): Aramäisch-Neuhebräisches Handwörterbuch zu
Targum, Talmud und Midrasch, 2., verbesserte und vermehrte Aufl.,
Frankfurt (Main).
Davidovich, T. (2007): The Mystery of the House of Royal Women: Royal
Pīlagšīm as Secondary Wives in the Old Testament, Uppsala.
Dumézil, G. (1940): Mitra Varuna: Essai sur deux représentations indoeuropéennes de la souveranité, Paris.
Ellenbogen, M. (1962): Foreign Words in the Old Testament, London.
Engelken, K. (1989): Pilœḡœš, in: H.-J. Fabry, H. Ringgren (hrsg.), Theologisches Wörterbuch zum Alten Testament, Bd. VI, Stuttgart-Berlin-Köln:
Kohlhammer, Sp. 586–589.
Epstein, L. M. (1934–35): The Institution of Concubinage among the Jews,
Proceedings of the American Academy for Jewish Research 6, S. 153–
188.
Erdmann, K. (1942): Paelex, in: RE XVIII.2, Sp. 2225–2227.
– (1949): , in: RE XVIII.3, Sp. 226–229.
Ernout, A. (1930): Les éléments étrusques du vocabulaire latine, BSL 30, S.
82–124.
Ernout, A. / Meillet, A. (1959): Dictionnaire étymologique de la langue
latine, 4e éd., Paris, (= DELL).
Ewald, H. von (1859): Über den heutigen Stand der phönikischen
Forschungen, ZDMG 13, S. 343–358.
Fick, A. (1897):  und , BB 22, S. 125–126.
Focke, F. (1953): Pallas Athene, Saeculum 4, S. 398–413.
Frisk, H. (1960–72): Griechisches etymologisches Wörterbuch, Bd. I–III,
Heidelberg, (= GEW).
Fröhde, F. (1891): Griechische und lateinische Etymologien, BB 17, S. 303–
319.
Furnée, E. J. (1972): Die wichtigsten konsonantischen Erscheinungen des
Vorgriechischen. Mit einem Appendix über den Vokalismus, The HagueParis.
Garbini, G. (1997): I Filistei. Gli antagonisti di Israele, Milano.
Georgiev, V. (1936): Die Träger der kretisch-mykenischen Kultur, ihre
Herkunft und ihre Sprache, Teil I: Urgriechen und Urillyrier (ThrakoIllyrier), Sofia.
– (1941–45): Vorgriechische Sprachwissenschaft, Lief. I–II, Sofia.
Gesenius, W. (1815): Geschichte der hebräischen Sprache und Schrift. Eine
philologisch-historische Einleitung in die Sprachlehre und Wörterbücher
der hebräischen Sprache, Leipzig.
ipabo_66.249.78.36
192
Rafał Rosół
Görg, M. (1979): Piggul und pilaegaeš – Experimente zur Etymologie,
Biblische Notizen 10, S. 7–11.
Gutschmid, A. von (1890): Über den letzten Band von Movers’ Phöniziern,
in: Ders., Kleine Schriften, Bd. II, hrsg. von F. Rühl, Leipzig, S. 1–19.
Hester, D. A. (1965): “Pelasgian” – a New Indo-European Language?,
Lingua 13, S. 335–384.
Hofmann, J. B. (1950): Etymologisches Wörterbuch des Griechischen,
München.
Horn, P. (1893): Grundriss der neupersichen Etymologie, Strassburg.
Jastrow, M. (1903): A Dictionary of the Targumim, the Talmud Babli and
Zerushalmi, and the Midrashic Literature, vol. I–II, New York.
Klein, E. (1987): A Comprehensive Etymological Dictionary of the Hebrew
Language for Readers of English, Jerusalem.
Koehler, L. / Baumgartner, W. (1967–96): Hebräisches und aramäisches
Lexikon zum Alten Testament, Bd. I–VI, 3. Aufl., neu bearbeitet von W.
Baumgartner, J. J. Stamm, Leiden, (= HALAT).
Krause, W. (1970): Griechisch-orientalische Lehnwortbeziehungen, in: D.
Ableitinger, H. Gugel (hrsg.), FS für K. Vretska, Heidelberg, S. 89–115.
Lagarde, P. de (1856): Reliquiae iuris ecclesiastici antiquissimae, Lipsiae.
Leumann, M. (1977): Lateinische Laut- und Formenlehre, München.
Levin, S. (1983): Hebrew pîleḡeš, Greek , Latin paelex: the Origin
of Intermarriage among the Early Indo-Europeans and Semites, General
Linguistics 23, S. 191–197.
– (1995): Semitic and Indo-European: the Principal Etymologies,
Amsterdam.
Levy, J. (1881): Chaldäisches Wörterbuch über die Targumim und einem
großen Theil des Rabbinischen Schriftthums, Bd. I–II, 3., unveränderte
Aufl., Leipzig.
Lewy, H. (1895): Die semitischen Fremdwörter im Griechischen, Berlin.
Lexikon des frühgriechischen Epos (1955–2010): begründet von B. Snell,
hrsg. von M. Meier-Brügger, Bd. I–IV, Göttingen, (= LfgrE)
Liddell, H. G. / Scott, R. / Jones, H. S. (1996): A Greek-English Lexicon.
With a Revised Supplement, Oxford, (= LSJ).
Lipiński, E. (1997): Semitic Languages. Outline of a Comparative Grammar,
Leuven.
– (2006): On the Skirts of Canaan in the Iron Age. Historical and Topographical Researches, Leuven.
Lottner, C. (1858): Über die Stellung der Italer innerhalb des indoeuropäischen Stammes, KZ 7, S. 161–193.
Macalister, R. A. S. (1914): The Philistines. Their History and Civilization,
London.
193
Mallory, J. P. / Adams, D. Q. eds. (1997): Encyclopedia of Indo-European
Culture, Chicago-London.
– (2006): The Oxford Introduction to Proto-Indo-European and the ProtoIndo-European World, Oxford.
Masson, É. (1967): Recherches sur les plus anciens emprunts sémitiques en
grec, Paris.
Matasović, R. (2009): Etymological Dictionary of Proto-Celtic, LeidenBoston.
Mayer, M.-L. (1967): Note etimologiche IV, Acme 20, S. 287–291.
Mayrhofer, M. (1992–2001): Etymologisches Wörterbuch des Altindoarischen, Bd. I–III, Heidelberg, (= EWAia).
Mayser, E. (1970): Grammatik der griechischen Papyri aus der Ptolemäerzeit, Bd. I: Laut- und Wortlehre, 1. Teil: Einleitung und Lautlehre, 2.
Aufl. bearbeitet von H. Schmoll, Berlin.
Meier-Brügger, M. (1975): --. Zur Geschichte eines griechischen Nominalsuffixes, Göttingen.
Movers, F. C. (1856): Die Phönizier, Bd. II.3, Breslau.
Müller, A. (1877): Semitische Lehnworte im älteren Griechisch, BB 1, S.
273–301.
Muss-Arnolt, W. (1892): On Semitic Words in Greek and Latin, TAPhA 23,
S. 35–157.
Olsen, B. A. (1999): The Noun in Biblical Armenian. Origin and WordFormation - with Special Emphasis on the Indo-European Heritage,
Berlin-New York.
Palmer, L. R. (1983): Mycenaean Religion: Methodological Choices, in: A.
Heubeck, G. Neumann (hrsg.), Res Mycenaeae, Akten des VII. Internationalen Mykenologischen Colloquiums (Nürnberg, 6.–10. April 1981),
Göttingen, S. 338–366.
Pisani, V. (1938): Paleontologia linguistica. Note critiche e metodologiche,
Cagliari.
– (1968): Sanskrit pālāgalÏ und pālāgalás, ZDMG 118, S. 124–127.
Plautz, W. (1963): Monogamie und Polygamie im Alten Testament, ZAW 75,
S. 3–27.
Pokorny, J. (1959): Indogermanisches etymologisches Wörterbuch, BernMünchen, (= IEW).
Prellwitz, W. (1905): Etymologisches Wörterbuch der griechischen Sprache,
2. verbesserte Aufl., Göttingen.
Pusey, E. B. (1885): Daniel the Prophet, New York.
Rabin, Ch. (1974): The Origin of the Hebrew Word pileḡeš, JJS 25, S. 353–
364.
Renan, E. (1878): Histoire générale et systèm comparé des langues sémitiques, 1e partie, 5e édition revue et augmentée, Paris.
ipabo_66.249.78.36
194
Rafał Rosół
Rosół, R. (2010): Wschodnie korzenie kultu Apollona. Studium
lingwistyczno-historyczne, Kraków.
Saalfeld, G. A. E. A. (1884): Tensaurus Italograecus. Ausführliches
historisch-kritisches Wörterbuch der griechischen Lehn- und Fremdwörter im Lateinischen, Wien.
Schrader, O. (1890): Sprachvergleichung und Urgeschichte. Linguistischhistorische Beiträge zur Erforschung des indogermanischen Altertums, 2.,
vollständig umgearbeitete und beträchtlich vermehrte Aufl., Jena.
Slater, W. J. (1986): Aristophanis Byzantii Fragmenta, Berlin-New York.
Sokoloff, M. (1990): A Dictionary of Jewish Palestinian Aramaic of the
Byzantine Period, Ramat-Gan, (= DJPA).
Stiegner, R. G. (1979): Die Königin von SabaD in ihren Namen. Beitrag zur
vergleichenden semitischen Sagenkunde und zur Erforschung des Entwicklungsganges der Sage, Graz.
Szemerényi, O. (1974): The Origins of the Greek Lexicon: Ex oriente lux,
JHS 94, S. 145–157.
Thesaurus linguae Latinae (1900–): Lipsiae, (= ThLL).
Thurneysen, R. (1924): Keltisches, IF 42, S. 143–148.
Uhlenbeck, C. C. (1902): Zur gotischen Etymologie, PBB 27, S. 112–136.
Vaan, M. de (2008): Etymological Dictionary of Latin and the Other Italic
Languages, Leiden-Boston.
Wadler, A. (1935): Der Turm von Babel. Urgemeinschaft der Sprachen,
Basel.
Walde, A. (1921): Lateinische Etymologien, IF 39, S. 74–93.
– (1930–32): Vergleichendes Wörterbuch der indogermanischen Sprachen,
hrsg. und bearbeitet von J. Pokorny, Bd. I–III, Berlin-Leipzig.
– (1938–54): Lateinisches etymologisches Wörterbuch, Bd. I–III, 3. Aufl.
von J. B. Hofmann, Heidelberg, (= LEW).
Weise, O. (1882): Die griechischen Wörter im Latein, Leipzig.
Wiedemann, O. (1904): Etymologien, BB 28, S. 1–83.
Windekens, A. J. van (1954): , BzN 5, S. 221–224.
La dislocation à droite en latin
Par OLGA SPEVAK, Toulouse
Résumé: L’objectif du présent article est d’examiner le phénomène de
la «dislocation à droite» qui, dans les grammaires traditionnelles
latines, entre dans la catégorie d’apposition. En me fondant sur un
corpus constitué d’œuvres de Plaute, de Térence, de Cicéron et de
Pétrone, j’essaierai de proposer une typologie des constituants disloqués à droite. Ces derniers se distinguent par certaines propriétés:
coréférentiels avec un terme (nom ou pronom), énoncé au préalable,
ils fournissent une spécification de son contenu (générique > spécifique) ou apportent des évaluations subjectives d’un contenu. Les constituants disloqués à droite remplissent une fonction pragmatique
particulière en spécifiant, a posteriori, un terme énoncé au préalable.
Cependant, ils ne se définissent pas par le placement en fin de phrase:
un constituant en position finale n’est pas nécessairement «disloqué à
droite».
1. Introduction
La dislocation à droite en latin, à la différence de la dislocation à gauche, n’a pas encore fait l’objet d’une étude systématique. C. Cabrillana leur a réservé une brève section dans la
nouvelle syntaxe de J. M. Baños Baños (2009: 120); moi-même
(Spevak 2010: 111–114), je leur ai consacré quelques mots dans
mon livre sur l’ordre des constituants dans la prose latine
classique. Parmi les articles, on peut mentionner, tout d’abord,
l’étude d’A. Toth (1994) qui annonce, dans son titre, les constituants appelés «Koda» mais qui, en réalité, n’accorde que très
peu d’attention à la dislocation à droite; ensuite, celle de R.
Amacker (2001), assez discutable à mon avis, qui avance une
hypothèse sur la segmentation prosodique de la phrase latine.
En revanche, des linguistes travaillant sur les langues modernes se sont vivement intéressés au phénomène de dislocation, à
gauche aussi bien à droite. La théorie de la grammaire fonctionGlotta 89, 195–221, ISSN 0017–1298
ipabo_66.249.78.36
196
Olga Spevak
nelle selon S. C. Dik (1997 II: 389 sq.) attribue à ces constituants des fonctions pragmatiques particulières. Les constituants
qui précèdent une proposition indiquent l’entité ou le domaine
dans le cadre desquels la phrase est à interpréter. Ces constituants sont appelés «Thème»,1 par exemple:
(a) Ta sœur, je l’ai vue hier.
(b) Quant aux étudiants, ils ne seront pas invités.
Les constituants ajoutés en fin de phrase spécifient un
constituant exprimé dans la phrase (c) ou l’ensemble du contenu
de la phrase (d). Ils représentent une sorte d’afterthought, une
addition qui spécifie, clarifie ou corrige quelque chose qui a déjà
été dit. Il peut aussi s’agir de préciser une idée qui découle du
contexte. De tels constituants sont appelés Tails (Dik 1997 II:
401):
(c) I didn’t like it very much, that book of yours.
(d) John gave the book to a girl, in the library.
La position en fin de phrase est habituelle pour ce type de
constituants; cependant, S. C. Dik (1997 II: 401) ajoute qu’ils
peuvent aussi apparaître en tant que parenthèses à l’intérieur
d’une proposition après le terme qu’ils spécifient:
(e) I saw John hand it – the money I mean – to the girl.
(f) He pretended that it was there – in the library – that the whole
thing took place.
R. Geluykens (1987), dans son étude sur corpus, a retenu les
caractéristiques principales de ces constituants disloqués à droite:
- ils sont typiques de la conversation spontanée;
- ils représentent une stratégie conversationnelle de «réparation». Le locuteur, après avoir énoncé un élément pronominal
dans la proposition (cf. l’exemple (c)), fournit une information
__________
1
Pour plus de détails et des références aux travaux antérieurs, voir O.
Spevak (2010: 107 sqq.).
197
plus explicite en utilisant un constituant disloqué à droite. Il a
recours à cette stratégie lorsqu’il estime que sa phrase pourrait
souffrir d’un manque de clarté;
- ils sont précédés d’une brève pause;
- ils sont souvent accompagnés d’une expression métalinguistique de type I mean (cf. l’exemple (e)).
En outre, R. Geluykens (1987: 126) fait remarquer que la
majorité de Tail constituents dans son corpus de l’anglais
contemporain concernent les sujets de la phrase. Et, comme les
sujets fonctionnent souvent comme des «topiques»2 en anglais,
les constituants disloqués à droite sont typiquement des afterthoughts concernant les topiques.
K. Lambrecht (1994: 203 sq.), pour sa part, appelle les
constituants disloqués à droite «antitopics». Il montre que le
référent du constituant disloqué à droite est, sur le plan
informationnel, saillant, et il introduit des exemples en allemand
pour prouver ce rapport entre l’antitopique et la saillance
pragmatique.
Il convient de mentionner encore une étude moderne sur
corpus (comprenant le finnois), celle de M. Fernandez-Vest
(2008). Elle soutient, elle aussi, que la dislocation à droite est
typique de la langue parlée, comme cela ressort du tableau 1: les
constituants disloqués à droite sont quasi-absents des textes
écrits.
Tableau 1: Dislocation à gauche (détachement initial) et à droite
(détachement final) d’après M. Fernandez-Vest
Corpus
oral
texte écrit (édité)
monologues
dialogues
(total)
détachement
initial
70 %
20 %
0,5 %
détachement
final
30 %
80 %
0%
__________
2
Selon la grammaire fonctionnelle, le topique est une fonction pragmatique attribuée à l’entité «dont on parle».
ipabo_66.249.78.36
Olga Spevak
198
L’exemple (g), emprunté à M. Fernandez-Vest, illustre les
deux détachements, à gauche et à droite. Tous deux concernent
le topique, l’entité dont on parle – en l’occurrence, l’escabeau.
(g) [Tu ne m’avais pas parlé d’une histoire d’escabeau?] Oui!
Alors il est tombé / un jour il a voulu monter sur un escabeau
(information saillante) / et pis il avait pas vu que l’escabeau
(détachement initial) il avait pas la corde! Tu sais / on met une
ficelle / pour pas que ça s’ouvre / l’escabeau (détachement
final)! (extrait d’un corpus de français du sud-ouest)
Le latin est une langue que nous ne connaissons que sous la
forme écrite; nous ne disposons pas de locuteurs natifs pour
examiner la langue spontanée. Dans les textes latins, un exemple
comme (g) n’a pratiquement aucune chance d’être relevé. Mais,
faut-il en conclure que la dislocation à droite en est absente?
2. Les constituants disloqués à droite en latin.
Problèmes d’identification
L’identification des constituants disloqués à droite (exemples
(c-f)3 est rendue difficile en latin par le fait que nous ne pouvons
pas utiliser l’intonation comme critère; celle-ci n’est établie que
d’une manière hypothétique. Selon C. Cabrillana (2009: 120),
ces constituants disloqués à droite ne se définissent que d’un
point de vue pragmatique: ce sont des constituants qui complètent et clarifient, de façon additionnelle, une information déjà
présentée dans la proposition. En même temps, C. Cabrillana
signale qu’ils sont parfois difficiles à distinguer des appositions. Il convient de s’attarder brièvement sur ces dernières:
__________
3
À titre provisoire, j’appellerai ces constituants «disloqués à droite» dans
le présent article. Ils se distinguent non pas seulement par un placement
spécial (à droite de la phrase) mais aussi par le fait qu’ils remplissent une
fonction pragmatique particulière, comme on l’a vu dans l’introduction. Ce
sont les Tail constituents dans la terminologie de la grammaire fonctionnelle.
En allemand, A. Toth (1994) les appelle Koda; en espagnol, C. Cabrillana
(2009: 120) parle d’appendice.
199
(1a) Dionysius, seruus meus, qui..., aufugit. (Cic. Fam. 13.77)
«Dionysius, mon esclave, qui..., s’est enfui.»
Les appositions de type (1a) se distinguent par certaines propriétés (voir Longrée 1990 et Lavency 1997: 120):
- l’accord en cas (éventuellement en genre et en nombre)4
avec le nom (ou le syntagme) sur lequel elles portent;
- l’identité référentielle du nom et de l’apposition.
L’identité référentielle se manifeste dans la possibilité d’omettre,
tout au moins dans le cas d’appositions déterminatives,5 l’un ou
l’autre élément:
(1b) Dionysius aufugit.
(1c) Seruus meus aufugit.
En outre, D. Longrée (1990: 11) a mis en évidence que
l’ordre des termes, Dionysius, seruus meus, signale lequel des
deux noms précise l’autre; dans l’expression seruus meus,
Dionysius ce serait le nom propre qui apporterait une information sur seruus meus. L’apposition ne peut alors pas être
antéposée, car le terme antéposé ne sert plus d’apposition:
(1d) Seruus meus, Dionysius, aufugit.
Dionysius est, dans de tels cas, à interpréter comme une apposition s’il est délimité par des pauses à l’oral ou par des virgules à
l’écrit. Ces délimitations sont difficiles à établir pour le latin. En
effet, l’absence de pause entraîne un changement de fonction de
ce nom propre: dans seruus meus Dionysius (cf. Cic. Att. 9.3.1),
le nom propre ne fonctionne pas comme une apposition mais
__________
4
Pour ce qui est de l’accord en genre et en nombre, cf. Ch. Touratier
(1994: 442): «accord en cas et, dans la mesure du possible, en nombre et en
genre». Pour Ernout-Thomas (1953: 135), «l’accord en genre et en nombre
n’a rien de nécessaire».
5
En fonction de la nature de l’expansion, M. Lavency (1997: 120) établit
une distinction entre l’apposition déterminative (1a) et l’apposition qualificative (cf. ci-dessous, exemple (2)). Cf. également H. Pinkster (1995: 122 et
note p. 344) sur l’apposition restrictive (qui détermine le nom) et l’apposition
non-restrictive (qui ne le détermine pas).
ipabo_66.249.78.36
200
Olga Spevak
comme une épithète. Dans cet article, il est impossible de traiter
en détail ce phénomène, les appositions «étroites» de type urbs
Roma ou rex Ancus. Ces constructions sont, traditionnellement,
analysées comme appositives (Szantyr 1972: 426, ErnoutThomas 1953: 135, Fugier 1983: 242, parmi d’autres). Cependant, D. Longrée (1990), M. Lavency (1997: 119) et F. Heberlein (1996) ont mis en évidence que, par leurs propriétés
syntaxiques, les éléments Roma et Ancus fonctionnent comme
des épithètes.6 M. Lavency (1997: 119) les appelle «épithètes
nominales»:
(1e) Seruus meus Dionysius aufugit.
Une autre difficulté d’analyse des éléments dits «appositionnels» réside, pour nous, dans la variabilité de l’ordre des
constituants en latin. La propriété typique de l’apposition est
d’être apposée, c’est-à-dire d’être adjacente et de former ainsi,
avec le nom qu’elle complète, une séquence contigüe et noninterrompue (Serbat 1996: 35). Cependant, il n’est pas rare de
rencontrer des appositions séparées du nom qu’elles complètent.
Par exemple, l’apposition – qualificative – hominis honestissimi
en (2) n’est pas placée au contact du nom qu’elle modifie, Sex.
Pompei Chlori.
(2) Quid ego hic nunc Sexti Pompei Chlori testimonium recitem
qui causam Dionis egit, qui omnibus rebus interfuit, hominis
honestissimi, tametsi ciuis Romanus... (Cic. Verr. 2.2.23)
«Est-ce le lieu et le moment de vous donner lecture du
témoignage de Sextus Pompeius Chlorus, qui a plaidé la cause
de Dion, qui s’est intéressé à toutes ces affaires? C’est un
homme très honorable; et quoique citoyen romain...»
__________
6
Cf. également Ch. Touratier (1994: 443 sq.). Pour F. Heberlein (1996),
urbs et Roma n’entretiennent pas un rapport de dépendance mais un rapport
de «Soziation». H. Fugier (1983: 242) a proposé de paraphraser ces
syntagmes en: haec urbs Roma est. Toutefois, je considère que les «associations» telles urbs Roma, flumen Garumna, arbor ficus... représentent des
expressions où il s’agit de préciser la classe générique (urbs, flumen, arbor) à
laquelle appartient l’entité désignée par un nom propre (Roma, Garumna) ou
un nom commun plus spécifique (ficus).
201
En (3), le rapport entre meus sodalis et Apoecides est de type
«déterminatif». Cette phrase est prononcée lorsqu’un personnage déjà connu du public arrive sur la scène. Sur le plan
informatif, meus sodalis est l’élément le plus informatif de sa
phrase. Faute d’intonation, on peut hésiter entre une interprétation par apposition (cf. (1d)) ou une interprétation par épithète
nominale (cf. (1e)). Cependant, dans le contexte donné, on peut
considérer que meus sodalis suffirait à lui seul à identifier le
référent et, partant, l’élément Apoecides7 ne serait qu’une
spécification additionnelle. En d’autres termes, nous pouvons
avoir affaire à un constituant disloqué à droite (auquel cas ce
nom serait précédé d’une pause). Toutefois, une lecture comme
une épithète nominale déterminative disjointe du reste du syntagme ne peut être exclue avec sûreté; dans ce cas, le locuteur
rappellerait au public l’information sur l’identité et le statut du
personnage («mon ami Apécide»).8
(3) Sed meus sodalis it cum praeda Apoecides. (Plaut. Epid. 394)
«Mais voici mon camarade Apécide qui revient avec son
butin.»
Revenons-en à la dislocation à droite. C. Cabrillana (2009:
120) affirme qu’à la différence des appositions, l’accord en
genre, en nombre et en cas n’est pas nécessaire pour les constituants disloqués à droite; ils peuvent le présenter (4), mais
l’accord n’est pas requis. En (5), argentum et uiginti minas9 ont
le même cas mais non pas le même genre ni le même nombre.
__________
7
Cependant, les noms propres présentent des particularités référentielles
spécifiques, point sur lequel je reviendrai (voir le commentaire qui suit
l’exemple (10)).
8
En effet, dans la comédie, un personnage peut «extérioriser» ses pensées
et décrire ce qui se déroule sur la scène.
9
Sans doute par erreur, dans l’exemple (5), la forme uiginti minae (au
lieu de minas) est éditée dans la syntaxe latine espagnole (p. 120). Les
éditions critiques que j’ai consultées (Budé, Teubner, Oxford Classical Texts)
sont unanimes sur minas.
ipabo_66.249.78.36
202
Olga Spevak
(4) Em istic homo te articulatim concidit, senex, / tuos seruos.
(Plaut. Epid. 488–489)
«Eh bien, c’est cet individu qui t’a réduit en bouillie, mon
vieux, c’est ton esclave.»
(5) Ait se ob asinos ferre argentum atriensi Saureae, / uiginti
minas. (Plaut. Asin. 347–348)
«Il dit qu’il apporte l’argent des ânes à l’intendant Sauréa,
vingt mines.»
Un autre exemple (sans accord en nombre), donné par C.
Cabrillana (2009: 685), est emprunté à la prose historique de
Tite-Live:
(6) Priori Remo augurium uenisse fertur, sex uultures, ... (Liv.
1.7.1)
«Le premier augure fut, dit-on, pour Rémus: c’étaient six
vautours, ...»
Dans cette section, j’ai mentionné les problèmes majeurs
d’identification des constituants disloqués à droite. Dans les
sections suivantes, je chercherai à établir les propriétés qui leurs
sont caractéristiques et qui les distinguent des appositions.
3. Objectif et corpus
L’objectif de cet article sera d’étudier les constituants disloqués à droite qui ont été relevés dans un corpus comprenant huit
comédies – sept de Plaute (Asin., Aul., Bacch., Cas., Cist., Curc.
et Epid.) et une de Térence (Eun.) –, le volume II de la
correspondance de Cicéron, le Satyricon de Pétrone,10 ainsi que
__________
10
Les comédies sont destinées à une présentation orale; les lettres de
Cicéron représentent des lettres privées rédigées par un homme érudit; le
Satyricon de Pétrone est une œuvre littéraire présentant des traits caractéristiques de la langue spontanée, en particulier dans les discours directs.
203
des exemples trouvés passim.11 L’enjeu principal est de proposer
une typologie fondée sur la nature du terme coréférentiel.
4. Constituants disloqués à droite
4.1. Constituants coréférentiels à un terme nominal
Les constituants disloqués à droite coréférentiels à un terme
nominal représentent une catégorie dont mon corpus a fourni
une dizaine d’exemples. Chez Plaute, on relève plusieurs occurrences de type (5), mettant en œuvre le mot argentum et un
certain nombre de minae. Le constituant disloqué à droite
spécifie le contenu de son terme coréférentiel, argentum. Il est
au même cas que le terme coréférentiel mais le genre et le
nombre sont différents.
(7) Memento promisisse te... mihi omne argentum redditum iri, /
minas triginta. (Plaut. Curc. 491–2)
«Souviens-toi que... tu as promis de me rendre tout l’argent, à
savoir trente mines.»
(8) Hercle te hau sinam emoriri, nisi mi argentum redditur, /
uiginti minae. (Plaut. Pseud. 1222–1223) (exemple cité par
Panhuis 1982: 85)
«Par Hercule, je ne te laisserai pas mourir si tu ne me rends
pas l’argent, mes vingt mines!»
Un exemple analogue se rencontre chez Pétrone (9); il ne
présente pas de référent «nominal» au sens stricte du terme mais
une proposition «relative nominale» (Lavency 1997: 268), quod
collocari non potuit, qui commute avec un nom, par exemple,
argentum.
__________
11
Cf., en particulier, Ch. Bennett (1914), W. Havers (1926) et J. B.
Hofmann (1985) sur les appositions et sur le nominatiuus pendens. Exception
faite des évaluations (section 4.3.), qui sont relativement fréquentes chez
Plaute, des appositions partitives de type (18) et les constituants disloqués à
droite signalés par des particules, par exemple quidem, tous les exemples que
j’ai rassemblés sont cités dans cet article.
ipabo_66.249.78.36
204
Olga Spevak
(9) Eodem die: in arcam relatum est quod collocari non potuit,
sestertium centies. (Petron. 53.4)
«Même jour: on a remis dans le coffre ce qu’on n’a pas pu
placer, dix millions de sesterces.»
Il est intéressant de signaler qu’en (10), minas quadraginta
n’est pas rejeté en fin de phrase; c’est un constituant disloqué à
droite parenthétique (cf. (e) et (f), cités en introduction), et non
pas une apposition. En effet, le même procédé qu’en (5), (7) et
(8) y est mis en œuvre: la spécification du contenu de argentum.
En outre, les deux termes ne présentent pas l’accord en genre et
en nombre.
(10) Emi atque argentum dedi, / minas quadraginta, ipsi in
manum. (Plaut. Trin. 125–126)
«Je la lui ai achetée, et je lui ai donné l’argent, quarante
mines, à lui-même, en mains propres.»
Les exemples précités permettent de saisir la première
catégorie de constituants disloqués à droite: coréférentiels avec
un terme énoncé au préalable, ils spécifient son contenu. Les
constituants disloqués ne sont pas nécessairement placés au
contact de leur terme coréférentiel (théoriquement, ils pourraient
l’être mais le plus souvent, ils ne le sont pas). Ils se rencontrent
à une place proéminente, en l’occurrence, en début de vers. Leur
référent est un nom générique (homo, augurium, argentum).
Voilà une différence remarquable par rapport aux appositions:
tandis que, dans le cas des appositions (tout au moins, les
appositions déterminatives), une équation peut être établie entre
les deux termes, telle Dionysius = seruus meus «Dionysius est
mon esclave», ou seruus meus = Dionysius «mon esclave est
Dionysius», dans le cas de constituants disloqués, il ne s’agit pas
d’équations: augurium ≠ sex uultures et argentum ≠ uiginti
minas. Certes, sex uultures et uiginti minae représentent, respectivement, un augure et une somme d’agent, mais augurium et
argentum ne se réduisent pas à sex uultures ou à uiginti minae.
Ce trait explique l’absence, éventuelle, de l’accord en genre et
en nombre.
205
En termes syntaxiques, dans l’exemple Dionysius, seruus
meus, les deux constituants peuvent, dans une phrase attributive,
assumer le rôle de sujet et d’attribut,12 tandis que uiginti minae
ne peut pas figurer en position d’attribut, comme le montre la
reconstruction des énoncés suivants:
(1f) Dionysius seruus meus est.
(1g) Seruus meus Dionysius est. (Seruus meus Dionysius
appellatur.)
(5a) Viginti minae argentum est.
(5b) ?Argentum uiginti minae sunt.
Il convient de noter que l’identité référentielle, qui se manifeste dans les exemples (1f) et (1g), n’exclut pas la dislocation à
droite; les exemples de dislocation à droite comme (1h) et (1i)
sont tout à fait envisageables:
(1h) Dionysius aufugit, seruus meus (scilicet).
(1i) Seruus meus aufugit, Dionysius (scilicet).
Seule l’intonation nous apporterait une preuve décisive sur leur
statut, celui de constituant disloqué ou celui d’apposition; face à
de tels cas, la question reste difficile à trancher. D’un autre côté,
il faut souligner que les exemples (1a) et (1d) se distinguent par
un référent spécifique et unique:
(1a) Dionysius, seruus meus aufugit.
(1d) Seruus meus, Dionysius, aufugit.
En effet, telle est la propriété référentielle des noms de parentés
et des noms qui expriment des personnes appartenant au proche
entourage ou à la familia: c’est le cas de seruus dans notre
exemple. En revanche, un nom générique comme auctor ou
hospes possède un référent moins accessible et appelle d’autant
plus une spécification. Ainsi, tout comme H. Dik (2007: 35), je
pencherai, dans le cas de l’occurrence (11), pour une interpréta__________
12
Cette équivalence semble se justifier pour les noms génériques
accompagnés d’un déterminant, comme le possessif ici: seruus meus.
ipabo_66.249.78.36
206
Olga Spevak
tion par constituant disloqué à droite; il est tiré d’un dialogue
tragique:
(11)        
    
     
. (Soph. El. 954–957)
«J’en suis sûre, tu n’hésiteras pas à t’unir à ta sœur pour tuer
l’auteur du meurtre de ton père, Égisthe.»
Le terme  «Égisthe» complète  
  «l’auteur du meurtre de ton père». Le référent
y est représenté par un syntagme nominal, le constituant disloqué est un nom propre, placé en début de vers. De même dans
mon corpus, un exemple emprunté à Plaute présente les mêmes
caractéristiques; le constituant disloqué s’y trouve en fin de vers:
(12) Continuo antiquom hospitem nostrum sibi / Mnesilochus
aduocauit, Pelagonem senem. / (Plaut. Bacch. 261–262)
«Aussitôt, Mnésiloque convoqua notre ancien hôte, le vieux
Pélagon.»
Si l’on admet que les constituants disloqués à droite sont,
typiquement, des spécifications a posteriori d’un terme énoncé
au préalable, dans le sens GENERIQUE > SPECIFIQUE, on peut
assigner un tel statut aussi aux exemples suivants, empruntés à
Cicéron. Dans les deux cas, il s’agit de spécifier un terme
général (Asia et dies) par un terme concret (Cyzicum et VIII Kal.
Feb.).
(13) Nobis iter est in Asiam, maxime Cyzicum. (Cic. Att. 3.6)
«J’ai l’intention d’aller en Asie, de préférence à Cyzique.»
(14) Venit tandem concilio de me agendi dies, VIII Kalendas
Februarias. (Cic. Sest. 75)
«Enfin, le huitième jour avant les Kalendes de février, mon
affaire vint à l’assemblée.»
207
En revanche, l’exemple suivant, relevé chez Pétrone, montre
le procédé inverse, allant du spécifique vers le générique;13
«l’émeraude verte» y est caractérisé de «verre précieux». C’est
surtout à propos de cas similaires que les grammaires (cf.
Ernout-Thomas 1953: 135) parlent d’appositions dépourvues
d’accord en genre et en nombre:14
(15) Smaragdum ad quam rem uiridem, pretiosum uitrum? (Petron.
55.6)
«À quoi destines-tu la verte émeraude, ce verre si précieux?»
À la différence des autres exemples cités dans cette section,
l’élément pretiosum uitrum qualifie son référent. Cependant, cet
exemple pourrait être considéré comme un cas de dislocation à
droite, comme une qualification a posteriori.
L’identification des constituants disloqués à droite est facilitée par la présence des particules comme et quidem (16),
uidelicet, scilicet (17) et d’autres similaires au sens de «à
savoir». En particulier, les additions à l’aide de et sont qualifiées
par H. Rosén (1990 et 2009: 413) d’«épitaxe» ou de «coordination assymétrique». Dans ces cas, un terme, additionnel, est
ajouté au moyen de et (au sens de «et ce») à une proposition
saturée sur le plan syntaxique et informationnel.15 Toutefois,
l’exemple (17), par la nature pronominale du référent, fait partie
de la catégorie décrite dans la section suivante (section 4.2).
(16) [Hic mecum Balbus, Hirtius, Pansa.] Modo uenit Octavius, et
quidem in proximam uillam Philippi. (Cic. Att. 14.11.2)
«[Balbus, Hirtius et Pansa sont ici avec moi.] Octave est arrivé
récemment et ce, dans la villa de Philippe, toute proche d’ici.»
(17) Sed erant permulti alii ex quibus id facillime scire posset,
omnes scilicet Lanuuini. (Cic. Mil. 46)
__________
13
C’est le seul exemple de ce type que mon corpus ait offert.
Cf. ne canes quidem, sollicitum animal (Liv. 5.47.3) «les chiens, ces
animaux attentifs».
15
Sur l’épitaxe, voir H. Rosén (2009: 413 et 2008).
14
ipabo_66.249.78.36
Olga Spevak
208
«Bien d’autres auraient pu le lui apprendre, c’est-à-dire tous
les habitants de Lanuvium.»
4.2. Constituants coréférentiels à un terme pronominal
La deuxième catégorie de constituants disloqués à droite,
offerts par mon corpus, est représentée par ceux qui sont coréférentiels à un terme pronominal. L’exemple (18) est emprunté à
Ch. Bennett (1914: 6–7) qui le cite, dans la section sur les appositions chez Plaute et Térence, au titre de partitive apposition:16
(18) Quia nos libertinae sumus, et ego et tua mater, ambae /
meretrices fuimus. (Plaut. Cist. 38–39)
«C’est parce que nous ne somme que des affranchies que, ta
mère et moi, nous avons fait le métier de courtisanes»
Un terme pronominal, nos, est précisé par et ego et tua mater.
On pourrait se demander comment interpréter cet exemple: comme une apposition? Ou comme un constituant disloqué à droite
et parenthétique? En effet, l’identité référentielle y est assurée et
le segment apposé rappelle plutôt une information connue. D’un
autre côté, il comporte les particules focalisantes, et... et..., qui
mettent en relief les deux éléments.
En revanche, le référent d’un pronom autre que personnel
(1ère ou 2e personne) n’est pas véhiculé par le pronom lui-même,
comme l’illustre l’exemple (19), avec le pronom utrumque.
L’élément apposé, et multiloqua et multibiba, ne découle pas du
contexte et apporte une information nouvelle.
(19) Vtrumque haec, et multiloqua et multibiba, est anus. (Plaut.
Cist. 149)
«La vieille possède deux qualités: elle parle beaucoup et elle
boit beaucoup.»
__________
16
Ch. Bennett (1914: 6–7) donne cinq exemples d’«apposition partitive»
pour la première et la deuxième personne et sept pour la troisième; sa liste
n’est pas exhaustive.
209
Il est moins aisé d’exemplifier les constituants disloqués à
droite dont le référent serait un pronom démonstratif (cf.
l’exemple (c) cité dans l’introduction). À ce titre, R. Amacker
(2001: 195) cite un exemple, tiré du De finibus de Cicéron (dans
un discours direct):
(20) Tum ille: ‘Finem’, inquit, ‘interrogandi, si uidetur, quod
quidem ego a principio ita me malle dixeram hoc ipsum
prouidens, dialecticas captiones’. (Cic. Fin. 2.17)
«Trêve aux questions, veux-tu, dit-il alors, je le préfère
comme je te l’avais dit en commençant: car voici justement ce
que je prévoyais, les pièges de la dialectique.»
Or, hoc ipsum n’y est pas anaphorique, comme dans (c), mais
cataphorique: il sert à annoncer dialecticas captiones. Pour cette
raison, l’interprétation de (20) comme relevant de la dislocation
à droite devrait être écartée. De même en (21), illud est
cataphorique et orationis genus ne peut pas être considéré
comme disloqué à droite.
(21) Vnum illud mihi uideris imitari, orationis genus. (Cic. Leg.
2.17)
«Il n’y a qu’un point par lequel tu me parais l’imiter, c’est le
style.»
En revanche, le pronom ille en (22) n’est pas cataphorique
mais anaphorique: il fait référence au savoir partagé entre
Cicéron et Atticus. Le constituant Pompei, nostri amici peut de
ce fait être pris pour une spécification a posteriori: c’est un
constituant disloqué.
(22) Illam manum tu mihi cura ut praestes, quoniam propius abes,
Pompei, nostri amici. (Cic. Att. 1.1.3) (exemple cité par
Amacker 2001: 195)
«Tâche de m’assurer, puisque tu es moins loin de lui que moi,
la forte troupe d’électeurs dont dispose notre ami Pompée.»
L’exemple (23), comportant le pronom eccos, illustre, lui
aussi, le phénomène de dislocation à droite. En tant que
ipabo_66.249.78.36
210
Olga Spevak
déictique, il réfère à une entité extralinguistique et est spécifié
par patrem sodalis et magistrum, un constituant nominal disloqué à droite. Ils fonctionnent ensemble pour former une sorte de
phrase présentative qui accompagne l’arrivée de nouveaux
personnages sur la scène.
(23) Sed eccos uideo incedere / patrem sodalis et magistrum; hinc
auscultabo quam rem agant. (Plaut. Bacch. 403)
«Mais les voici que je vois arriver le père de mon camarade et
son précepteur; je vais écouter, d’ici, de quoi ils parlent.»
En revanche, on pourrait hésiter sur le statut de L. Pontius en
(24); R. Amacker (2001: 196) cite cet exemple pour montrer un
nom propre disloqué à droite. En tout cas, is y appelle la relative
en quem, et non pas immédiatement L. Pontius;17 le nom propre
spécifie le référent de cette relative.
(24) Vna agebant ceteri creditores, in quibus erat <L.> Lucullus et
P. Scipio et is quem putabant magistrum fore si bona uenirent,
L. Pontius. (Cic. Att. 1.1.3)
«Les autres créanciers s’associent à la poursuite, entre autres
L. Lucullus et P. Scipion, et celui qui sera, pensent-ils, syndic
si les biens sont mis en vente, L. Pontius.»
En somme, les constituants coréférentiels à un pronom
personnel (nos) semblent fonctionner comme un rappel d’un
contenu plutôt qu’une spécification et se rapprochent, de ce fait,
des appositions. La coréférence à un quantifieur impliquant une
idée de pluralité (cf. uterque, multi...) favorise l’interprétation du
constituant comme une spécification a posteriori (constituant
disloqué à droite). Les exemples mettant en œuvre des pronoms
cataphoriques, tels (20) et (21), ne relèvent pas de la dislocation
à droite.
__________
17
À noter que cet exemple est différent de is quem quaero, uir sacerrumus, cité ci-dessous en (30); celui-là présente une spécification du référent,
celui-ci une évaluation subjective.
211
4.3. Constituants disloqués à droite exprimant une évaluation
subjective
Avant d’aborder l’analyse de ce type de constituants disloqués à droite, il convient de considérer le cas du «nominatif
proche d’exclamation», mentionné par A. Szantyr (1972: 28):
(25) Sequebatur raeda cum lenonibus, comites nequissimi. (Cic.
Phil. 2.58)
«Suivait un chariot avec des proxénètes, une escorte de
vauriens.»
Comites nequissimi «escorte de vauriens» porte sur raeda cum
lenonibus «chariot avec des proxénètes» mais à la différence du
type décrit dans la section 4.1., où les constituants disloqués à
droite portaient sur un nom, dans ce cas, nous n’avons pas seulement affaire à une spécification: il s’agit ici en outre d’un jugement fait par l’auteur à propos du contenu précédent: le chariot
avec des proxénètes est qualifié d’escorte de vauriens.
Ce trait, à savoir l’évaluation subjective, est caractéristique
des exemples suivants:
(26) Nunc neque quid uelim neque nolim facitis magni, pessumae.
(Plaut. Asin. 214)
«Maintenant, ce que je veux et ce que je ne veux pas vous
importe peu, méchantes!»
(27) Quam tu uirginem / me reposcis? – Quam ab lenone abduxti
hodie, scelus uiri. (Plaut. Curc. 614)
«Quelle fille me réclames-tu? – Celle que tu as prise chez le
marchand aujourd’hui, crapule.»
(28) ‘Tu autem’, inquit, ‘etiam tu rides, cepa cirrata? (Petron. 58.1)
«Alors, toi aussi tu ris, lui dit-il, espèce d’oignon frisé?»
(29) [uxor Trimalchionis] Est sicca, sobria, bonorum consiliorum
tantum auri uides; est tamen malae linguae, pica puluinaris.
(Petron. 37.7)
«[l’épouse de Trimalchion] Elle est sobre, tempérante, de bon
conseil, comme tu la vois, elle vaut son pesant d’or; mais elle
a une méchante langue, une véritable pie domestique.»
ipabo_66.249.78.36
212
Olga Spevak
Ces constituants disloqués à droite concernent l’interlocuteur
(26–28) – auquel cas il s’agit d’apostrophes – ou une tierce
personne (29). Seul l’exemple (28) manifeste une coréférence
explicite avec tu. Les évaluations, essentiellement négatives,18
énoncées par le locuteur, résultent du contenu de la phrase et
représentent une sorte de conclusion, par exemple en (26), «vous
y portez peu d’attention donc vous êtes méchantes», en (27),
l’interlocuteur rusé, par le fait d’avoir emmené la jeune fille en
question, mérite l’appellation de scelus uiri ou en (29), les
qualités de la dame en question portent à conclure qu’elle est
comme une pie. Ces évaluations, ajoutées à des phrases saturées
sur le plan syntaxique et informationnel, semblent avoir un
statut énonciatif particulier et représenter des «prédications
secondaires» (cf. Touratier 1994: 445).19
Il convient de mentionner encore deux exemples. Tout comme en (29), le jugement évaluatif n’y est pas directement adressé
à l’interlocuteur mais il apparaît dans un monologue. À la
différence des exemples précités, les termes coréférentiels y sont
exprimés: dans le premier cas, uir sacerrumus est coréférentiel à
is quem; dans le second, thensaurum meum à eccum lenonem.20
(30) Vtinam is sit quem ego quaero, uir sacerrumus. (Plaut. Rud.
158)
«Si seulement c’était l’homme que je cherche, cette crapule.»
(31) Sed eccum lenonem, incedit, thensaurum meum. (Plaut. Curc.
676)
«Mais voici le marchand de filles qui arrive, mon coffre-fort!»
Les constituants disloqués exemplifiés dans cette section ne
présentent pas systématiquement de coréférence explicite: dans
ces cas, le référent est implicite (abduxisti en 27) ou découle du
__________
18
Selon E. Dickey (2002: 369), pessimus est une expression de la colère,
scelus (uiri) une insulte très forte.
19
Cf. H. Fugier (1985: 112), sur le statut énonciatif des exclamations:
«L’exclamation injurieuse est explicitement prise en charge par le locuteur:
Ambitiose! ‘Je te dis prétentieux.’ C’est un jugement porté sur tu.»
20
Cf. ci-dessus, l’exemple (23). Cette expression, eccum lenonem, sert à
introduire un personnage sur la scène.
213
contexte, comme uxor Trimalchionis (29). L’absence de coréférence explicite n’infirme cependant pas l’interprétation de ces
constituants comme disloqués à droite. R. Geluykens (1987:
124; cf. également Fernandez-Vest 2004: 80)21 les considère
comme tels et donne à titre d’exemple:
(h) He did it to me again, the bastard.
4.4. Constituants disloqués à droite avec valeur résomptive
Si les grammaires latines traitent le phénomène qui nous
intéresse ici, elles le présentent dans des sections sur l’apposition. Ernout-Thomas (1953: 24) font une mention spéciale de
l’«apposition de phrase»,22 en l’illustrant par:
(32) Cuius hoc dicto admoneor, ut aliquid etiam de humatione et
sepultura dicendum existimem, rem non difficilem, iis
praesertim cognitis, quae de nihil sentiendo paulo ante dicta
sunt. (Cic. Tusc. 1.102)
«Ce trait me fait songer qu’il serait bon de dire un mot des
enterrements et des sépultures. La chose est sans difficulté,
surtout quand on est au courant de ce que nous avons dit tout à
l’heure relativement à l’absence de sentiment chez les morts.»
Le syntagme rem non difficilem, constitué d’un nom générique
et d’un adjectif qualificatif, résume le contenu de la phrase
précédente. En même temps, la résomption traduit l’opinion de
l’auteur: elle a le caractère d’une évaluation subjective qui est,
dans ce cas, davantage développée dans la proposition subséquente (iis praesertim cognitis...). Pour Ch. Touratier (1994:
445), il s’agit d’une sorte de «prédication secondaire».
Kühner-Stegmann (1914 I: 248), eux aussi, exemplifient ce
type d’«appositions», en ajoutant qu’elles peuvent apparaître au
__________
21
R. Geluykens (1987: 124) précise que ce type de constituants disloqués
à droite ne relève pas de la stratégie conversationnelle de «réparation» (cf. cidessus, § 1).
22
De manière similaire, A. Szantyr (1972: 429).
ipabo_66.249.78.36
214
Olga Spevak
nominatif aussi bien qu’à l’accusatif. Ils les qualifient d’«explications» ou de «jugements».
(33) Hoc enim identidem dicitis, non intellegere nos quam dicatis
uoluptatem. Rem uidelicet difficilem et obscuram. (Cic. Fin.
2.75)
«C’est, du reste, ce que vous ne cessez de dire. Selon vous,
nous ne comprenons pas ce que vous entendez par le plaisir.
Évidemment, c’est quelque chose de difficile et d’obscur.»
Les constituants disloqués à droite à valeur résomptive
rejoignent la catégorie précédente (section 4.3). Ils expriment,
eux aussi, une évaluation subjective mais leur portée est plus
large.
5. Bilan
Il convient à présent de résumer les caractéristiques principales des constituants disloqués à droite qui ont été examinés dans
la section 4:
- Coréférentiels à un nom, les constituants disloqués à droite
fournissent une spécification de ce terme énoncé au préalable.
Leurs termes coréférentiels sont, tout au moins dans un certain
nombre de cas, saillants sur le plan informationnel. Les
particules telles et quidem «et ce» sont un signal explicite de ce
procédé de spécification.
- Coréférentiels à un pronom, ils fournissent le contenu exact de
leur référent pronominal; ce dernier est pragmatiquement saillant.
- Les constituants disloqués à droite peuvent exprimer des
évaluations subjectives de l’auteur à propos de l’interlocuteur,
d’une tierce personne ou de la situation. Ce type de dislocation a
souvent un référent implicite ou découlant du contexte et n’est
pas lié à la saillance pragmatique.
- Les constituants disloqués à droite peuvent servir à résumer le
contenu d’un énoncé précédent.
215
- Les constituants disloqués à droite peuvent être placés au
contact de leurs référents mais le plus souvent, ils s’en tiennent à
distance.
6. Les phrases «à rallonges»
Aux constituants disloqués à droite que je viens de décrire, on
pourrait ajouter encore un type particulier: les «phrases à
rallonges», décrites par D. Longrée (1991 et 1996). Elles se
rencontrent tout particulièrement chez les historiens latins et il
s’agit de propositions participiales (essentiellement des ablatifs
absolus), ajoutées comme une information complémentaire au
contenu énoncé au préalable. De telles propositions précisent
une circonstance du procès décrit dans la proposition principale
(34). Elles peuvent également se rapporter à un nom (ou à un
syntagme nominal), exprimé dans le contexte précédent; la
«rallonge» en (35) fournit une information sur legiones.
(34) Germanicus... expeditum exercitum in Chattos rapit, L.
Apronio ad munitiones uiarum et fluminum relicto. (Tac. Ann.
1.56.1)
«(Germanicus)... fond sur les Chattes avec une armée sans
bagages, L. Apronius ayant été laissé en arrière à l’aménagement des routes et des fleuves.»
(35) (Germanicus... legiones...) obuias extra castra habuit, deiectis
in terram oculis uelut paenitentia. (Tac. Ann. 1.34.1)
«(Germanicus...) trouva les légions en dehors du camp, venant
à sa rencontre, les yeux baissés vers la terre comme par
repentir.»
7. Phénomènes apparentés qui ne sont pas à analyser
comme des dislocations à droite
Une fois saisies les propriétés des constituants disloqués à
droite, il convient de procéder à une confrontation avec un autre
phénomène, tout à fait distinct: le placement d’un constituant en
ipabo_66.249.78.36
Olga Spevak
216
fin de phrase. En effet, le constituant qui se rencontre en fin de
phrase – même s’il est précédé d’une virgule – n’est pas nécessairement à interpréter comme un constituant disloqué à droite.
7.1. Les sujets en fin de phrase
Récemment, R. Amacker (2001: 196 sq.) a essayé de définir
les constituants disloqués à droite à partir d’une hypothèse sur la
structure prosodique de la phrase latine. Cela le conduit à interpréter certains sujets comme disloqués à droite, par exemple:
(36) [Caelius... studuit Catilinae iterum petenti...] At23 studuit
Catilinae, cum iam aliquot annos esset in foro, Caelius; et
multi hoc idem ex omni ordine atque ex omni aetate fecerunt.
(Cic. Cael. 12)
«[Caelius... on le vit s’attacher à Catilina, candidat pour la
seconde fois...] Caelius, après plusieurs années passées dans le
forum, s’attacha à Catilina; bien d’autres, de tout état et de
tout âge, ont fait comme lui.»
Cicéron a déjà dit, dans le paragraphe précédent, que Caelius
soutenait Catilina; il répète la même affirmation avec plus
d’insistance. Or, Caelius n’était pas le seul à accorder son
soutien à Catilina et c’est précisément cette information que
Cicéron veut communiquer ici. Caelius est, dans un sens,
confronté à multi «beaucoup d’autres» (je préférerais une
virgule après Caelius) et entre en contraste avec eux. En conséquence, Caelius n’est pas un constituant disloqué à droite.
L’exemple (37) est encore plus explicite. Cicéron plaide pour
Sextus Roscius, accusé d’avoir tué son père, qui porte le même
nom. Un parent, Titus Roscius, est également impliqué dans
l’affaire:
(37) Nam cum hic Sextus Roscius esset Ameriae, T. autem iste
Roscius Romae, cum hic filius assiduus in praediis esset...,
__________
23
At est une leçon minoritaire; la plupart des manuscrits l’omettent.
217
iste autem frequens Romae esset, occiditur ad balneas Pallacinas rediens a cena Sex. Roscius. (Cic. S. Rosc. 18)
«En effet, alors que Sextus Roscius (fils) se trouvait à Amérie,
ce Titus Roscius était à Rome; et alors que le fils Roscius ne
quittait pas les propriétés..., on voyait celui-ci (Titus) partout
dans Rome; et voilà que Sextus Roscius (père) est tué, près
des bains de Pallacine, de retour d’un dîner.»
Dans le contexte précédent, Cicéron a parlé de deux Roscii (Sex.
Roscius fils et Titus Roscius); le sujet de occiditur (Sex. Roscius
père) doit alors être exprimé dans cette phrase parce que les
conditions permettant le maintien du sujet ne sont pas remplies.
La phrase complexe s’ouvre par un cadre temporel (deux subordonnées temporelles en cum) et la principale répond à la
question «que s’est-il passé?». Dans une telle configuration, il
n’est pas inhabituel en latin d’avoir le verbe en tête et le sujet en
position finale. Si la principale commençait par Sex. Roscius,
elle fournirait une information sur ce qui est arrivé à Sex.
Roscius. En outre, j’interpréterais cette phrase comme répondant
à la question secondaire «qui a-t-il été tué?», avec Sex. Roscius
comme élément saillant. L’analyse de cet exemple comme
illustrant un constituant disloqué à droite est, à mes yeux, incorrecte. On notera au passage que ni (36), ni (37) ne présentent de
coréférence, contrairement à ce qu’on a vu au 4.1.
On se gardera d’analyser comme disloqués à droite les
constituants qui se retrouvent en fin de phrase – que ce soient
des sujets, des objets ou des circonstants, par exemple:
(38) Vbi ubi erit, iam inuestigabo et mecum ad te adducam simul /
Pleusidippum. (Plaut. Rud. 1210)
«Où qu’il soit, je le chercherai et je te l’amènerai, ton
Pleusidippe.»
(39) Reliqua exspectabam, ut tibi placebat, Thessalonicae. (Cic.
Att. 3.10)
«J’attends la suite, selon tes avis, à Thessalonique.»
(40) Quamquam nec in hac oratione spes est posita causae, nec in
eorum studiis qui a te pro Ligario petunt tui necessarii. (Cic.
Lig. 31)
ipabo_66.249.78.36
218
Olga Spevak
«Mais ce n’est ni sur ma plaidoirie que repose l’espoir de
notre cause ni dans les efforts de ceux qui te demandent grâce
pour Ligarius et qui sont tes amis.»
De tels constituants ne présentent pas les caractéristiques
typiques des constituants disloqués à droite: la coréférence, la
spécification d’un terme, l’évaluation subjective. Dans les trois
exemples précités, il s’agit d’un objet direct (Pleusidippum),
d’un complément de lieu (Thessalonicae) – on n’a aucune raison
de les considérer comme ne faisant pas partie syntaxique de la
phrase – et d’un constituant prédicatif tui necessarii «en tant que
tes amis». Sur le plan informationnel, ces constituants sont
saillants.
7.2. Les syntagmes nominaux disjoints
La dislocation à droite ne devrait pas être confondue non plus
avec la disjonction du syntagme nominal lorsque l’une de ses
composantes est rejetée en fin de phrase. Contrairement à R.
Amacker (2001: 195–196), j’interprète les exemples (41) et (42)
comme une disjonction. Il s’agit des syntagmes hoc de magistro
et aedes tuae; le constituant occupant la place finale d’une
phrase n’est pas eo ipso disloqué à droite.
(41) Verum hoc ridiculum est de magistro. (Cic. Att. 1.1.3)
«Mais cette histoire de syndic est ridicule.»
(42) Tum mihi aedes quoque arridebant, quom ad te ueniebam,
tuae. (Plaut. Asin. 207)
«En ce temps-là, ta maison elle-même était tout sourire,
lorsque je venais te voir.»
8. Conclusions
Les constituants disloqués à droite (Tails) ont pour fonction
essentielle de préciser un terme énoncé au préalable. Ces constituants, généralement décrits à propos des langues modernes
comme des constituants typiques de la langue parlée spontanée,
219
se rencontrent aussi en latin. Or, d’une part, on ne devrait pas
établir d’équation entre «constituants disloqués à droite» et «bas
niveau de langue»: il ne s’agit pas uniquement d’un phénomène
«décadent», témoignant d’un niveau de langue peu soigné,
comme l’exemple (g) pourrait le faire penser. Dans des contextes pragmatiques favorables, les constituants disloqués à droite
trouvent une pleine justification. D’autre part, ces constituants
sont relativement rares en latin, non pas seulement dans les
textes prosaïques hautement stylisés mais aussi dans les comédies de Plaute et de Térence, dans la correspondance de Cicéron
et dans le Satyricon de Pétrone. Pour les besoins de mon
analyse, j’ai isolé quatre catégories: constituants coréférentiels à
un nom, constituants coréférentiels à un pronom, jugements
évaluatifs et résomptions, afin de dégager les caractéristiques
principales (coréférence, spécification, évaluation) de ces constituants. Toutefois, faute de pouvoir tenir pour pertinent le critère
intonatif, il est difficile de trancher, dans certains cas, entre
apposition et constituant disloqué à droite. En outre, j’ai proposé
quelques critères d’identification pour ne pas confondre les
constituants disloqués à droite avec les sujets ou d’autres constituants placés en fin de phrase ou encore, avec les syntagmes
nominaux disjoints dont un des éléments est rejeté en position
finale.
Bibliographie
Amacker, R. (2001): «Indices de thématisation segmentée non initiale chez
Cicéron» dans C. Moussy (éd.), De lingua Latina novae quaestiones.
Actes du Xe Colloque international de linguistique latine, Paris-Sèvres,
19–23 avril 1999, Louvain-Paris, Peeters, 185–200.
Bennett, Ch. E. (1914): Syntax of Early Latin, vol. II, The Cases, LeipzigBoston, Theodor Stauffer, Allyn et Bacon.
Cabrillana, L. C. (2009): «Nominativo y Vocativo» dans J. M. Baños Baños
(dir.), Sintaxis del latín clásico, Madrid, Liceus, 111–130.
Dickey, E. (2002): Latin Forms of Address: From Plautus to Apuleius,
Oxford, Oxford University Press.
ipabo_66.249.78.36
220
Olga Spevak
Dik, H. (2007): Word Order in Greek Tragic Dialogue, Oxford, Oxford
University Press.
Dik, S. C. (1997): The Theory of Functional Grammar, I–II, 2ème édition
revue, Berlin-New York, M. de Gruyter.
Ernout, A., Thomas, F. (1953): Syntaxe latine, Paris, Klincksieck.
Fernandez-Vest, M. M. J. (2004): «Mnémème, Antitopic: le post-Rhème, de
l’énoncé au texte», dans M. M. J. Fernandez-Vest et S. Carter-Thomas
(éds.), Structure informationnelle et particules énonciatives. Essai de
typologie, Paris, Harmattan, 63–93.
– (2008): «Vers une typologie informationnelle des constructions à détachement», communication présentée au Centre Alfred Ernout en février
2008.
Fugier, H. (1983): «Le syntagme nominal en latin classique», dans W. Haase
(éd.), Aufstieg und Niedergang der römischen Welt 2, 29, 1, Berlin-New
York, W. de Gruyter, 212–269.
– (1985): «Le vocatif dans la phrase latine», dans Ch. Touratier (éd.),
Syntaxe et latin, Aix-en-Provence, Presses universitaires, 105–120.
Geluykens, R. (1987): «Tails (right-dislocation) as a repair mechanism in
English conversation», dans J. Nuyts et G. de Schutter, Getting One’s
Words into Line, Dordrecht, Foris, 119–129.
Havers, W. (1926): «Der sog. ‘Nominativus pendens’», Indogermanische
Forschungen 43, 207–257.
Heberlein, F. (1996): «Über ‘enge’ Appositionen», dans H. Rosén (éd.),
Aspects of Latin. Papers from the Seventh International Colloquium on
Latin Linguistics, Jerusalem, April 1993, Innsbruck, 343–359.
Hofmann, J. B. (1985): La lingua d’uso latina [trad. it. de Lateinische
Umgangssprache, 19513 par L. Ricottilli], Bologna, Pàtron.
Lambrecht, K. (1994): Information Structure and Sentence Form. Topic,
Focus, and Mental Representations of Discourse Referents, Cambridge,
Lavency, M. (19972): VSVS. Grammaire latine, Louvain-la-Neuve, Peeters.
Longrée, D. (1990): «À propos du concept d’‘apposition’», Information
grammaticale 45, 8–13.
‒ (1991): «La phrase à rallonge chez Tacite», dans W. Haase (éd.), Aufstieg
und Niedergang der Römischen Welt, II, 33, 4, Berlin-New York, 2539–
2580.
– (1996): «La concurrence entre ‘rallonge’ et ‘parataxe’ dans l’œuvre de
Tacite: conditionnements linguistiques, choix stylistiques», dans H.
Rosén, Aspects of Latin, Papers from the Seventh International Colloquium on Latin Linguistics, Jerusalem, 19–23 avril 1993, Innsbruck,
Innsbrucker Beiträge zur Sprachwissenschaft, 483–496.
221
Panhuis, D. G. J. (1982): The Communicative Perspective in the Sentence. A
Study of Latin Word Order, Amsterdam, Benjamins.
Pinkster, H. (1995): Sintaxis y semántica del latín [édition révisée et augmentée de Latin Syntax and Semantics, 1990], Madrid, Ed. Clásicas.
Rosén, H. (1990): «La coordination asymétrique comme critère de fonction
syntaxique en latin», Information grammaticale 46, 34–37.
– (2008): «Latin epitaxis in historical and typological view», dans G.
Calboli (éd.), Papers on Grammar X, Roma, Herder, 205–242.
– (2009): «Coherence, sentence modifiation, and sentence-part modification
– the contribution of particles», dans Ph. Baldi et P. Cuzzolin, New
Perspectives on Historical Latin Syntax, I. Syntax of the Sentence, BerlinNew York, Mouton de Gruyter, 317–441.
Serbat, G. (1996): L’emploi des cas en latin. I. Nominatif, vocatif, accusatif,
génitif, datif, Grammaire fondamentale du latin, tome VI, Louvain-Paris,
Peeters.
Spevak, O. (2010): Constituent Order in Classical Latin Prose, Amsterdam,
Benjamins.
Szantyr, A. (19722): Lateinische Syntax und Stilistik mit dem allgemeinen
Teil der lateinischen Grammatik, Munich, C. K. Beck.
Toth, A. (1994): Thema, Topik und Koda im Lateinischen. Zu einigen syntaktischen, semantischen und pragmatischen Problemen der lateinischen
Grammatik, dans G. Calboli (éd.), Papers on Grammar IV, Bologna,
CLUEB, 177–210.
Touratier, Ch. (1994): Syntaxe latine, Louvain-la-Neuve, Peeters.
ipabo_66.249.78.36
Πάλαι / τὸ πάλαι, πρίν / τὸ πρίν, παραχρῆµα /
τὸ παραχρῆµα
By JESÚS DE LA VILLA, Madrid
Abstract: In this paper, following the systematic exploration of the
differences between πρότερον / τὸ πρότερον carried out by Rijksbaron (2006), I will consider the occurrences of three temporal
adverbs when used with and without an article, namely πάλαι, πρίν,
and παραχρῆµα. The main result of the investigation is that the use of
the article triggers that these three adverbs refer to a complete and
delimited period, the past (τὸ πάλαι), the whole precedent period (τὸ
πρίν) or the immediate following time (τὸ παραχρῆµα). Apart from
that, there seems to be a clear tendency to use the forms with article
where there is some kind of contrast or focalisation on the temporal
reference given by the adverb.
1. Introduction*
Until an article by A. Rijksbaron (2006), the differences
between the use of certain temporal adverbs with and without
articles – in pairs such as νῦν / τὸ νῦν, πρότερον / τὸ πρότερον, παραχρῆµα / τὸ πραχρῆµα – had not been the subject of
any detailed study. In parallel, translations of Ancient Greek
texts into modern languages have not reflected any possible
distinctions between both series of elements.
__________
*
The research for this paper has been carried out with the financial
support of the Spanish Ministry of Education through the research project
HUM2005–06622–C04–01. A first version of it was presented in Katwijk,
The Netherlands, on december 2005, during a conference to honour Prof.
Albert Rijksbaron with the occasion of his retirement. I want to thank the
observations and remarks to the oral presentation of the paper made by
several participants during the Katwijk conference. Dr. M. E. Torrego also
read a previous written version of the text and offered me valuable commentaries. The revision of the English version has been done by Ms. H. Ferguson.
Glotta 89, 222–241, ISSN 0017–1298
Πάλαι / τὸ πάλαι, πρίν / τὸ πρίν, παραχρῆµα / τὸ παραχρῆµα
223
Before Rijksbaron’s contribution, we find only a few mentions of the meaning of the forms with an article. For example,
in the whole manual of Kühner-Gerth (18983: 595) the only
remark is the following: <Es gibt> viele adverbale Ausdrücke
mit dem Neutrum des Artikels, wenn der ganze Umfang einer
Zeitperiode bezeichnet werden soll (“There are many adverbial
expressions with the neuter of the article, when the whole
duration of a temporal period is referred to”). In a very different
interpretation, Cooper (1998: 410) proposes the following: “The
article shows a contrast between one time period and another or
others. So τὸ νῦν constrast the present, the currently true
situation, with the past or the future or both”.
It seems, then, that there are at least two possible justifications for the use of the neuter form of the article with adverbs: a
semantic one – the reference to a whole period instead to a
certain moment – and a pragmatic one – contrast between
different periods.
In this context, Rijksbaron’s work is important, because it is
the first systematic study of the use of one adverb, namely
πρότερον / τὸ πρότερον, in a large corpus of classical and
postclassical authors. His conclusions seem to confirm, at least
partially, Kühner-Gerth’s proposal. Moreover, he has suggested
that in certain conditions, and only for the form with an article,
the combination with different aspectual stems of the the verb
yields differences in the meaning of the adverb. His proposal
can be summarized in the following schema.
(1) πρότερον | τὸ πρότερον according to Rijksbaron (2006)
i) Without an article: ‘before, earlier’
ii) With an article: a) with a present form: ‘during the preceding
period’
b) with an aorist form: ‘the previous time, the last time’
Apparently, then, the expression with an article actually
refers to a period, but only when combined with a present form
of the verb.
ipabo_66.249.78.36
Jesús de la Villa
224
In this paper, I want to follow the systematic exploration of
the data of the uses of adverbs with an article. I will consider the
occurrences of three temporal adverbs when used with and
without an article, namely πάλαι, πρίν, and παραχρῆµα. The
main questions I want to answer are the following:
i) Does the presence of the article change the interpretation of
these adverbs?
ii) If so, is the article used when a whole period is meant?
iii) Are aspectual stems also relevant for the interpretation of the
different uses of those adverbs?
iv) Is there any evidence of a pragmatic or contrastive use of the
article, as proposed by Cooper (1998)?
For methodological reasons, I have restricted myself to the
data of some authors where both possibilities for those adverbs –
with and without an article – occur. My corpus contains the texts
of Herodotus, Thucydides, and Plato. Additionally, examples
from other authors are introduced to support my argument.
In the following sections I will first consider the data for each
adverb separately (§§ 2–4). Then, I will compare the partial
results, looking for a patron of relevancy for the article when
used with adverbs. Finally, I will draw some conclusions (§ 5).
2. πάλαι | τὸ πάλαι
The distribution of the two possibilities, with and without an
article, in the three authors I am considering is as follows:
Distribution:
Hdt. Πάλαι 12x / τὸ πάλαι
Th. Πάλαι
7x / τὸ πάλαι
Pl. Πάλαι 123x / τὸ πάλαι
1
12x
2
8x
3
3x
__________
1
See Powell (1938: s.u.). I exclude the attributive uses of the adverb of
the type οἱ πάλαι.
2
Πάλαι: Th. 1.2.1, 1.15.3, 1.39.3, 2.54.2, 3.13.1, 3.39.5, 6.10.5, 7.75.5.
τὸ πάλαι: 1.5.1, 1.6.5, 1.13.5, 2.15.5, 3.45.3, 3.104.3, 3.104.5, 4.42.2,
3
The three examples of τὸ πάλαι are: Phdr. 251b, R. 573b, Lg. 693e.
225
In many dictionaries and glossaries two meanings for this
adverb are proposed, one temporally relative: it implies that the
event lasts from a previous time until a certain moment; it
corresponds to English “from old” (Powell 1960), and French
despuis longtemps (Bailly). In other cases, it expresses an
absolute temporality referring to the past; the proposed translation is “formerly” (Powell), autrefois, jadis (Bailly). The Oxford
Greek-English Lexicon of Lidell-Scott-Jones (LSJ) proposes a
single translation for all the cases: “long ago”. Neither LSJ nor
Bailly make any distinction between the uses of πάλαι and of
τὸ πάλαι. Powell apparently keeps the relative meaning “from
old” only for πάλαι, whereas the absolute meaning would be
shared by the forms with and without an article.
A detailed consideration of the data in my corpus offers the
following results:
1) Contrary to Powell’s proposal, τὸ πάλαι can also express the
relative meaning “from old”, as in the example (2).4 There-fore,
the presence of the article is not the feature that can distinguish
between the relative and absolute uses of πάλαι.
(2) Ἄλλως τε τούτων ἀπεόντων αἰτίους γενέσϑαι δουλοσύνης
τοῖσι Ἕλλησι Ἀϑηναίους οὐδαµῶς ἀνασχετόν, οἵτινες αἰεὶ
καὶ τὸ πάλαι φαίνεσϑε πολλοὺς ἐλευϑερώσαντες (Hdt.
8,142,3)
“Apart from that, it is unbearable that not all this alone but slavery too should be brought upon the Greeks by you Athenians,
who have always been known as givers of freedom to many”
2) The relative meaning “from old” is always and only
associated with tenses that express present time, this is the present indicative, as in (2) and (3), or the perfect indicative, as in
(4).
__________
4
Translations are always taken from or based on those of the Loeb Classical Library. They are only given as a support for the reading of the examples;
they do not necessarily agree with the actual interpretation of the adverbs
proposed in this paper.
ipabo_66.249.78.36
226
Jesús de la Villa
(3a) τὸν δὲ αὐτῷ ὑποκρίνεσϑαι ὡς ταῦτα πάλαι παρασκευάζεται ποιέειν (Hdt. 2,162,4)
“He answered that he had long been preparing to do just that”.
(3b) ἀλλὰ καὶ σοῦ πάλαι ϑαυµάζω αἰσϑανόµενος ὡς ἡδέως
καϑεύδεις (Pl. Cri. 43b)
“I have been wondering at for you for some time, seeing how
sweetly you sleep”.
(4) πάλαι τὰ καλὰ ἀνϑρώποισι ἐξεύρηται … (Hdt. 1,8,4)
“Men have long ago made wise rules ...”.
Nevertheless, it is important to notice, that the meaning “from
old” seems to be related to the temporal, and not to the aspectual, content of the verb. Actually, both πάλαι (5) and τὸ πάλαι
(6), are associated with the imperfect, that is, a past tense of
imperfective aspectual content, but in these cases does not yield
the sense “from old”, but better “formerly”.
(5) ἐτίϑεσαν τὸ πάλαι τρίποδας χαλκέους (Hdt. 1,144,2)
“In the past, they offered bronze tripods”
(6) Λέγω δὲ ὅτι Mαρδονίῳ τε καὶ τῇ στρατιῇ τὰ σφάγια οὐ
δύναται καταϑύµια γενέσϑαι· πάλαι γὰρ ἐµάχεσϑε· νῦν δὲ
οἱ δέδοκται τὰ µὲν σφάγια ἐᾶν χαίρειν … (Hdt. 9,45,2)
“I tell you, then, that Mardonius and his army cannot get omens
to his liking from the sacrifices. Otherwise you would have
fought long before this. Now, however, it is his purpose to pay no
heed to the sacrifices...”
Probably, then, the sense “from old” can be considered simply a
contextual interpretation triggered by the apparent incompatibility between a present time expressed by the verb and the past
meaning of the adverb. The contact between past and present is
established through the sense “from old <until now>”, that connects both temporal periods.
3) When used with present or perfect stem, as in (2)–(6) above,
and (7), both πάλαι and τὸ πάλαι always refer to a period, not
to a single point in the past. In contrast, when used with an aorist
227
stem, this sense is not implied, as in (8a) above, or it is directly
excluded, as in (8b).
(7) ἐκαλέοντο δὲ πάλαι πρὸς πάντων Ἄριοι (Hdt. 7,62,1)
“They were formerly called Arians by everyone”.
(8a) οὓς χρῆν … πάλαι κοινώσαντας τὴν δύναµιν κοινὰ καὶ τὰ
ἀποβαίνοντα ἔχειν (Th. 1,39,3)
“They should have shared their power with you before, and so
they could also share now the consequences of their acts”.
(8b) εἰ ἐγὼ πάλαι ἐπεχείρησα πράττειν τὰ πολιτικὰ πράγµατα,
πάλαι ἂν ἀπολώλη (Pl. Ap. 31 d)
“If I had undertaken to go into politics, I should have been put to
death long ago”
This difference in meaning is obviously related to the aspectual
content of the verb: the imperfective value of the present stem
triggers a durative and, in principle, unlimited extension of the
event, whereas the perfective meaning of the aorist implies that
the event – either punctual, repeated or durative – finally ended.5
This result is apparently in accordance with Rijksbaron’s proposal for the use of τὸ πρότερον. Nevertheless, the data for
πάλαι show that here it is not related to the presence or absence
of the article; both constructions, with and without the article –
compare (2) and (3), for example –, behave in the same way.
4) The fact that τὸ πάλαι, at least with present and perfect stem,
always refers to the whole past seems connected with the use of
the article, as in (2), (5) above, and (9), whereas πάλαι may
refer to the past as a whole, as in (7) above, and (10),6 or may
not, as in (3)–(4) above, and (11).
__________
5
About the basic meanings of the aspectual stems of the Greek verb see,
for example Rijsbaron (20033). For the implications of this explanation of
aspectual system in other syntactic and semantic phenomena in Ancient
Greek, see (Villa 2004).
6
It is interesting that in at least one example, namely Hdt. 9,46,3, we find
the combination πάλαι ἀπ᾽ ἀρχῆς + aorist, which may be interpreted as a
sign that πάλαι by itself could not refer to the whole past, but only
contextually.
ipabo_66.249.78.36
228
Jesús de la Villa
(9) Ἀλλὰ οὔτε Aἰγυπτίους δοκέω ἅµα τῶ/ ∆έλτα τῶ/ ὑπὸ
Ἰώνων καλεοµένῳ Aἰγύπτῳ γενέσϑαι αἰεί τε εἶναι ἐξ οὗ
ἀνϑρώπων γένος ἐγένετο, προιούσης δὲ τῆς χώρης
πολλοὺς µὲν τοὺς ὑπολειποµένους αὐτῶν γενέσϑαι,
πολλοὺς δὲ τοὺς ὑποκαταβαίνοντας. τὸ δ j ὦν πάλαι αἱ
Θῆβαι Aἴγυπτος ἐκαλέετο, ... (Hdt. 2,15,3)
“I maintain, rather, that the Egyptians did not come into existence
together with what the Ionians call the Delta, but have existed
since the human race came into being; and as the land grew in
extent, there were many of them who stayed behind, and many
who spread down over it. Be that as it may, the Theban district
was in the past called Egypt”.
(10) φαίνεται γὰρ ἡ νῦν Ἑλλὰς καλουµένη οὐ πάλαι βεβαίως
οἰκουµένη (Th. 1,2,1)
“It is evident that the country now called Hellas had in ancient
times no settled population”
(11) Oὐκοῦν καὶ τόδε πάλαι ἐλέγοµεν, ὅτι ἡ ψυχή, ὅταν µὲν τῶ/
σώµατι προσχρῆται εἰς τὸ σκοπεῖν τι ... (Pl. Phaedo 79 c)
“Now we have also been saying for a long time, have we not,
that, when the soul makes use of the body for any inquiry,...”
When used with aorist, τὸ πάλαι also implies the whole past. In
this case the aorist should receive a complexive intepretation, as
in (12).
(12a) οἱ γὰρ Ἕλληνες τὸ πάλαι ..., ἐπειδὴ ἤρξαντο µᾶλλον
περαιοῦσϑαι ναυσὶν ἐπ’ ἀλλήλους, ἐτράποντο πρὸς
λῃστείαν (Th. 1,5,1)
“For in early times the Hellenes ..., as communication by sea
became more common, were tempted to turn into pirates.”
(12b) ... ὅσαι ἤδη ἔξοδοι κοιναὶ ἐγένοντο Πελοποννησίοισι καὶ
τὸ πάλαι καὶ τὸ νέον (Hdt. 9,26,2)
“...in all campaigns of the united Peloponnesian armies, both
ancient and recent,...”
These data basically confirm the proposals by Kühner-Gerth
and Rijksbaron, but they also force us to introduce precisions:
First, πάλαι can also refer to the whole past. In this sense, it
implies that the form without an article is the unmarked member
229
of the opposition, as it can refer both to the whole past or not,
whereas the form with an article always implies the whole past.
A precision affects the distribution of the construction with an
article: at least for τὸ πάλαι, it cannot be confirmed that, as
proposed for τὸ πρότερον, there is any difference when used
with present or with aorist. In both distributions τὸ πάλαι refers
to the whole past.
5) Finally, concerning the possible pragmatic or contrastive use
of the article, it can be confirmed that τὸ πάλαι, as proposed by
Cooper, is also very frequently used when a contrast between
past and present should be stressed, as in (13), but it is not
always so, as in the examples (2), (9) or (12a) above. Πάλαι,
apparently, is not used in such contexts.
(13a) τὸ δὲ πάλαι ... διαζώµατα ἔχοντες περὶ τὰ αἰδοῖα οἱ
ἀϑλεταὶ ἠγώνιζοντο, καὶ οὐ πολλὰ ἔτη ἐπειδὴ πέπαυται.
(Th. 1,6,5)
“Formerly ... the athletes who contended wore belts across their
middles; and it is but a few years since that the practice ceased”
(13b) Ἀϑηναῖοι δὲ καὶ Πέρσαι τὸ µὲν πάλαι οὕτω πως, τὸ δὲ
νῦν δὲ ἦττον (Pl. Lg. 693 d)
“So it was for the Athenians and Persians in old times, in contrast
to their present condition”
To summarize, particular temporal or aspectual contexts seem to
trigger different interpretations for this adverb, but this appears to
happen irrespective of the use of the adverb with or without an
article. On the other hand, the use of the article in τὸ πάλαι,
first, always introduces a form of determination, that is, the
reference to a whole period, and, second, it is only the form with
an article that is used in contrastive, viz. pragmatically focalised,
contexts.
ipabo_66.249.78.36
230
Jesús de la Villa
3. πρίν, τὸ πρίν
Among the authors in my corpus, Thucydides and Plato do not
use τὸ πρίν. Only Herodotus uses both the form with an article
(7x) and without an article (3x).7 Therefore, I will concentrate in
the data from Herodotus.
The dictionary of LSJ proposes two different meanings for
this adverb. One is temporally relative: it establishes a sequence
between two events irrespective of the moment when those
events occur; the translation into English would be “before”.
The other meaning is temporally absolute; this means that it is
related to the present moment of the speaker and always referred
to the past; its meaning in English would be “formerly”, “once”.
In this dictionary, however, this difference is not associated with
the presence or absence of the article.
In contrast, Bailly apparently associates a distinction in
meaning with the presence of the article: πρίν auparavant,
avant “before” / τὸ πρίν dans le passé, autrefois “in the past”,
“once”.
Neither in LSJ nor in Bailly’s dictionary is anything said
about the reference to the whole past or to the eventual contrastive use of the article. Moreover, the distribution proposed
by Bailly, associated with relative / absolute temporality, does
not seem to be connected to the use of the article. Actually, if we
carefully consider the data of my corpus, we can discover the
following facts:
1) The adverb πρίν, when used without an article, actually
corresponds to the relative meaning “before”. This relative
meaning, however, is also clearly present in uses of τὸ πρίν,
such as (14) and (15). This clearly contradicts the distribution
suggested by Bailly.
__________
7
Instances of πρίν are: Hdt. 5.29.2, 5.30.2, 5.58.1, 5.66.1, 7.1.1, 8.37.2,
9.109.3. Of τὸ πρίν: Hdt. 1.129,4, 2.108.2, 2.172.2.
231
(14) ἐποίευν ... Aἴγυπτον, τὸ πρὶν ἐοῦσαν ἱππασίµην καὶ
ἁµαξευοµένην πᾶσαν, ἐνδέα τούτων (Hdt. 2,108,2)
“and they made Egypt, that had been a land of horses and carts,
empty of them”
(15) ἅτε δὴ δηµότην τὸ πρίν ἐόντα καὶ οἰκίης οὐκ ἐπιφανέος·
µετὰ δὲ ... (Hdt. 2,172,2)
“...on the ground that he was a common man and of no high
family; but presently ...”
Therefore, the presence or absence of the article does not
seem to be related to a difference in its meaning that can follow
an oppositional relative / absolute temporal reference.
2) Both πρίν and τὸ πρίν associated with a present (16) or
perfect (17) stem always refer to a period of time.8 In contrast,
when used with an aorist, at least in the sole example we have of
this combination, it refers to a fact of undetermined duration
(18).
(16a) τοὺς δὲ ἄλλους Mιλησίους τοὺς πρὶν στασιάζοντας
τούτων ἔταξαν πείϑεσϑαι (Hdt. 5,29,2)
“and they ordained that the rest of the Milesians who had been at
feud should obey these men”
(16b) a Πέρσας δὲ δούλους ἐόντας τὸ πρὶν Mηδῶν νῦν
γεγονέναι δεσπότας (Hdt. 1,129,4)
“and the Persians, who were slaves, are now the masters of the
Medes”
(17) ἐπεὶ δὲ ἡ ἀγγηλίη ἀπίκετο περὶ µάχης τῆς ἐν Mαραϑῶνι
γενοµένης παρά βασιλέα ∆αρεῖον ... καὶ πρὶν µεγάλως
κεχαραγµένον τοῖσι Ἀϑηναίοισι διὰ τὴν ἐς Σάρδις ἐσβολήν
(Hdt. 7,1,1)
__________
8
I have not found any examples of τὸ πρίν combined with perfect in my
corpus, but examples from other authors suggest that the result is the same as
when it is only πάλαι that combines with the perfect. See, for instance, the
following passage by Euripides: (17b) Πελασγιώτας δ᾽ ὠνοµασµένους τὸ
πρὶν ∆αναοὺς καλεῖσϑαι νόµον ἔϑηκ᾽ ἂν Ἑλλάδα. (E. frg. 228,7) ) “He
established by law that the Pelasgians, formerly named Danaans, should be
called Hellad.”
ipabo_66.249.78.36
232
Jesús de la Villa
“When the message concerning the fight at Marathon came to
Darius ..., already greatly angry against the Athenians for their
attack upon Sardis ...”
(18) ἐπιγίγνεταί σφι τέρεα ἔτι µέζονα τοῦ πρὶν γενοµένου τέρεος
(Hdt. 8,37,2)
“They were visited by miracles yet greater than the aforesaid”
In consequence, there does not seem to be a direct connection
between the use of the article and different interpretations related
to aspectual contents.
3) When used with an article, this adverb always implies that the
event is supposed to have lasted for the whole period, that is, the
whole past before the main event, as in the examples already
presented in (14), (15), and (16b). In contrast, πρίν can also refer
to the whole past (19), but in other instances only to the period
that immediately precedes a certain point, as in (16a), (17).
(19a) ἐσήγαγον διδασκάλια ἐς τοὺς Ἕλληνας καὶ δὴ καὶ γράvµµατα, οὐκ ἐόντα πρὶν Ἕλλησι ὡς ἐµοὶ δοκέειν (Hdt. 5,58,1)
“[The Phoenicians] brought with them to Hellas, among many
other kinds of learning, the alphabet, which had been unknown
before this”
(19b) Ἀϑῆναι, ἐοῦσαι καὶ πρὶν µεγάλαι, τότε ἀπαλλαχϑεῖσαι
τυράννων ἐγίνοντο µέζονες (Hdt. 5,66,1)
“Athens, which had been great before, now grew even greater
when her tyrants had been removed”
In consequence, the use of the form without an article also
appears as the unmarked form, whereas the article introduces a
marked determination referring to the whole period.
4) Finally, when τὸ πρίν is used in the text of Herodotus, a
strong contrast is usually stressed between a previous period and
a later moment. This constrast can even be reinforced by the
presence of some other term opposed to τὸ πρίν, such as µετὰ
δὲ (15) or νῦν (16b). This does not seem, however, to be a
generalized case for all the uses of τὸ πρίν with every author.
233
Actually, there are clear instances in other authors where the
form with an article is not used in contrastive contexts, as that of
(20).
(20) ἐσορῶ ... τοὺς τοῦ µεγάλου δή ποτε παῖδας τὸ πρὶν
Ἡρακλέους (E. HF. 442‒4)
“I see the children of Heracles who was once so great”
On the other hand, when the adverb is not accompanied by the
article, there can be some constrast, as in the examples of (19),
or none at all, as in (17) or (21).
(21) ὅτε οἱ Nάξιοι ἦλϑον, ξεῖνοι πρὶν ἐόντες τῶ/ Ἱστιαίῳ (Hdt.
5,30,2)
“when the Naxians, who had been his guest and friends, arrived”
The conclusion, therefore, is that the article is often used when
there are cases of contrastive focalisation, but not always, and,
more important, not exclusively, because πρίν without an article
could also be used in contrastive contexts. What we have, then,
is not a rule, as could be deduced from the Coopers’s assertion,
but only a tendency.
The main conclusions that can be obtained about the use of
πρίν and τὸ πρίν are, then, basically coincident with what we
had previously seen for πάλαι: the article always seems to indicate that a whole period is meant. Second, it seems to be mainly
used when a contrast between two periods is stressed. Thus, the
proposal of Kühner-Gerth is confirmed, and this result also
agrees with Rijksbaron’s conclusion for πρότερον, except that
aspect does not seem to have any relevance for the use of this
adverb.
ipabo_66.249.78.36
Jesús de la Villa
234
4. παραχρῆµα
The data in my corpus are distributed as follows:
9
Hdt. παραχρῆµα 3x / τὸ παραχρῆµα 1x
10
Th. παραχρῆµα 9x / τὸ παραχρῆµα 2x
11
Pl. παραχρῆµα 7x / τὸ παραχρῆµα 1x
The information given for this adverb in dictionnaries and
glossaries is rather confusing and cannot be systematized. As a
general meaning, LSJ gives “on the spot, forthwith,” without
any distinction between those uses with and without an article.12
In particular notes, they add that τὰ παραχρῆµα can be
opposed to τὰ µέλλοντα. Finally, for the expression ἐκ τοῦ
παραχρῆµα the translation “offhand” is proposed.
Bailly offers the following sequence of possible translations
into French: παραχρῆµα “sur le fait, sur-le-champ, à l'instant
même”; ἐκ τοῦ παραχρῆµα “sur-le-champ”; τὸ παραχρῆµα
“ce qui arrive” (noun phrase), “présentement”. The most important indication is the last one; this is the possibility that τὸ
παραχρῆµα could refer to a general period, “the present moment”. Although not completely correct, as we will see, this
indication seems to reflect a right intuition.
The analysis of our data gives the following results:
1) The use of this adverb with a verb in the present stem refers
to the beginning of an event that will go on, this is, an inceptive
sense, as in (22).13 It is the aspectual imperfective content of the
__________
9
Παραχρῆµα: Hdt. 2.93.5, 3.15.4, 7.10.3. τὸ παραχρῆµα: Hdt. 6.11.2.
Παραχρῆµα: Th. 1.20.2, 1.134.3, 2.6.2, 4.7, 4.15.1, 5.15.2, 6.57.4,
7.75.5, 8.92.2. τὸ παραχρῆµα: Th. 6.55.3, 8.44.3.
11
Παραχρῆµα: Pl. Phdr. 243b, Ti. 26a, Lg. 800d, 865a, 866e, Thg. 129e,
131a, . τὸ παραχρῆµα: Pl. Lg. 670d.
12
So also Powell: παραχρῆµα “forthwith, at first”. Tὸ παραχρῆµα “at
first”.
13
On the concept “inceptive”, see Ruijgh (1985). I prefer to use this
terminology and leave for some uses of the aorist the traditional label
“ingressive”.
10
235
present stem that triggers this interpretation: “an event starts
inmediatly and then continues”.
(22a) οἷς δ j ἂν συλλάβηται τῆς συνουσίας ἡ τοῦ δαιµονίου
δύναµις, οὗτοί εἰσιν ὧν καὶ σὺ ᾔσϑησαι· ταχὺ γὰρ
παραχρῆµα ἐπιδιδόασιν (Pl. Thg. 129 e)
“Those who are assisted in their intercourse by that spiritual
power are the persons whom you have noticed; for they make
rapid progress [i.e., “they start to progress”] there and then.”
(22b) ... ἵνα ... καὶ ᾄδοντες αὐτοί τε ἡδονὰς τὸ παραχρῆµα
ἀσινεῖς ἥδωνται (Pl. Lg. 670 d)
“... and in the singing may not only enjoy [i.e., “start to enjoy”]
innocent pleasure themselves at the moment...”
On the contrary, the use with aorist implies the accomplishment
of the event in a definitive way, as in (23).
(23a) καὶ ποιήσας δὴ πᾶσαν τὴν καλουµένην Παλινῳδίαν
παραχρῆµα ἀνέβλεψεν (Pl. Phdr. 243 b)
“And when he had written all the poem, which is called the
recantation, he saw again at once”
(23b) τὸ µὲν παραχρῆµα ἀπέπλευσαν ἐς Xαλκην, ἐντεῦϑεν δ jἐς
Σάµον, ὕστερον δὲ ἐκ τῆς Xάλκης καὶ ἐκ τῆς Kῶ τοὺς
ἐπίπλους ποιούµενοι ἐπὶ τὴν Ῥόδον ἐπολέµουν (Th. 8,44, 3)
“[The Athenians] sailed off for the moment to Chalce, and from
thence to Samos, and subsequently wage war against Rhodes,
issuing from Chalce, Cos, and Samos.”
In both contexts, present and aorist, it is the aspectual content
of the verb what impose the interpretation. There is not any
difference between the use of the adverb with and without the
article.
2) In the uses of τὸ παραχρῆµα it is possible to understand that
a complete period of time and not only a temporal point is
meant. The sense would be one of a general character and could
be rendered into English by something similar to “the immediate
ipabo_66.249.78.36
236
Jesús de la Villa
following time”. This is the case of the sentence in (22b) above,
and also of other senteces such as those of (24).
(24a) νῦν ὧν ὑµεῖς ἢν µὲν βούλησϑε ταλαιπωρίας ἐνδέκεσϑαι,
τὸ παραχρῆµα µὲν πόνος ὑµῖν ἔσται οἶοί τε δὲ ἔσεσϑε
ὑπερβαλόµενοι τοὺς ἐναντίους εἶναι ἐλεύϑεροι· εἰ δὲ
µαλακίῃ τε καὶ ἀταξίῃ διαχρήσεσϑε, οὐδεµίαν ὑµέων ἔχω
ἐλπίδα µὴ οὐ δώσειν δίκην βασιλέι τῆς ἀποστάσιος (Hdt.
6,11,2)
“If you now consent to endure hardships, you will have toil for
the present time, but it will be in your power to overcome your
enemies and gain freedom; but if you will be weak and disorderly, I see nothing that can save you from paying the penalty to the
king for your rebellion”
(24b) οὐδ j ἂν κατασχεῖν µοι δοκεῖ ποτὲ Ἱππίας τὸ παραχρῆµα
ῥαδίως τὴν τυραννίδα (Th. 6,55,3).
“Nor can I ever believe that Hippias would have obtained immediatly the tyranny so easily”
The reference to a complete period is also possible for some
examples of παραχρῆµα such as (22a) above; this contradicts
the distribution of Bailly, who, apparently, reserved this value
for τὸ πρίν. Nevertheless, what is important is that the reference
to a period is absolutely excluded in other cases such as those of
(23a) above, and (25), where παραχρῆµα refers to the single
moment in which the event is accomplished.
(25a) καὶ αὐτίκα ... καὶ παραχρῆµα ἰχϑύων σµικρῶν πίµπλαται
πάντα (Hdt. 2,93,5)
“as soon as [those places are flooded], they are suddenly full of
little fishes”
(25b) καὶ Bοιωτῶν τε παραχρῆµα ξυνέλαβον ὅσοι ἦσαν ... (Th.
2,6,2)
“[The Athenians] had instantly seized all the Boeotians in
Attica...”
(25c) εἰ δὲ µή, τότε ἤδη παραχρῆµα βουλευσόµεϑα ὅτι δράσοµεν (Pl. Thg. 131a)
237
“If it does not, it will be time then for us to consider, at the
moment, what we shall do.”
It seems, then, that the article triggers an interpretation of the
adverbial expressions with παραχρῆµα as related to a whole
period of time: “the immediate following period”. This conclusion agrees with the conclusions already set up for other adverbs.
3) Although I have not found many cases where there is any
focalisation or contrast between the time reference expressed by
(τὸ) παραχρῆµα and any other temporal indication, the most
evident case of contrast in my corpus, that of (23b), presents the
form with an article. If it can be proved to be a clear tendency,
this would also be parallel to what happens to τὸ πάλαι, and τὸ
πρίν, as we have seen above.
In spite of the lack of definitive evidence about this point, it
is interesting to consider some particular uses of παραχρῆµα
with an article that can be of some help.
This adverb appears rather often in prepositional phrases such
as εἰς τὸ παραχρῆµα, ἐκ τοῦ παραχρῆµα, ἐν τῶ/ παραχρῆµα. Here, the preposition seems to determine the interpretation. With directional prepositions, “the immediate following
time” seems the most appropiate meaning, as in (26).
(26) τοὺς ἐπὶ τὰ γυµνάσια καὶ πόνους ἰόντας οὐκ ἴσµεν ὡς
ἀσϑενεῖς εἰς τὸ παραχρῆµα γίγνονται (Pl. Lg. 646 c)
“And we know that men who go to the gymnasia for hard training
commence by becoming weaker?”
Moreover, in this instance, the only one of εἰς παραχρῆµα in
my corpus, there is some contrast bewtween two different temporal references: τοὺς ἐπὶ πόνους ἰόντας and ἀσϑενεῖς εἰς τὸ
παραχρῆµα (26b).
On the other hand, with prepositions that express origin or
position as ἐκ τοῦ παραχρῆµα (27) and ἐν τῶ/ παραχρῆµα
ipabo_66.249.78.36
238
Jesús de la Villa
(28),14 the restricted interpretation “forthwith, immediately”
seems the only possible one. In those cases there is no focalisation at all.
(27) ἤλπιζον γὰρ καὶ τοὺς µὴ προειδότας ... ἐκ τοῦ παραχρῆµα
ἔχοντάς γε ὅπλα, ἐϑελήσειν σφᾶς αὐτοὺς ξυνελευϑεροῦν.
(Th. 6.56.3)
“They hoped that those not in the plot ... would immediately use
the arms in their hands and would fight to recover their liberty”
(27b) ἔγνων ἔγωγε καὶ τῶ/ Kτησίππῳ τὸν νοῦν προσέχων ὡς
ταχὺ ὑµᾶς ἐκ τοῦ παραχρῆµα µιµεῖσϑαι οἷός τε ἦν. (Pl.
Euthd. 303e)
“This fact I perceived myself by watching Ctesippus and
observing how quickly he was able to imitate you on the spot”
(28a) [sc. σῖτος καὶ πότος καὶ ἀφροδίσιαι ἡδοναὶ] τὴν ἡδονὴν
ταύτην ἐν τῶ/ παραχρῆµα παρέχει (Pl. Prt. 353 d)
“Food, drinks and sexual acts produce those pleasures at the
moment”
(28b) καὶ οἷα δὴ τὰ πολλὰ ἀεὶ µετ j ἐµοῦ ξένοι τινὲς ἕπονται καὶ
συνακροῶνται πρὸς οὓς ἐγὼ σεµνότερος ἐν τῶ/ παραχρῆµα
γίγνοµαι (Pl. Mx. 235 b)
“And as I am generally accompanied by some strangers, who
listen along with me, I become in their eyes also all at once more
majestic”
These results show that, when used with a preposition, the construction article + παραχρῆµα does not follow, at least in part,
the same pattern of the uses without a preposition: only with the
preposition εἰς are the particular expressions used similarly to
other of instances of τὸ παραχρῆµα, whereas the use of the
prepositions ἐκ and ἐν yields expressions apparently identical to
single παραχρῆµα. This difference probably has its origin in
the lexical meaning of the prepositions. Nevertheless, it remains
an open question how the expressions ἐκ τοῦ παραχρῆµα and
__________
14
More examples with ἐκ: Cra. 399a, Smp. 185c, Mx. 236b, Crit.. 107d,
Lg. 768b, 867a. With ἐν: Plt. 310c, Phlb. 21c, Prt. 354b, Mx. 235b, R. 408b,
455a, Lg. 799d, 915d.
239
ἐν τῶ/ παραχρῆµα appeared as alternative expressions for the
use of the adverb without an article. The answer to this question
needs more research on a wider corpus of data.
Summing up παραχρῆµα, and leaving apart the uses within
prepositional phrases, the use of this adverb with and without an
article seems to follow the same patterns of πάλαι and πρίν:
first, the presence of the article triggers an interpretation of τὸ
παραχρῆµα as referring to a whole period, and, second, there is
some evidence for the focalisation of those expressions.
5. Interpretation of the data and conclusions
The data we have considered so far show, first, that the fact
that πάλαι, πρίν, and παραχρῆµα sometimes refer to a period,
and other times to a single temporal point is not directly related
to their semantic meaning or to the presence or absence of the
article. Rather, this distinction can be attributed to their distributional combination either with the imperfective tenses – present
and, partially, perfect – or the perfective one – aorist –.
Second, the use of the article triggers that these three adverbs
refer to a complete and delimited period, the past (τὸ πάλαι),
the whole precedent period (τὸ πρίν) or the immediate following
time (τὸ παραχρῆµα). This is very much in accordance with
the short indication of Kühner-Gerth, which has been confirmed,
and with a part of the data of Rijksbaron (2006) for τὸ πρότερον.
It cannot be considered strange that the article introduces a
form of determination in the otherwise unspecified temporal
reference referred to by the adverb. This is the main function of
the article whenever it is used in Greek.15 This determination
requires that clear limits are established on the temporal reference. For a period, those limits can only be the begining and the
end of it. In consequence, the whole period is meant.
__________
15
See, e.g., Kühner-Gerth (19893: 578 ss.; 589 ss.)
ipabo_66.249.78.36
240
Jesús de la Villa
Depending on their lexical meaning, for πρότερον and πρίν,
it is the period before the reference point that becomes determined, that is, “all the time before”. In a similar way, πάλαι
“formerly” becomes “the whole past” and παραχρῆµα “immediatly” becomes “the immediate period”.
Third, there seems to be a clear tendency to use the forms
with article where there is some kind of contrast or focalisation
on the temporal reference given by the adverb. This confirms
what Cooper has indicated. Once again, the article is used in its
specifying and delimiting function. Of course, this possibility is
not independent of our second conclusion: only if there is a
delimited and concrete temporal reference – “the time before”,
“the past”, etc. – can we introduce a contrast with another concrete moment such as “now” or “the future”. Nevertheless,
Cooper’s indication should be reformulated to show that the
focalisation, apparently associated with the use of the article, is
only a tendency, and not an obligation. We have seen some
instances where the article is present but are not focalised, and,
on the contrary, that there are also some uses of the adverbs
alone that are introduced in contrastive contexts.
Bibliography
Bailly, M. A. (1929): Dictionnaire grec-français, Paris: Hachette.
Cooper, G. L. (1998): Attic Greek Prose Syntax, Ann Arbor: The University
of Michigan Press.
Kühner, R. - Gerth, B. (18983): Ausführliche Grammatik der griechischen
Sprache. II Satzlehre, Hannover: Hahn.
LSJ = Lidell, H. G - Scott, R. - Jones, H. S. (19409): A Greek-English
Lexicon, Oxford: Clarendon.
Powell, J. E. (1938): A Lexicon to Herodotus, Cambridge: CUP.
Rijksbaron, A. (20033): Syntax and semantics of the verb in Classical Greek,
Amsterdam: Gieben.
– (2006): “The meaning and word class of πρότερον and τὸ πρότερον”. In:
E. Crespo et alii (eds), Parts of Speech and related topics in Ancient
Greek, Louvain: Peeters, 441–454.
241
Ruijgh, C. J. (985): “L’emploi ‘inceptif’ du thème du présent du verb grec”,
Menmosyne 38, 1–61.
Villa, J. de la (2004): “Aspectos del aspecto en griego antiguo”. In: P.
Quetglas et alii (eds.) Actas del XIV Simposio de Estudios Clásicos.
Sección Catalana de la SEEC, Vich, 2002, Barcelona, Univ. de
Barcelona, 97–124.
ipabo_66.249.78.36
Deontic meaning of the auxiliary verb
construction μέλλω + INF in the Septuagint
and Hellenistic Greek
By ANSSI VOITILA, Joensuu
Abstract: In his monograph on the expression of future meanings in
Greek, Markopoulos (2009) deals with the Auxiliary Verb Construction μέλλω + INF from Ancient to late Medieval Greek. He suggests
that a new meaning, a deontic one, developed from the earlier intention and/or prediction meaning during the first century of the Common
Era. This is remarkable because deontic meaning is usually an early
meaning for futures. This article examines the attestations of the
deontic meaning in the development of μέλλω + INF suggested by
Markopoulos. It is argued that this meaning is already found earlier in
the material from the Hellenistic period, namely the Septuagint.
Keywords: future, trajectory to future, intention, prediction, deontic
meaning, auxiliary verb construction μέλλω, Septuagint, (Classical,
Hellenistic-Roman) Greek
In his recent monograph on the expression of future meanings in
Greek, Markopoulos (2009) deals with the Auxiliary Verb Construction1 μέλλω + INF as a part of his study on ‘the notional
category “Future” as manifested in three future-referring “periphrases” in Greek’, from Ancient to late Medieval Greek (p. 1).2
In this article, I focus on the attestations of a new meaning in the
development of μέλλω + INF suggested by Markopoulos, a
deontic one (2009:48). Such a development would be remarkable because the deontic meaning is usually an early meaning
________

Senior Lecturer at the Philosophical Faculty, University of Eastern
Finland, and a Research Associate at the Department of New Testament
Studies, University of Pretoria, South Africa.
1
He prefers the term “Auxiliary Verb Construction” to “periphrasis”
(Markopoulos 2009:11–13).
2
The rest are ἔχω + INF (to have) and (ἐ)θέλω + INF (to wish/will).
Glotta 89, 242–252, ISSN 0017–1298
Deontic meaning of the auxiliary verb construction μέλλω
243
for futures.3 In fact, on the basis of data collected from 75
languages, Bybee, Pagliuca & Perkins (1991:26‒29) and Bybee,
Perkins & Pagliuca (1994:254) have argued that obligation
meaning belongs to the first stage and probability meaning to
the final stage of a general universal trajectory of constructions
expressing future meanings.4 In their model, all future-referring
constructions pass through the semantic stage of intention
meaning.5 I will argue that if this meaning did in fact emerge
with the μέλλω + INF construction, this had already happened
earlier in the Hellenistic period.
Already a century ago, Magnien (1912:100) notes that the
Auxiliary Verb Construction (AVC) μέλλω + INF denotes ‘un
état du sujet qui le détermine à une action,’ which means the
AVC denotes the intention of the agent.6 Markopoulos argues
that ‘μέλλω itself should be seen as a modal verb with an
“intention” meaning’. Intention is, of course, something that
needs a conscious animate agent and the realisation of this intention belongs clearly to the future. Markopoulos concludes that in
contexts where intention is excluded, the third-person μέλλω
with an inanimate subject gives rise to a prediction meaning,
which he views as a later development (2009:21).7 Markopoulos
________
3
Markopoulos notes this himself when discussing his results concerning
the AVC in the Late Medieval Greek (see Markopoulos 2009:230).
4
See Bybee, Perkins & Pagliuca (1991:26–33; 1994:253–266), Poplack
& Malvar (2007) on Portuguese and Aaron (2010) on Spanish.
5
So-called FUTAGE, pathways to future:
desire
> intention
> future > probability
imperative
obligation
ability > root possibility > future >
possibility
6
The original meaning of the construction remains controversial (Markopoulos 2009:20). According to Liddell, Scott & Jones (1996:1099), the AVC
indicates: “I) an estimated certainty or strong probability in the present, past,
or future: to be destined or likely to”; “II) to be about to, in purely temporal
sense”; or “III) to be always going to do without ever doing: hence delay, put
off”.
7
The context, by frequent implicature, triggers a semantic change in the
meaning of the AVC. Such contexts are called “bridging” contexts by Heine
(2002:84) or “critical” contexts by Diewald (2002:109, 2006). Agentivity (or
animacy) is an operating factor mentioned in several studies when a grammatical form is spreading to new contexts and uses; see Hopper & Traugott (2003:
ipabo_66.249.78.36
244
Anssi Voitila
provides examples of this phenomenon already from the
classical period.
The μέλλω + INF - AVC as a future-tense periphrasis is rare
in Ptolemaic papyri (Mayser 1926:226).8 According to Markopoulos (2009:48), the only development attested in the
Hellenistic-Roman period – and the one we are concentrating on
in this article – is the emergence of the deontic, or at least
‘deontic-future’, meaning of μέλλω. This meaning comes close
to a future certainty marker.9 In his study, Markopoulos presents
three attestations, all of which are dated to the 2nd to 3rd centuries CE. Given this evidence, he concludes that ‘the μέλλω
AVC probably acquired this deontic meaning in the first century
10
AD’ (2009:48). The deontic meaning is illustrated in (1) – (3),
taken from Markopoulos (2009:48).11
(1) καὶ περὶ τοῦ βορρᾶ εἰ μέλλει ποτίζεσθαι μελησάτω
σο̣ι π̣ε̣ρ̣ὶ̣ α̣ὐ̣τ̣ο̣ῦ̣
‘and for the north [field], if it needs to be watered, take care of
it’ (POslo, 155 / 2nd C. AD)
(2) εἰ οἶδας ὅτι θέλεις μετενέγκαι τὸν σῖτον εἰς Πέψ̣α,
γράψον πῶς μέλλομεν αὐτὸν μετενέγκαι.
‘if you know that you are going to (lit.: want to) transfer the
grain to Pepsa, write how we should transfer it’ (PMichael, 17
/ 2nd-3rd C. AD)
________
104–106) in connexion of generalization of grammatical function, Poplack &
Malvar 2007 and Aaron 2006 and 2010. Aaron (2006:43–5, 145) shows how
Spanish future ir a + INF spread from animate to inanimate subjects, such
that the intention meaning is bleached. With animate subjects, predictive
meaning is still there.
8
In the Greek Pentateuch, see Evans (2001:227). For statistics on all the
AVCs, see Markopoulos (2009:48–49).
9
For this term, see Bybee, Perkins & Pagliuca (1994:248).
10
I have checked the corpus and there are not any other examples of
deontic μέλλω AVC in http://papyri.info/ (Duke Database of Documentary
Papyri), which was provided by NYU Digital Library Technology Services
and the Institute for the Study of the Ancient World.
11
The translations of (2) and (3) are the author’s. These texts are found in
the Duke Database of Documentary Papyri (2011).
245
(3) καὶ ταχύ μοι πέμψον δι᾽ οὗ ἐὰν εὕρῃς, ἐπεὶ ὁ χιτὼν
ὑφανθῆναι μέλλει.
‘and quickly send it for me, through anybody you happen to
find, for the tunic needs to be woven’ (Oxy, 113 / 2nd C. AD)
These clauses share some important features – though
Markopoulos does not state this explicitly – that implicitly
exclude the intention or future meanings all the cases: in (1), the
subject is inanimate; in (2), the subject of the AVC is not in the
control of the realisation of the action predicated, since the agent
of the action is the “you” addressed in the letter; and in (3), the
construction is in passive voice, and the weaving of the tunic is
not under the full control of the speaker. Further, in all of these
cases, the AVC denotes present (or continuing) realities rather
than future, although there is reference to some desired/needed
future action in these sentences (i.e. it is volitive), i.e. in all the
cases the speaker has an interest in an action ‘to be performed
volitionality by the addressee or some other person understood
in the context’ (as the deontic must in Narrog 2005:683). Moreover, the sentences suppose either the existence of social (as in
(2) and (3)) or physical conditions (in (1)) independent of the
speaker and the present speech situation that compel an agent to
complete the predicate action12 that exists or may exist (conditionnal sentences) already when these texts were written. Thus,
we may not label these uses as future certainty if ‘future
certainty indicates that the speaker is emphasizing that the future
event is sure to come about’ (Bybee, Perkins & Pagliuca 1994:
248). On the contrary the μέλλω + INF constructions (1) – (3)
meet the requirements of event-oriented modality (vs. speakeroriented, for the distinction, see Narrog 2005:685–90).
As noted earlier, in the schema of universal diachronic paths
to futurity proposed by Bybee, Pagliuca & Perkins, deontic
meanings appear in the earlier stages (first stage) of the trajec-
________
12
This definition of “necessity” is taken from Bybee, Perkins & Pagliuca
(1994:177). In (1), by agent, of course, I mean the actual agent of the action
of “watering”.
ipabo_66.249.78.36
246
Anssi Voitila
tory, while epistemic meanings belong to the final stages. The
matter is further complicated by the fact that, as Magnien (1912)
already noted, μέλλω + INF may also be understood as already
indicating inferred probability in the Homeric epic poetry, Iliad
and Odyssey.13 Of course the directionality in the change of
modal meaning from epistemic to deontic is not unheard of in
cross-linguistic study, but in the languages that point to such a
development, however, the directionality goes from eventoriented to speaker-oriented modality (Narrog 2005). In contrast, as shown above, the deontic μέλλω AVC is used clearly
event-orientedly, thus the deontic μέλλω + INF considered a
development from an epistemic modal meaning would not fit
into this path of diachronic development either. Rather, it is
much more convincing that when one modal marker is moving
towards new meanings the older meanings are demonstrating
indeterminacy, like the Middle English must (see Traugott &
Dasher 2002:128–9; Narrog 2005:684). Therefore, it may be
assumed that these instances of deontic μέλλω AVC are not the
whole picture. In fact, Markopoulos nowhere states how this
deontic meaning emerged; μέλλω only ‘seems to have
developed’ it (Markopoulos 2009:48). Indeed, it seems that the
deontic meaning of the AVC does not occur in Markopoulos’s
(2009:88–93, 113–114) Early Medieval Greek (5th–10th century
th
th
AD) material, but it reappears in the subsequent period (11 –15
century AD), as it ‘developed a new modal meaning: its semantics “shifted” from the domain of intention/prediction to the
domain of deontic modality.’ This development is the manifesta-
________
13
For the term epistemic deductive modality, see Palmer (2001:8, 24ff.);
on inferred probability, see Bybee, Pagliuca & Perkins (1994:24). The fact
itself is illustrated by examples taken from Liddell, Scott & Jones (1996:
1099), in which the AVC indicates inferred probability: μέλλω που
ἀπεχθέσθαι Διὶ πατρί ‘I must have become hateful to father Zeus’, (Iliad
21.83); κελευσέμεναι δὲ σ’ ἔμελλε δαίμων ‘a god must surely have
bidden thee’, (Odyssey 4.274). In these examples, the speaker refers to a
judgement he deduces from the facts he has observed, which is based on the
events that already happened. Naturally, there is no way the speaker is able to
know what the gods really think. See also the examples in Magnien
(1912:100).
247
tion of a contextual specialisation of the AVC along with a
decline in its use: ‘μέλλω was used in the religious texts as
future where there is little room for doubt regarding the future’,
which led to obligation or ‘destiny future’ meanings in lowerregister texts (2009:136–140).
I have gone through all 43 μέλλω AVC cases in the
Septuagint, the translation of the Old Testament from the
Hellenistic period (3rd–1st century BCE), which Markopoulos
(2009:245) claims to have included in his sources but never
cites.14 Most of the AVC in the Septuagint express an intention
or prediction meaning. I found that two instances with deontic
meaning already occur in the Greek Pentateuch, which goes
back to the 3rd century BCE; however, none occur in the later
translated books or in the books originally written in Greek. The
cases are somewhat similar to Markopoulos’s examples, e.g., the
verb μέλλω has an inanimate subject (inf. γίνεσθαι ‘happen’),
and, therefore, an intention reading is excluded. The μέλλειphrases do not denote intention or mere futurity, but necessity
already extant in the present. Consider the cases in (4) and (5).15
(4) ἐσκίρτων δὲ τὰ παιδία ἐν αὐτῇ· εἶπεν δέ Εἰ οὕτως μοι
μέλλει γίνεσθαι ἵνα τί μοι τοῦτο; ἐπορεύθη δὲ
πυθέσθαι παρὰ κυρίου·
‘And the children were cavorting within her, and she said, “If
it has to happen like this with me (that she is suffering because of her pregnancy), why do I have this (i.e., the cavorting)?” Then she went to inquire of the Lord’ (Gen 25:22)
(5) καὶ νῦν πορεύου καὶ ἐγὼ ἀνοίξω τὸ στόμα σου καὶ
συμβιβάσω σε ὃ μέλλεις λαλῆσαι.
‘And now go, and I will open your mouth and instruct you
what you need to/should speak’ (Exod 4:12).
________
14
His sources consist of only low-register texts to avoid Atticism (Markopoulos 2009:15–16).
15
Septuagint text: Accordance. Version 8.4, November 2009. Oak Tree
Software, Inc. http://www.accordancebible.com.
ipabo_66.249.78.36
248
Anssi Voitila
In (4), the AVC refers to the present situation of the speaker.
The future is only implied in the sense that the actual event of
giving birth may or may not be realised in the future; in this
way, it resembles (1). The translator interpreted Rebecca’s utterance to be a complaint: Rebecca refers to the fact that her
extended barrenness, which has been long and difficult for her,
has ended with pregnancy, which is a happy thing. But now the
pregnancy is not proceeding without additional hardships and
she asks why it has to happen like this. The AVC here is clearly
event-oriented.
In (5), the AVC resembles that of (2) in that the agent of the
AVC is not in control of the realisation of the action that the
speaker (i.e., God) is referring to. Thus, God does not mean to
state a mere future fact, nor is He denoting Moses’s intention.
Moses will not speak his own words, but will be given the words
suitable to any coming situation. These words are words he has
an obligation to deliver; thus, this μέλλει has a deontic
meaning. Further the AVC may be considered as event-oriented,
since although God-speaker is going to give the words to Moses
(volitionality) he himself is not their instigator but the coming
situation where the words need to be uttered. A similar sentence
structure and content, but now with a different nuance, is
encountered a bit later, as shown in (6).
(6) καὶ ἐγὼ ἀνοίξω τὸ στόμα σου καὶ τὸ στόμα αὐτοῦ καὶ
συμβιβάσω ὑμᾶς ἃ ποιήσετε.
‘And I will open your mouth and his mouth and instruct you
what you will do’ (Exod 4:15)
The only difference between (5) and (6) is that the relative
clause now has the FUT.IND of the verb ποιεῖν ‘to do,’ and the
addressee constitutes both Moses and Aaron. Moses has not
trusted God and God gets angry. He gives Aaron to Moses to
speak God’s words on His behalf. In (5), God’s words mean that
you should not worry about what to say, and He is polite and
persuasive. However, in (6), His attitude has changed: by using
the FUT.IND, he commands Moses. He is not referring just to
249
speaking anymore, but to everything they must do: ‘you will
certainly do as you are told’. The future in (6) expresses
speaker-oriented modality16 as opposed to the deontic modality
like that expressed by μέλλω in (5).
The deontic reading in (5) is further confirmed by Philo in his
book Who is the heir of the divine things? (20 BC–50 AD). There
(in (7) below), the content of (5) has been used as evidence that
God has given to ‘the man of worth’ both what (σὺ τὰ λεκτέα
συνεβίβασας εἰπεῖν ‘thou has taught it to say what ought to
be said’) and when (ἡνίκα δεῖ φθέγξασθαι ‘when I ought to
speak’) he ought to speak. It may, of course, be argued that
Philo reflects the new usage of the AVC illustrated by Markopoulos; after all, he was living in the 1st century CE. But to give
such an interpretation by an eminent exegete like Philo for a text
that had been used for two or three centuries before him would
suggest that he follows a reading of the AVC here that was
already familiar to him.
(7) σὺ γὰρ ἐθέσπισάς μοι μὴ φοβεῖσθαι, σύ μοι γλῶσσαν
παιδείας ἔδωκας τοῦ γνῶναι ἡνίκα δεῖ φθέγξασθαι,
σὺ τὸ στόμα ἀπερραμμένον ἐξέλυσας, σὺ διοίξας ἐπὶ
πλέον ἤρθρωσας, σὺ τὰ λεκτέα συνεβίβασας εἰπεῖν
τὸν χρησμὸν βεβαιούμενος ἐκεῖνον· ‘ἐγὼ ἀνοίξω τὸ
στόμα σου, καὶ συμβιβάσω σε ἃ μέλλεις λαλήσειν’
‘for thou hast given me an oracular command not to fear, (25)
thou hast given me a tongue of instruction, that I should know
when I ought to speak; thou hast opened my mouth that was
sewed up, and when thou hadst opened it, thou hast also
strengthened its nerves to articulate; thou has taught it to say
what ought to be said, confirming the oracle, “I will open thy
mouth and teach thee what thou oughtest to speak.”’17 (Philo,
Who is the Heir, 25)
________
16
See Bybee, Pagliuca & Perkins (1991:23); the speaker has authority
over the addressee, so the second-person IND.FUT is interpreted as a
command (Bybee, Pagliuca & Perkins 1991:28).
17
I have only slightly altered F.H. Colson and G.H. Whitaker’s translation when I felt this was necessary in order to make my point.
ipabo_66.249.78.36
Anssi Voitila
250
The instances from the Septuagint treated here show that the
μέλλω AVC with deontic meaning was already in use earlier in
the Hellenistic period. If, as is commonly believed, the Septuagint Pentateuch was translated before the year 250 BCE, this
meaning must have emerged during the final decades of the 4th
century. This is all the more interesting when we realise that the
translators worked in a relatively literal manner18 and that the
use of the μέλλω AVC in these cases was not consonant with
this literalness, i.e., it was not dictated by the original Hebrew
text. In Exod 4:12, the FUT.IND serves as the most common
equivalent of the Hebrew verb form in question (yiqtol) and the
AVC compels the translator to use two Greek words to render
one Hebrew word; in Gen 25:22, there is no equivalent in the
Hebrew – thus, the AVC must reflect the everyday Greek usage
of the translators.
In the case of the μέλλω AVC then, Markopoulos’s analysis
has proved inadequate with regards to the deontic meaning and
its usage in the Septuagint. If the AVCs are read as deontic, they
add at least three centuries to what Markopoulos considers the
date of origin for the deontic use of the AVC. This meaning was
already well established by the beginning of the Hellenistic
period. This fact may indicate that this usage is even older, but
such a determination is nearly impossible, since this meaning of
the AVC do not appear in the corpus of the Classical period,
which consists almost entirely of material from higher-register
texts, which may not have allowed such a usage for μέλλω +
INF. Nevertheless, deonticity is not necessarily, as Markopoulos
concludes, a new meaning in the development of the μέλλω +
INF, but may well be seen as a remnant of its/one of its original
meaning(s) and as such accords with the larger cross-linguistic
universal paths of change described by Bybee and colleagues
(1991, 1994).
________
18
See, for example, Thackeray (1909:12–13) and Siegert (2001:131–135).
251
Bibliography
Aaron, J. E. (2006): Variation and change in Spanish future temporal expression. Ph.D. dissertation. University of New Mexico.
– (2010): Pushing the envelope: Looking beyond the variable context.
Language Variation and Change 22:1.1–36.
Bybee, J., W. Pagliuca, & R. Perkins (1991): Back to the Future. Approaches
to Grammaticalization II. Eds. E. Closs Traugott and B. Heine. TSL 19:2.
Amsterdam-Philadelphia: Benjamins Publishing, 17–58.
– (1994): The evolution of grammar: Tense, aspect, modality in languages
of the world. Chicago: University of Chicago Press.
Diewald, G. (2002): A model for relevant types of contexts in grammaticalization. New Reflections on Grammaticalization. Eds. I. Wischer and
G. Diewald. TSL 49. Amsterdam-Philadelphia: Benjamins Publishing,
103–120.
– (2006): Context types in grammaticalization as constructions. Constructions SV1–9/2006. Available at: http://www.constructions-online.de/
articles/specvol1/686.
Duke Database of Documentary Papyri (2011): NYU Digital Library Technology Services and the Institute for the Study of the Ancient World.
Available at: http://papyri.info/.
Evans, T. V. (2001): Verbal syntax in the Greek Pentateuch. Oxford: Oxford
University Press.
Heine, B. (2002): On the role of context in grammaticalization. New reflections on grammaticalization. Eds. I. Wischer and G. Diewald. TSL 49.
Amsterdam-Philadelphia: Benjamins Publishing, 83–101.
Hopper, P. J. & E. Closs Traugott (2003): Grammaticalization. Second
edition. Cambridge: Cambridge University Press.
Liddell, H. St., R. Scott & Sir H. St. Jones (1996): A Greek English Lexicon.
Oxford: Oxford University Press.
Magnien, V. (1912): Emplois et origines du futur grec. Paris: Librairie
Ancienne Honoré Champion.
Markopoulos, Th. (2009): Future in Greek: From Ancient to Medieval.
Oxford: Oxford University Press.
Mayser, E. (1926): Grammatik der griechischen Papyri aus der Ptolemäerzeit mit Einschluß der gleichzeitigen Ostraka und der in Ägypten
verfassten Inschriften, Band 2:1 Satzlehre. Berlin-Leipzig: Walter de
Gruyter.
Narrog, H. (2005): Modality, mood, and change of modal meanings: A new
perspective. Cognitive Linguistics 16–4:677–731.
Palmer, F. R. (2001): Mood and modality. Second edition. Cambridge:
ipabo_66.249.78.36
252
Anssi Voitila
Philo of Alexandria. Who is the heir of the divine things? In: Philo. Volume
IV. Loeb Classical Library 261, trans. F. H. Colson and G. H. Whitaker.
Cambridge, MA: Harvard University Press, 269–447.
Poplack, Sh. & E. Malvar (2007): Elucidating the transition period in
linguistic change. Probus 19:1.121–169.
Siegert, F. (2001): Zwischen Hebräischer Bibel und Altem Testament: Eine
Einführung in die Septuaginta. Institutum Judaicum Delitzschianum
Münsteraner Judaistische Studien 9. Münster-Berlin-Hamburg-London:
Lit Verlag.
Thackeray, H. St. J. (1909): A Grammar of the Old Testament in Greek
according to the Septuagint I. Cambridge, MA: Cambridge University
Press.
Traugott, E. Closs & R. B. Dasher (2005): Regularity in semantic change.
Cambridge Studies in Linguistics Cambridge: Cambridge University
Press.
λυκάβας, λυκηγενής, ἀμφιλύκη
By MARTIN L. WEST, Oxford
Odyssey 14. 161–2 = 19. 306–7
τοῦδ᾽ αὐτοῦ λυκάβαντος ἐλεύσεται ἐνθάδ᾽ ᾽Οδυσσεύς,
τοῦ μὲν φθίνοντος μηνός, τοῦ δ᾽ ἱσταμένοιο.
The phrase τοῦδ᾽ αὐτοῦ λυκάβαντος evidently indicates a timeunit that is currently running, ‘in this very ‒’, and the second
verse provides either the same or a more exact specification of
the time in question: ‘as one moon wanes and the next one is
established’, in other words, at the transition from one month to
the next. λυκάβας, which is found nowhere else in archaic or
classical literature, was anciently understood to mean ‘year’. In
the Hellenistic period learned poets took it up in that sense, and
it is even used in some verse inscriptions. Its attribution to the
Arcadians in the γλῶσσαι κατὰ πόλεις (AB 1095) seems to be
based only on its occurrence in late Arcadian epitaphs.1 The
statement in LSJ that Dio Prus. 7. 84 interpreted it as ‘month’ in
the Homeric verse is incorrect.2 The only variant meaning found
in ancient texts is ‘time’ in two late epitaphs from Amorgos.3
Not all modern scholars have accepted the meaning ‘year’.
There is after all a regular word for ‘year’, ἔτος, which is
__________
I am indebted to Andreas Willi for his helpful comments on a first draft
of this article and for directing me to some relevant literature.
1
Manu Leumann, Homerische Wörter, Basel 1950, 273.
2
See C. J. Ruijgh, L’élément achéen dans la langue épique, Amsterdam
1957, 147; Chantraine, DELG 649. Dio writes (i. 205. 18 Arnim): ἂν ἄρα
φανῆι ἀληθεύων περὶ τοῦ ᾽Οδυσσέως, ὅτι ἐκείνου τοῦ μηνὸς ἥξοι, which is
sufficiently accounted for by τοῦ μὲν φθίνοντος μηνός, τοῦ δ᾽ ἱσταμένοιο.
3
IG XII. 7. 120 (cf. 127), Roman period, τὸ τέλος [λ]υκάμαντι ἀπέδωκα,
‘I have paid my debt to Time’ (‘pro χρόνωι usurpatum’ Delamarre in IG). For
the μ see Schwyzer, Griech. Gramm. i. 257 and especially P. Kretschmer,
ZVS 35, 1899, 603‒8: it is due to the following nasal. For the semantic shift
cf. the contrary development by which χρόνος comes to mean ‘year’.
Glotta 89, 253–264, ISSN 0017–1298
ipabo_66.249.78.36
254
Martin L. West
common enough in Homer; and even if an obscure synonym
existed in the epic language, the prophecy of Odysseus’ return
to Ithaca (made by himself in his beggar guise, to Eumaios and
Penelope) would seem to call for a greater immediacy than ‘this
very year’, especially as a much more precise calendrical datum
is added in the next line. Some, accordingly, have conjectured
that λυκάβας meant ‘month’, ‘lunation’, or perhaps more
specifically the darker part of the month, the days around New
Moon.4
The question is bound up with that of the word’s etymology,
which has not yet been satisfactorily resolved. The author of the
latest etymological dictionary holds it to be pre-Greek because
of the variant form λυκάμας on the Amorgos inscriptions (but
see above, n. 3) and the stem in -αντ- (but see below).5 Most
scholars, no doubt rightly, have assumed that the λυκ- represents
the IE word for ‘light’ which still glimmers in Greek in
ἀμφιλύκη νύξ (Il. 7. 433, of the morning twilight) and in the
later attested words λυκόφως, λυκοψία, λυκαυγής, all referring
to twilight; Hesych. λ 1384 preserves an unharvested Hellenistic
poetic fragment λυκοειδέος ἀοῦς.6 The root appears also in
λευκός, λύχνος, and perhaps λεύσσω.
__________
4
So Leumann, op. cit. 212 n. 4, ‘Zusammenhang und Zeitbestimmung
führen auf den bevorstehenden “Neumondstag” zwischen zwei Monaten, an
dem Apollon ein Opfer dargebracht wird, υ 276 ff.’; Η. Koller, Glotta 51,
1973, 30, ‘Λυκάβας ist also die ἕνη καὶ νέα, die νουμηνία, zugleich aber auch
der laufende Monat’; Norman Austin, Archery at the Dark of the Moon,
Berkeley-Los Angeles 1975, 244–6.
5
R. Beekes, Etymological Dictionary of Greek, Leiden-Boston 2010, i.
876. Beekes seems in general to resort too readily to pre-Greek solutions.
6
Macrobius, Sat. 1. 17. 37–41, in a learned discussion of Apollo Lycius,
after citations of Cleanthes (SVF i. 123. 16) and Antipater of Tarsus (SVF iii.
249. 24), writes: prisci Graecorum primam lucem, quae praecedit solis
exortus, λύκην appellauerunt ἀπὸ τοῦ λευκοῦ. … idem Homerus εὔχεο δ᾽
᾽Απόλλωνι Λυκηγενέϊ κλυτοτόξωι (Il. 4. 101), quod significat τῶι γεννῶντι
τὴν λύκην, id est qui generat exortu suo lucem. … neque minus Romani, ut
pleraque alia ex Graeco, ita et lucem uidentur a λύκη figurasse. annum
quoque uetustissimi Graecorum λυκάβαντα appellabant τὸν ἀπὸ τοῦ λύκου
(id est sole) βαινόμενον καὶ μετρούμενον. … ipsos quoque λύκους a λύκη, id
est a prima luce, appellatos quidam putant, quia hae ferae maxime id tempus
aptum rapiendo pecori obseruant, quod antelucanum post nocturnam famem
ad pastum stabulis expellitur. The alleged word λύκη is no doubt abstracted
255
λυκάβας then looks like a compound. But what is the second
element? August Fick thought he had found it in Hesych. α 42
ἄβα·τροχός, ἢ βοή, the compound then meaning literally
‘Lichtkreislauf’.7 But even if we grant the existence of this ἄβα,
we do not know whether its first alpha is long or short, and it
remains unexplained why a compound λυκ-άβα should sprout a
suffix -ντ-.
Of more recent discussions cited in Frisk’s and Chantraine’s
etymological dictionaries, the most profitable is that of H.
Koller in Glotta 51, 1973, 29–34. He formulates the problem as
follows (31):
Das Wort muß zusammengesetzt sein, doch widerspricht es allen
Möglichkeiten griechischer Wortzusammensetzung: Eine Wortfuge
mit kurzem α, ein Hinterglied -βας sind nicht denkbar, ob man nun
das Vorderglied mit λύκος „Wolf“ oder *lǔk- „Licht“ in Verbindung
bringt oder in -βας ein Hinterglied zum Verb βαίνω sucht. Für das
Vorderglied „Wolf“ käme nur λυκο- in Frage, für „Licht“ allenfalls
λυκη-; zu βαίνειν könnte das Hinterglied nur -βατης heißen.
His solution is to postulate an original accusative expression
such as *τόνδ᾽ αὐτὸν λύκα βάντα, or *αὖθι μένων λύκα βάντα,
where λύκα βάντα means ‘das weggegangene Licht’, i.e. the
lightless night of the New Moon. After the old word *luk- fell
out of use, he supposes, the phrase was interpreted as a noun
λυκάβαντα, ‘New Moon’; hence the Odyssey poet could create
the genitive phrase τοῦδ᾽ αὐτοῦ λυκάβαντος, ‘during this New
Moon’, and later poets by misinterpreting the Homeric text
could change the meaning to ‘year’.
Some objections spring to mind. First, since *λυκ- corresponds exactly to Sanskrit ruc-, we may expect it to be a feminine, not a masculine. Second, in its other survivals in Greek it
always refers to twilight; at an earlier stage it may have meant
__________
from the Homeric ἀμφιλύκη νύξ. Macrobius’ source may be Cornelius
Labeo, De oraculo Apollinis Clarii, a work which he cites in 1. 18. 21.
7
Fick, GGA 1894, 240; followed by F. Bechtel, Lexilogus zu Homer,
Halle/Saale 1914, 41 and 218.
ipabo_66.249.78.36
256
Martin L. West
more generally ‘daylight’.8 It is a little bold to assume that with
no qualification it could signify the moon or the moon’s light.
And we have to suppose that the simple βάντα, with no modifier
or directive case, is adequate to convey the meaning ‘weggegangen’. The perfect βέβηκε can in some circumstances mean ‘is
gone hence’, ‘has perished’ (LSJ βαίνω A. I. 3), and the aorist
can be used in expressions such as οὓς μὴ κῆρες ἔβαν θανάτοιο
φέρουσαι (Il. 2. 302), where only one destination comes into
question. But the moon is ‘going’ all the time. To say that it ἔβη,
without further definition, would be uninformative.
More recently Michael Janda has offered a new exegesis.9
While agreeing that λυκα- should be the accusative of ‘light’, he
is reluctant to see -βαντ- as the participle, as participles do not
normally figure in compounds. He analyses it as *-gwh2-ent-, the
verbal root with a supposed ‘Caland suffix’. The parallels
adduced, however, are unimpressive. As to the meaning of the
compound, he reverts to ‘year’, with the explanation that ‘going
to the (sun)light’ was a designation of the Milky Way, which
later came to stand for ‘year’. I find this quite far-fetched and
fanciful.
I propose something closer to Koller’s derivation. I agree
with him that λυκα is an accusative and that βας is the participle
of βαίνω. But in my view λυκάβας or λύκα βάς was originally a
qualification of μείς, and *μεὶς λυκάβας, or *méns lúka gwánts,
meant ‘the moon going to the daylight’.10 As the moon wanes
towards the end of the month and approaches the sun, it
__________
8
Cf. RV 6. 39. 4 ayáª rocayad arúco rucānó … ayám īyata ṛtayúgbhir
áśvaiÙ suvarvídā n£bhinā car≠aºipr£Ù, ‘this (Soma, as sun) has illumined the
lightless (worlds) by illumining … this (Soma) goes with the rightly-yoked
steeds, with his sun-bringing nave filling the lands’.
9
Michael Janda, Elysion: Entstehung und Entwicklung der griechischen
Religion, Innsbruck 2005, 268–70.
10
For the simple accusative with the verb of motion cf. Chantraine,
Grammaire homérique, ii. 45 f.; A. A. Macdonell, A Vedic Grammar for
Students, Oxford 1916, §197. 1; my Old Avestan Syntax and Stylistics,
Berlin-Boston 2011, §51; B. Delbrück, Vergleichende Syntax d. idg. Sprachen, i. 363–5. Of Indra it is said ágacho rocanáṃ diváḥ, ‘you went to the
light of heaven’, RV 8. 98. 3, and so of the sun in 10. 170. 4.
257
becomes progressively less visible at night, its time above the
horizon coinciding more and more with the hours of daylight. In
Greek calendars the waning month is generally μεὶς φθίνων; the
moon could be seen shrinking from day to day. It is imaginable
that when it finally became invisible in the sun’s glare, this was
at one time referred to as *méns lúka gwánts. With univerbation
the less significant accent was lost, giving *lúkagwαnts, which
by the law of limitation resulted in paroxytone λυκάβας. And
just as calendar dates came to be given in the form τετάρτη
φθίνοντος, with ellipse of μηνός, so it is plausible enough that
λυκάβας could come to be used by itself without μείς.
I have tried to find ethnographic parallels for the idea of the
moon ‘going to the light’ at the end of the month. I have not
found much. The Basuto say, or used to say, that ‘the moon has
gone into the dark’; the Ho of Bengal that ‘it shines unto
daybreak’; and the Toradja of the Dutch East Indies name(d) the
28th of the month polioenja ‘passing’, with reference to the
moon’s going past the sun.11 None of these is quite what I was
looking for, but they do have it in common that they refer not to
the moon’s own aspect but to its changed position in the sky.
The λυκάβας of the Odyssey, the New Moon at which
Odysseus reveals himself, coincides with a festival of Apollo
(20. 156, 276–8; 21. 258). Apollo’s association with the New
Moon is well attested later.12 Does it have anything to do with
his epithet λυκηγενής?
This occurs only in the passage of the Iliad (4. 101 ~ 119) in
which Athena advises Pandaros to pray, and he does pray,
᾽Απόλλωνι λυκηγενέϊ κλυτοτόξωι, while aiming his arrow at
Menelaos. The bT and D scholia explain it as ‘born in Lycia; or
because a wolf led Leto, after she had given birth, to the
Xanthos to cleanse herself’ (cf. Ant. Lib. 35). In either case the
commentator was thinking of Apollo’s oracular shrine at Patara.
Aristarchus, however, affirmed that it did not relate to the well__________
11
M. P. Nilsson, Primitive Time-Reckoning, Lund 1920, 149 f., 161.
M. P. Nilsson, Die Entstehung und religiöse Bedeutung des griechischen Kalenders, 2. Aufl., Lund 1962, 38, 40 f.
12
ipabo_66.249.78.36
258
Martin L. West
known Lycia in the south-west of Asia Minor but to the Lycia in
the Troad from which Pandaros came.13 It certainly seems
probable that the poet used the word here (and nowhere else) on
account of Pandaros’ ‘Lycian’ provenance, and therefore that he
understood it to mean ‘Lycian-born’. But it is unlikely to be his
own coinage, and we should consider the possibility that its
original meaning was a different one.
The interpretation as ‘Lycian-born’ is problematic on
grounds both of form and of substance. As to the form,
*Λυκιηγενής would have been acceptable, but how can Λυκηbe justified? Paul Kretschmer, followed by A. Heubeck and R.
Beekes, claimed that it is an old form corresponding to the
Lukka of Hittite archives.14 But there is no evidence that Gk.
Λυκία replaced an older *Λύκα or *Λύκη; and if there ever was
such a form, one might expect the natal adjective to be formed
with the locative, *Λυκαιγενής, as in Θηβαιγενής and Πυλοιγενής. As for the substance, Apollo’s birthplace in all archaic
and classical sources is Delos, and his natal title is Δαλογενής
(Bacchyl. 3. 58, 11. 15; PMG 950(a)). Later we hear of his birth
being celebrated in cult in other places, including Lycia (Semos,
FGrHist 396 F 20), but that will mean Patara, not the Troad.
Further evidence against his being native to Lycia is provided by
the fact that in Lycia itself the Greek Apollo was equated with a
local god Natr-.15
The Patara legend, as represented in the Iliadic scholion cited
above, introduced a wolf. Here we must take account of
Apollo’s titles Λύκειος and Λύκιος. These are not Homeric.
Λύκειος appears in Alcman (PMGF 49, 50(a)) and is a genuine
cult title attested in many places. It cannot mean ‘Lycian’; it can
mean ‘of the wolf’, and it was so understood: Aesch. Sept. 146
καὶ σὺ Λύκει᾽ ἄναξ, λύκειος γενοῦ στρατῶι δαΐωι with sch.
__________
13
64.
On this northern Lycia see my The Making of the Iliad, Oxford 2011,
14
P. Kretschmer, Glotta 21, 1933, 227–9; A. Heubeck, Praegraeca,
Erlangen 1961, 54; R. Beekes, Journal of Ancient Near Eastern Religions 3,
2003, 15 f.
15
Cf. M. Egertmeyer, Res antiquae 4, 2007, 211.
259
quoted below; Soph. El. 6 f. τοῦ λυκοκτόνου θεοῦ | ἀγορὰ
Λύκειος with Finglass’s commentary.16 Λύκιος does mean
‘Lycian’, but it is not found of Apollo before Pind. Pyth. 1. 39,
where three centres of his worship are referred to (Lycia, Delos,
Delphi). The idea of a ‘Lycian’ Apollo was no doubt influenced
by the Homeric λυκηγενής, but it does not throw light on it.
Does λυκηγενής then have something to do with wolves?
Besides the Patara legend that a wolf led Leto to the river to
cleanse herself, there is another account that contrives to bring a
wolf into Apollo’s birth story: it tells that Leto herself took the
form of a wolf to avoid Hera’s jealousy, and other wolves
brought her in twelve days from the Hyperboreans to Delos.17
λυκηγενής could in theory mean ‘wolf-born’, with -η- serving as
a metrical alternative for -o-, as in ἐλαφηβόλος, etc.18 But these
myths are evidently late aetiological inventions, and it is hard to
believe that any such story underlies the (pre-)Homeric epithet
λυκηγενής.
If the first element of λυκηγενής is not convincingly
explicable as either ‘Lycian’ or ‘wolf’, does it have to do with
light? The idea was already broached in antiquity, as may be
seen from the passage of Macrobius quoted above (n. 6).
Heraclitus, Alleg. Hom. 7. 10, writes:
λυκηγενέτην δὲ προσαγορεύουσιν αὐτὸν οὐχ ὡς ἐν Λυκίαι
γεγενημένον ‒ ἔξω γὰρ οὖν τῆς ῾Ομηρικῆς ἀναγνώσεως οὗτος ὁ
νεώτερος μῦθος ‒ ἀλλ᾽ ὥσπερ οἶμαι τὴν ἡμέραν ἠριγένειαν
ὀνομάζει, τὴν τὸ ἦρ γεννῶσαν, ὅπερ ἐστὶν ὄρθρον, οὕτω
λυκηγενῆ προσηγόρευσε τὸν ἥλιον, ἐπειδὴ τοῦ κατὰ τὴν αἴθριον
ὥραν λυκαυγοῦς αὐτός ἐστιν αἴτιος· ἢ ὅτι τὸν λυκάβαντα
γεννᾶι, τουτέστι τὸν ἐνιαυτόν.
__________
16
P. J. Finglass, Sophocles. Electra, Cambridge 2007, 94 f.
Arist. Hist. anim. 6. 35, 580a16; Philostephanos FHG iii. 33 ap. sch.
Ap. Rhod. 2. 123–9a. Aelian, Hist. anim. 10. 26, explains the Homeric
λυκηγενής from this.
18
Cf. A. Debrunner, Griechische Wortbildungslehre, Heidelberg 1917, 68
f., and in particular for -ᾱ/η-γενής F. Solmsen, Untersuchungen zur griech.
Laut- und Verslehre, Strassburg 1901, 27 f.
17
ipabo_66.249.78.36
260
Martin L. West
Cf. sch. Aesch. Sept. 145a, Λύκιόν φησι τὸν ᾽Απόλλωνα ἢ διὰ
τὸ ἐν Λυκίαι τῆι πόλει τιμᾶσθαι αὐτόν, ἢ διὰ τὸ ἀνατιθέναι
τούτωι λύκον ὡς ἱερὸν ἢ διὰ τὸ λύκους ἀνελεῖν, ἤ, ὅ γε
κρεῖττον, διὰ τὸ λυκόφως ποιεῖν τῆς νυκτὸς ὑποχωρούσης·
᾽Απόλλων γὰρ ὁ ἥλιος. And 145b, διὰ τὸ λύκον ποτὲ ἀποκτεῖναι
᾽Απόλλω τοῦτον κεκλῆσθαί φασιν. τοῦτο δὲ πέπλασται ἀπὸ τοῦ
τὸ λυκόφως ὑπὸ ἡλίου διαλύεσθαι, εἰς τοῦτον ἀλληγορουμένου
᾽Απόλλωνος. λυκόφως δὲ καλεῖται διὰ τὸ ἔχειν τὸν λύκον
τοιάνδε χροιάν.
In the nineteenth century the ‘light’ interpretation was readily
embraced by those who saw Apollo as the god of light. In
Roscher’s Lexikon, for example, we read that Lykeios is a
‘Beiname des Apollon als des Lichtgottes … ohne Zweifel
herzuleiten von der Wurzel λυκ- leuchten (vgl. ἀμφιλύκη,
λυκόφως, λυκάβας, λυκαυγής, λυκοειδής, λύχνος …)’.19 But the
Homeric Apollo, whom we first see coming under cover of
night and shooting plague arrows at the Achaean army, cannot
be characterized as intrinsically a ‘god of light’, and his
identification with the sun, which Heraclitus’ and Macrobius’
explanations presuppose, cannot be traced earlier than the fifth
century. Kretschmer wrote in 1936 (Glotta 24, 227 n. 3), ‘Von
dem „Lichtgott“ Apollon ist man längst zurückgekommen, so
daß auch die Übersetzung „lichtgeboren“ wegfallen muß’.
However, even in the archaic period an epiphany of Apollo
may be accompanied by manifestations of brightness. In the
Pythian hymn, when he arrives at Crisa on the Cretans’ ship
from Cnossos, he leaps ashore,
ἀστέρι εἰδόμενος μέσωι ἤματι· τοῦ δ᾽ ἀπὸ πολλαί
σπινθαρίδες πωτῶντο, σέλας δ᾽ εἰς οὐρανὸν ἷκεν
ἐς δ᾽ ἄδυτον κατέδυσε διὰ τριπόδων ἐριτίμων.
ἔνθ᾽ ἄρ᾽ ὅ γε φλόγα δαῖε πιφαυσκόμενος τὰ ἃ κῆλα,
πᾶσαν δὲ Κρίσην κάτεχεν σέλας.
__________
19
Schirmer in Roscher, ii. 2175; cf. Bechtel, Lexilogus 41, ‘im Lichte
geboren’; W. Brandenstein, DLZ 85, 1964, 128.
261
(Hymn. Ap. 441–5.) When he dances and plays his cithara
among the gods, he is surrounded by an αἴγλη φαεινή (ibid.
202). His birth is not described in such explicitly radiant terms,
but it was a glorious event, filling Delos with an ambrosial
scent, and the earth ἐγέλασσε, which suggests a sparkling
brightness.20
There are other divine children whose birth is accompanied
by manifestations of light. In one of the Cretan cave cults the
annual birth of Zeus was signalled to the worshippers by a blaze
of fire: Ant. Lib. 19. 1–2 ἐν Κρήτηι λέγεται εἶναι ἱερὸν ἄντρον
μελισσῶν, ἐν ὧι μυθολογοῦσι τεκεῖν ῾Ρέαν τὸν Δία. … ἐν δὲ
χρόνωι ἀφωρισμένωι ὁρᾶται καθ᾽ ἕκαστον ἔτος πλεῖστον
ἐκλάμπον ἐκ τοῦ σπηλαίου πῦρ· τοῦτο δὲ γίνεσθαι
μυθολογοῦσιν ὅταν ἐκζέηι τὸ τοῦ Διὸς ἐκ τῆς γενέσεως αἷμα.
Similarly in the Eleusinian mysteries the birth of the divine child
was announced under a blaze of firelight.21
If λυκηγενής contains the ‘light’ root, the -η- might be
explained as a simple Bindevokal, taking the place of the more
regular -o- (cf. above). But it is more attractive to see it as the
old instrumental ending, *lukē, corresponding exactly to Vedic
rucƒ.22 Another example of a case-compound with instrumental
and -γενής is the name ᾽Ιφιγένεια; a corresponding masculine
᾽Ιφιγενής is attested in Et. Magn. 480. 15. I take the instrumental
here to have its elementary associative sense, not ‘born through
strength’ but ‘born with strength’, i.e. innately strong. By the
same token λυκηγενής might be applied to a god who brings
light into the world, or just to one whose birth is accompanied
by light.
Apollo was not necessarily the first or the only deity of
whom it was predicated. We had better leave open the question
__________
20
Theogn. 5–10; on γελᾶν cf. my note on Hes. Th. 40.
See N. J. Richardson, The Homeric Hymn to Demeter, Oxford 1974, 26
f., 317 f.
22
So M. Egertmeyer (as n. 15), 212, who renders ‘né par la lumière’. M.
Janda (as n. 9), 270, takes it to be either an instrumental used as a locative or
as an endingless locative of a collective *lukah2 [sic], anyway meaning ‘der
im Licht geborene’.
21
ipabo_66.249.78.36
262
Martin L. West
of whether in his case it alluded to light produced by ritual
artifice, as in the cases mentioned above, or to a natural
phenomenon: we have seen that from an early period he was
associated with the New Moon, though his birthday is regularly
placed not on the first but on the seventh day of the month (Hes.
Op. 771 with my note). Whatever the truth of it may be, I hope
to have established that the ‘light’ interpretation of λυκηγενής
deserves renewed attention and is more plausible than the
alternatives.
Finally let us return to ἀμφιλύκη. This too is a Homeric
hapax. The verse in which it appears is Il. 7. 433,
ἦμος δ᾽ οὔτ᾽ ἄρ πω ἠώς, ἔτι δ᾽ ἀμφιλύκη νύξ,
and it clearly signifies the morning twilight; the poet begins
‘when it was neither yet dawn’ as if he meant to continue ‘nor
still fully night’, but then eliminates the antithetical term in
favour of the intermediate one. Hellenistic poets took up the
word and used it as a noun. Apollonius Rhodius offers his own
gloss on it, 2. 669–71:
ἦμος δ᾽ οὔτ᾽ ἄρ πω φάος ἄμβροτον οὔτ᾽ ἔτι λίην
ὀρφναίη πέλεται, λ ε π τ ὸ ν δ᾽ ἐ π ι δ έ δ ρ ο μ ε ν υ κ τ ί
φ έ γ γ ο ς , ὅτ᾽ ἀμφιλύκην μιν ἀνεγρόμενοι καλέουσιν.
For him, then, ἀμφιλύκη νύξ meant literally ‘night with a weak
light encroaching on it’.23 We have seen that Macrobius
recognizes a word λύκη ‘light’ that must have been derived from
analysis of ἀμφιλύκη, and Apollonius appears to take the same
view.
Bechtel explains ἀμφιλύκη as formed by hypostasis from a
phrase ἀμφὶ λύκην, and suggests that one might even restore ἔτι
__________
23
Other passages are Arat. 747; [Opp.] Cyn. 1. 135, 3. 306; Paul. Sil. AP
5. 281. 4; Agathias AP 7. 583. 5; Apollin. 62. 1; Jo. Gaz. 1. 333; anon. in
Suda α 1743.
263
δ᾽ ἀμφὶ λύκη(ν) νύξ in Homer.24 He takes the ἀμφί to be
temporal. That would imply that λύκη itself meant ‘twilight’;
but ‘when the night was still around twilight-time’ is not a
convincing expression. In any case the temporal use of ἀμφί is
not found before the fifth century.
In λυκάβας and λυκηγενής we have found cases of a root
noun *luk-, corresponding to Skt. ruc, but we have no ground
for postulating an eh2-stem *lukā. *luk- is sufficient basis for an
adjective ἀμφίλυκος, of which ἀμφιλύκη could be the feminine.
(One might expect *ἀμφίλυκος νύξ; but cf. ἀσπίδος
ἀμφιβρότης.) It would mean, rather as Apollonius understood it,
‘with light around it’.25
However, there is another problem. As H. Koller observes in
the article previously cited, there is no other example in Homer
of an ἦμος-clause with no verb.26 He emends accordingly to
ἀμφὶ λύκ᾽ ἦν νύξ. But this is open to the same objection as
Bechtel’s restoration ‒ it is not a convincing form of expression
‒ while being rhythmically inferior.
The solution, I submit, is that λύκη is itself a verb form, or
rather that it was one when the phrase was first created. Beside
the root noun there was a root verb *luk-, reflected in Hittite luk-,
Sanskrit ruc-/roc-, and Latin lūcēre (see below). The -η is the
old stative suffix, from which certain old verbs in Greek derive
an intransitive or passive aorist, as in ἐάγη, ἐδάη, ἐδάμη, ἐκάη,
ἐμίγη, ἐμάνη, ἐπάγη, ῥύη, ἐτύπη, φάνη, ἐχάρη.27 It has been
claimed that these Homeric forms are hardly distinct in sense
from statives.28 This is not really the case; so far as I can see,
__________
24
ZVS 45, 1913, 228; Lexilogus 41 f.; cf. E. Risch, Wortbildung d. Hom.
Sprache, 2. Aufl., Berlin-New York 1974, 228; Schwyzer, Griech. Gramm. i.
437.
25
So Janda (as n. 9), 268 n. 557, ‘zu beiden Seiten Aufleuchten habend’.
But the twilight does not appear on both sides of the sky.
26
Glotta 51, 1973, 32. There are a couple of examples with ὄφρα ‘so long
as’ and one with πρίν; see Chantraine, Gramm. hom. ii. 4.
27
For a fuller list of those attested in Homer see Chantraine, Gramm.
hom. i. 399–401.
28
A. L. Sihler, New Comparative Grammar of Greek and Latin, New
York-Oxford 1995, 564.
ipabo_66.249.78.36
264
Martin L. West
they are used only as punctual aorists. The unaugmented λύκη
seems to be something more archaic: a yet older (obsolete) verb,
with the older (obsolete) function of the -η ending, ‘was light’,
‘was gleaming’. As with Koller’s solution, we still have to
understand ἦν with οὔτ᾽ ἄρ πω ἠώς, but the presence of a verb
in the second part of the line makes the construction easier.
Nearly the same form survives in the Latin lūcet ‘it is
(day)light’, except that the root syllable here shows a long
vowel. Alfred Bammesberger has satisfactorily explained this as
the result of fusion between the stative *lŭk-ē- and the causative
*louk-eye-, which underlies the rare transitive use of lūcēre.29
Though commonly impersonal, lūcet is also found with diēs as
subject: Sil. Ital. 16. 91 f. en unquam lucebit in orbe | ille dies,
Amm. Marc. 21. 15. 2 lucente iam die. This provides a good
parallel for (ἀμφὶ) λύκη νύξ.30
I take the ἀμφί to mean something like ‘round the edges’. As
dawn approaches, the glow of light spreads round an arc of the
horizon from the east. It makes no difference in practice, and
hardly even in theory, whether we write ἀμφὶ λύκη or ἀμφιλύκη;
I prefer the second, since the Iliad poet, in employing what must
have been a very old formula, will have had no sense of a λύκη
as a separate word that he might have used on its own. Indeed he
may well have taken ἀμφιλύκη to be an adjective.
__________
29
A. Bammesberger, ‘Lateinisch lūcēre’, Glotta 64, 1986, 101 f. Calvert
Watkins, TPhS 1971, 68 f. = Selected Writings i, Innsbruck 1994, 163 f., had
given a similar account but assumed full-grade *leuk-ē- as the stative form.
Bammesberger points out that the zero grade is the norm.
30
Cf. also Plin. HN 2. 100 lumen de caelo noctu uisum est C. Caecilio
Cn. Papirio consulibus et saepe alias, ut diei species nocte luceret; Fronto,
ad Ant. imp. 1. 5. 3 (p. 93. 16 van den Hout) et tu me ad pristinam illam
mensuram luciscentis amoris tui reuocas et iubes matutina dilucula lucere
meridie.
Negitanummius? Suggestion for an emendation
of Apuleius’ Metamorphoses 10.21.8
By BENEDICTE NIELSEN WHITEHEAD, København
The problem: poscinummius versus negatinummius
Apuleius (Met 10.21.8) relates a tender kiss that is neither like
‘the money-seeking kisses of the prostitutes, nor like the moneywithholding(?) kisses of the punters’: “basiola vel meretricum
poscinummia vel adventorum negatinummia”.
This passage contains two compound adjectives, poscinummius
and negatinummius, of an unusual structure; they would appear
to be verbal government compounds of a type that displays a
deverbal first member, followed by a noun that represents the
object of the verb contained in the first member. Such formations, for which I suggest the shorthand P I CK P O C K E T compounds, are rare in Classical Latin: the critical and probably exhaustive survey of Bork (1990: 165–258) boils the inventory
down to approximately 26 examples, of which the present author (2012: 114–151) suggests discarding at least 10; some of
these will be discussed below. However, the type is characteristic of the Romance languages from the earliest attestations onwards1 and is still productive in modern Romance, mainly in the
form of instrument nouns; thus Fr. porte-monnaie ‘wallet’, It.
batti-carne ‘meat-hammer’, Sp. abrelatas ‘can-opener’, Pt.
guardachuva ‘umbrella’.2 This pattern of attestation allows us to
assume that the type was a token of late and spoken Latin, the
__________
1
Save Romanian, where the type is attested, but (contrary to Schapira
1985: 25) where the known examples all seem to be due to a late influence
from the other Romance languages, possibly also from surrounding Slavic
languages. See Nielsen Whitehead 2012: 114–151.
2
See Bork 1990: 42–164 for the most recent and also the fullest historical
survey of Old Italian, Old and Middle French, Old Spanish, Old Occitan, Old
Catalan, Old Portuguese, Sardinian (p. 17 f.) and Romanian compounds of
this type, some dating back to the tenth century.
Glotta 89, 265–275, ISSN 0017–1298
ipabo_66.249.78.36
266
Benedicte Nielsen Whitehead
two formations under discussion here, attested in a secondcentury prose text, being no exception.
Our first example, poscinummius, is undoubtedly one of the
more secure examples of a Latin P I C K P O C K E T compound; Bork
(1990: 187 f.) thus classifies it as “ein zweifelsfreies V[erb]
E[rgänzung]-Kompositum”. The first member clearly is the stem
of the verb poscere ‘to ask for urgently; to beg, demand, request,
desire’, the second is nummus ‘coin’, followed by the suffix -ius.
Since Latin did not allow determinative or possessive compounds in which the subordinate, or ‘determining’, first member
was a verbal stem (the equivalent of Eng. drag-chain or Gm.
Schreibmaschine), the only sensible interpretation of the word is
indeed that of a P I C K P O C K E T type of compound, literally ‘demand-money → money-demanding’.
The context strongly suggests a similar interpretation for negatinummius, clearly the antonym of poscinummius. L&S translate
‘money-denying’, Lindner (1995: 122) ‘Geld oder Zahlung verweigernd’. This form, however, displays the past passive participle of negare ‘to deny, to withhold’ in the first member and is
thus not a morphological parallel to poscinummius, surprisingly
so, if we are to assume, as is likely, that Apuleius coined these
two compounds for the occasion. In fact, its morphology –
(verbal) adjective plus noun, followed by a suffix – is difficult to
reconcile with a P I C K P O C K E T interpretation, but is rather typical of possessive compounds. In the following, I shall review the
various possibilities of bringing poscinummius and negatinummius together on a joint formula.
1. The possessive interpretation
Possessive compounds with past passive participles are licit in
Latin, if not frequent; thus e.g. fissipes ‘cloven-footed’ with the
participle of findere in the first member. If we accepted a possessive analysis for negatinummius, it would be tempting to
speculate that if the two formations do not display parallel mor-
Negitanummius? An emendation of Apuleius’ Metamorphoses 10.21.8
267
phology, it is because the verb poscere does not form a passive
participle, and that posci- in fact stands in for the missing participle. However, a possessive interpretation of both compounds
leads to rather awkward semantics: ‘of desired money’; ‘of denied money’ and is therefore best discarded.
2. Latin pickpocket compounds with past passive participles
in the first member
Bork (1990: 233) argues in favour of a Latin category of P I C K compounds displaying a past passive participle in the
first member and is followed by Lindner (1995: 122). This suggestion meets with serious difficulties as concerns (1) an account of how a passive participle could retain the transitivity of
the base verb and (2) the, as we shall see, scanty Latin material
adduced in favour of such a category.
The closest we come to a satisfactory solution of problem (1) is
an observation made by AiGr 2,1: 276 § 108eα concerning Vedic
possessive compounds with -ta- participles in the first member,
such as hatá-mātar-, literally ‘killed-mother → whose mother
has been killed’. In cases where the person to which the compound refers is identical to the logical subject of the underlying
verb, the meaning becomes similar to that of an agentive compound – as when hatámātar- refers to someone who has killed
his or her own mother. Such an interpretation is of course facultative; and the compound still classifies as a possessive compound. On this background, we are justified to review the examples adduced with a degree of scepticism and to posit a possessive analysis when possible.
Perhaps the most persuasive instance is the enigmatic flexanimus, which Cicero attributes to Pacuvius. In Pacuvius’ Hermione, it is, so Cicero, applied to oratio ‘speech; eloquence’ in a
context which calls to mind the set phrase flectere animum ‘persuade, influence a person’s mind’ (as in Sallust (Jug 102.15.1)
“animus barbari ab amicis flexus” – ‘the barbarian had been per-
POCKET
ipabo_66.249.78.36
268
suaded by his friends.’) 3 and quite strongly suggests that the
word means ‘influential, persuasive’. Thus Cicero:
Sed tantam vim habet illa, quae recte a bono poeta dicta est
flexanima atque omnium regina rerum oratio
ut non modo inclinantem excipere, sed etiam adversantem ac
repugnantem, ut imperator fortis ac bonus, capere pos[s]it.
Cicero (de Or 2.44.87)
But so powerful is it that a good poet rightly called
eloquence “mind-bending and queen of all things”
because not only can it bring down the faltering, but like a strong
and good general, it can even capture the adverse and rebellious.
(Translation mine)
In Classical Latin, we find two further attestations of the word in
the sense ‘mind-soothing; mind-bending’, both based on emendations: Catullus (64.330) and Apuleius (Fl 3.3). From the third
century CE onwards, however, writers on music, like Martianus
Capella (9.906), who took much inspiration from Apuleius, revived it in this sense (Lindner 1995: 74).
It is rather enigmatic how this meaning could be derived from a
word of this morphological shape. Grimm (1826: 955) suggests
a P I C K P O CK E T compound derived from flexare, a frequentative
verb; but this rare verb is not attested with animus, and its meaning ‘to twist and turn’ renders it unsuited for the context. Pisani
(1934: 124) and others before him (see Bork 1990: 229f. for references) have suggested derivation from the perfect stem (as in
1sg. flexī) of flectere, an idea taken up by Bork (1990: 174) and
Lindner (1995: 74), who go as far as to establish a special class
of sigmatische verbale Rektionskomposita, solely on the basis of
this example and a small number of compounds in versi- to be
treated below. See Nielsen Whitehead 2011: 216 for a rejection
of this idea which runs counter to regular rules of Latin morphology, has too few members to be regarded a class and is also
__________
3
Further attestations of this usage in Terence, Seneca and Livius.
269
based on the unwarranted assumption of a significant Greek influence on Latin composition in general.
In spite of the difficulty of accounting for how the word acquired the meaning ‘mind-bending’, I shall accept that such was
its meaning, at least in the majority of its occurrences.
However, a complicating factor appears in the second context in
which Pacuvius employs the word, for here, it clearly has the
meaning of a possessive compound. In a fragment from Teucer
that has been handed down by both Varro (L 7.87.1) and Cicero
(Div 1.80), it presumably describes the dismay of Hesione at the
thought that her son Teucer is or may be dead (Boyle 2006:
101). Cicero cites it in the following context:
Fit etiam saepe specie quadam, saepe vocum gravitate et cantibus
ut pellantur animi vehementius, saepe etiam cura et timore, qualis
est illa...
Flexanima tamquam lymphata aut Bacchi sacris
commota, in tumulis Teucrum commemorans suum
Cicero (Div 1.80)
It often happens, too, that the soul is violently stirred by the sight
of some object, or by the deep tones of a voice, or by singing. Frequently anxiety or fear will have that effect, as it did in the case of
Hesione, who
Did rave like one by Bacchic rites made mad
And mid the tombs her Teucer called aloud
Falconer 1964: Div. 1.36.80
The possessive analysis is in good accordance with the morphology of the word: flexanimus is paralleled by such compounds indicating dispositions or states of the mind as aequanimus ‘even-tempered’, inanimus ‘lifeless’, exanimis ‘out of one’s
mind’ and magnanimus ‘magnanimous’. A similar meaning of
the first member occurs in flexiloquus ‘speaking in tongues, garbled’ applied by Cicero (Div 2.115.6) to the sayings of the oracle
at Delphi (Pacuvius himself refers to the same as flexa). Pisani
(1934: 124) suggests applying the grecist P I C K P O C K E T interpretation, which I have already rejected above, to this context,
ipabo_66.249.78.36
270
suggesting a reflexive use; however, there is no evidence that
flectere animum could ever be used reflexively. Note also the
other passive verb forms employed in the context, which would
tie in well with the passive meaning of flexus: Cicero’s pellantur
animi and Pacuvius’ lymphata and commota. Moreover, a further relevant example, flexipes ‘with pliable feet (about the
ivy)’, is likewise a possessive, as also acknowledged by Bork
(1990: 174f.), although he speculates that it could be reinterpreted as a ‘sigmatic’ P I C K P O C K E T compound.
A number of compounds with versi- in the first member are
classified by Bork (1990: 209) as ‘Zusammensetzungen Partizip
+ Substantiv, als sigmatische VE-Komposita interpretierbar’.
The earliest instance is versipellis which in Plautus (Amph. prol.
123; Bacch. 657) means ‘someone who will alter his appearance
whenever he wants; a turn-coat; a renegade’ and is, as also noted
by Oniga (1988: 162 n. 50), straightforwardly interpretable as a
possessive: ‘who wears his skin inside out’. The same interpretation can be given to versipellis ‘werewolf’ attested in Pliny (Nat
8.80.8) and Petronius (Sat 62.13.4).
In order to reinterpret versipellis as a P I C K P O C K E T compound,
it would, in my opinion, have to be related to the frequentative
verb versare ‘to turn often, keep turning, handle, whirl about,
turn over’. This verb seems semantically unsuited, as a werewolf is not supposed to be a creature whose outer appearance is
in constant flux, but someone who transforms secretly and thus
deceives regular humans. Tellingly, Pliny (Nat 8.80.5) employs
vertere, not versāre, in a discussion of the belief that “homines
in lupos verti”: ‘men may be turned into wolves’. A different
shade of meaning of this word is, however, attested in Gloss. II
475, 18, which translates versipellis as “vertipillo”, indicating
that the present stem of vertere is implied. However, the gloss
claims that the word designates the chameleon, an animal which
deceives its enemies precisely by constantly adapting its appearance to its surroundings.
Assuming that versipellis can thus also mean ‘of ever-changing
shape’ makes it tie in with a few compounds attested from the
271
1st century BCE onwards in which versi- seems to mean ‘multi-,
many-’: the widely attested versicolor/versicolorus/versicolorius
‘of many, varying colours (about feathers and clothes)’ and, in
Tertullian, versiformis ‘of varying, changing shape’. Apparently,
then, the first member versi- changed its meaning over time, going from participial ‘turned inside out’ via ‘two-sided’ to ‘multimany-’.
Is this an indication that the compound was reinterpreted as a
P I C K P O C K E T formation derived from versare? Form and function seem to speak against this: the Latin P I C K P O C K E T compounds are agent-nouns; and Latin agent-nouns are generally
animate. Our forms versicolor and versiformis on the other hand
are attributes of feathers and fabrics.
Most likely, then, versi- as a first member came to be associated
with the adjective versatilis ‘that turns or moves round, revolving, movable’, which, like versicolor and versiformis, is attested
from the 1st century BCE onwards.4 The compounds in versi- are
thus most safely analyzed as possessives.
Finally, tentipellium, which Festus ascribes to Afranius and
Titinius, would appear to have as its first member the past passive participle of tendo ‘to stretch’, as the second pellis ‘skin,
leather’. The meaning is obscure, as can be seen from Festus’
comment:
Tentipellium Artorius putat esse calciamentum ferratum, quo pelles
extenduntur, indeque Afranium dixisse in Promo: “Pro manibus
credo habere ego illos tentipellium”. Titinium autem Verrius
existimare id medicamentum esse, quo rugae extendantur, cum
dicat: “Tentipellium inducitur, rugae in ore extenduntur”…
Festus (354.28–34)
Artorius thought tentipellium was a shoe-last for stretching skins,
and that this is why Afranius said in Promus: “I believe they have a
__________
4
The suffixes -bilis and -tilis form ‘Verbaladjektiva der passiven Möglichkeit oder Wünschbarkeit’ derived from the present stem or the passive
participle (Leumann 1977: 348 § 312). Derivation from the participle only
became productive by the time of Cicero; this is why such an analysis could
not apply to the first member of flexanimus.
ipabo_66.249.78.36
272
tentipellium for hands”. 5 However, Verrius [claims] that Titinius
considers it to be a medicament for smoothing wrinkles, when he
says: “a tentipellium is applied, the wrinkles around the mouth are
evened out”.6
The context indicates that the word has the semantics of an instrument-noun P I C K P O C K E T compound, although it has the
morphology of a possessive compound, just like flexanimus. Pisani (1934: 122) suggests that it developed by dissimilation from
tendipellium, based on the present-stem of tendere. Others have
suggested that tenti- is a mere scribal error for tendi-; thus Bork
1990: 203: “Zusammensetzung mit t-Partizipium an erster Stelle, wenn nicht *tendipellium zu konjizieren ist, das ein zweifelsfreies VE-Kompositum wäre”.
Pisani’s proposed dissimilation is of course purely ad hoc, and
the idea of a scribal error is difficult because tentipellium is attested altogether three times in the above passage in Festus and
once in Gloss. II 196, 54. We shall have to accept the form tentipellium and the enigmatic P I C K P O CK E T semantics suggested by
the context.
In short, having discarded the compounds in versi- we may conclude that the evidence in favour of the postulated class of Latin
P I C K P O C K E T compounds with passive participles in the first
member amounts to three examples: (1) flexanimus ‘persuasive’,
which is also attested as a possessive ‘gone mad’, (2) tentipellium of uncertain meaning and (3) our candidate negatinummius,
the morphology of which is at variance with that of its antonym
poscinummius. Given the difficulties of interpretation and the
lack of comparative evidence it would be biased to conclude that
this material is indicative of any form of regular word-formation
pattern in Latin, rather than a mere coincidence. Under these
circumstances, it is justified to consider emendation.
__________
5 = Afranius (tog 281).
6 = Titinius (tog 173).
273
3. Previously suggested emendations
To solve the above-mentioned problems and to obtain a morphological parallel to the verbal-stem-plus-object structure of poscinummius, some editors have suggested reading negotinummia,
deriving the first member from the stem of negotiari ‘to trade,
do business’. But negotiari is intransitive in Classical Latin; I
therefore agree with Bork (1990: 233) who, with characteristic
resolution, regards this emendation as ‘sinnlos’.
Others have suggested negantinummia ‘money-withholding;
money-denying’, displaying the present participle of negare (literature in Bork: 187). This solution is unattractive first of all because, as noted by Bork (1990: 233), such a formation would be
morphologically isolated (present participles do not occur as
first members of Latin compounds), and secondly, as I have noted above, because the form would not be morphologically parallel to poscinummius. Granted that poscinummius was an acceptable structure, we would have expected that an antonym derived from negare had the shape *neginummius or perhaps
*neganummius (it is not certain what happens to a first-conjugation verb as first member of a Latin P I C K P O CK E T compound);
the formation of a morphologically isolated negantinummius
thus seems doubly unmotivated.
4. A new alternative: negitinummius
Bork (1990: 234) comes close to suggesting a very suitable
emendation of this difficult form when he remarks with regret
that “verben auf -atare nich gebildet werden”; in other words,
the first member of negatinummius is not the stem of a verb
*negatare. Bork seems to be thinking of a frequentative verb.
Whereas frequentatives of verbs of the second, third and fourth
conjugations are formed to the stem of the past passive participle with no modification other than loss of the stem vowel (examples: versare from versus, p.p.p. of vertere, cantare from can-
ipabo_66.249.78.36
274
tus, p.p.p. of canere, tractare from tractus, p.p.p. of trahere),
those of the first conjugation are formed with the suffix -ita-:
thus cantitare from cantare with the participle cantatus and indeed negitare from negatus, PPP of negare.
The formative -ita- adds a frequentative/intensive meaning to
the base verb; negitare therefore means ‘to deny repeatedly; to
persist in denying’. It thus provides a rather suitable antonym to
poscere ‘to ask for urgently; to beg, demand, request, desire’ and
hence fits the context even better than the neutral negare.
L&S classify negitare as ‘very rare’; it is, however, found in five
different classical authors: Plautus, Lucretius, Horace, Cicero
and Sallust. Even if it were not highly frequent, it is safe to assume that the word-formation pattern was productive enough for
speakers to understand it on first encounter.
If we may assume that negitare is at the base of our compound,
we must accept the emendation negitanummius, or perhaps
simply negitinummius, implying that the scribe miswrote -ita- or
-iti- as -ati-. Such an error is plausible enough, considering that
the scribe must have been more familiar with negat-, which occurs in negator, negatrix, negatio, negativus, negatum and of
course the participle negatus.
This solution is attractive, providing as it does a form that is, on
the one hand, morphologically parallel to poscinummius, and, on
the other, a semantically appropriate antonym to that form.
Bibliography
AiGr 2,1
L&S
= Wackernagel 1957
= Lewis & Short 1975
Bork, H. D. (1990): Die lateinisch-romanischen Zusammensetzungen Nomen
+ Verb und der Ursprung der romanischen Verb-Ergänzung-Komposita.
Bonn: Romanistischer Verlag.
Boyle, A. J. (2006): Roman Tragedy. London: Routledge.
275
Falconer, W. A., Ed. (1964): Cicero De senectute, De Amicitia, De
Divinatione. Loeb Classical Library. London, England/Cambridge,
Massachusetts: William Heinemann LTD/Havard University Press.
Grimm, J. L. C. (1826): Deutsche Grammatik. 2, Wortbildung. Gütersloh: C.
Bertelsmann.
Leumann, M. (1977): “Lateinische Laut- und Formenlehre”. Lateinische
Grammatik 1. O. Walter, Ed. München: C. H. Beck.
Lewis, C. T. & Short, C. (1975): A Latin dictionary. Oxford: Clarendon.
Lindner, T. (1995): Lateinische Komposita. Ein Glossar vornehmlich zum
Wortschatz der Dichtersprache. Innsbruck: Innsbrucker Beiträge zur
Sprachwissenschaft.
Nielsen Whitehead, B. (2011): “The alleged Greek influence on Latin
compounding”. Formal linguistics and the teaching of Latin: theoretical
and applied perspectives in comparative grammar. R. Oniga, R. Iovino &
G. Giusti, Eds. Cambridge: Cambridge Scholars: 215–225.
– (2012): Pickpocket compounds from Latin to Romance. PhD dissertation,
University of Leiden.
Pisani, V. (1934): “Presunti composti ‘bahuvrihi’ in latino”. Studi italiani di
filologia classica 11: 121–125.
Schapira, C. (1985): “Les composés roumains à thème verbal”. Revue de
linguistique romane 49 (193–194): 15–26.
Wackernagel, J. (1957): Altindische Grammatik. 2,1: Einleitung zur Wortlehre. Nominalkomposition. Göttingen: Vandenhoeck und Ruprecht.
ipabo_66.249.78.36
Stichwörterverzeichnis
Von KLAUS-JÜRGEN GRUNDNER, Berlin
Avestisch
pairikā 180
Vedisch
hatá-mātar- 267
Griechisch
ajmfiluvkh 253
a[n 37; 40ff; 68ff.
a[n (repeated) 99ff.
gaulov" 188
ge 99
ei[kw 171ff.
ejn o{sw/ 147ff.
ejn w/{ 147ff.
e[oika 171ff.
e[oiken 35
ejpieikhv" 174
e[st’ a[n 147ff.
e{w" 147ff.
i[aspi" 187
kavmhlo" 188
ke 37; 39
leuk- / luk- 254; 260
li'" 188
lukavba" 253
lukhgenhv" 253
melavndeto" 62ff.
mevllw+ Inf. 242
menoeikhv" 170ff.
mevno" 171ff.
mevcri 147ff.
nu'n/ to; nu'n 222
ojfeivlw 1ff.; 21; 30
pavlai/ to; pavlai 222
pallakeuvw 179
pallakhv 178
pallakiv" 178
pavllhx 179
pallikavrion 179, A. 8
(to;) paravcrhma 222
(to;) privn 222
(to;) provteron 222
calbavnh 188
w[felon/ w[feilon 17ff.
w[fellon/ w[felon 16ff.
Lateinisch
cetaria 47ff.
cetarius 47ff.
fissipes 266
flexanimus 267; 269ff.
flexiloquus 269
fuero 136
lūcet 264
negatinummius 265
negitinummius 273
paelex 178
paelicatus 180
poscinummius 265
reccido 126ff.
Stichwörterverzeichnis
rĕcĭdo 126
tentipellium 271ff.
versicolor 271.
versiformis 271
versipellis 270
Irisch (Mittelirisch)
airech 181
Hebräisch
pilegeš, pîlegeš 178
Fachwörter und
Eigennamen
Aktionsart 15; 27f.; 134
aorist 22ff.; 27ff.
Apollo Λυκειος and Λυκιος
258
aspect 22ff.;134
Atelisch 137
auxiliary 6ff.; 21
catégorie d’apposition 195
clausulae 129
coréference 219
counterfactional assertion(s)
16ff; 40ff
deontic meaning 32; 242
deontic modality 2ff.
dislocation à droite 195
emotional 86ff.
emotional context(s) 86
epistemic meaning 32
epistemic modality 2ff.
évaluation 219
277
Exon’s Law 126
future certainty marker 24f.
grammaticalization 1ff
imperfect 18, A. 23; 22ff.;
29
imperfektiv 134
Ionic and Attic 146ff.
late Greek prose 109
Lycian-born 258
mitigation 89ff
modal verb 243
modality 1ff
modality’s sem. space 30ff.
monosemous approach 71;
102ff.
mood(s) 1ff; 32; 36
optative 39f.
Pavlla" 188
participle (Greek) 109
Partizipium statt eines
Verbums finitum 111
Perfektfutur im Altlatein
130ff.
perfektiv 131ff.
pickpocket compd. 267
polysemic 30
polysemous view 71ff.;
102ff.
porte-monnaie 265
potential optative 68ff.
Präsensfutur 130ff.
Septuagint and Hellenistic
Greek 242
present 27
specification 219
subjectification 1ff.
subjunctive 36ff.
ipabo_66.249.78.36
278
Klaus-Jürgen Grundner
temporal adverbs with and
without an article 222
Thucydides, Xenophon and
Plato 146ff.
use of the article 239
uso sciolto e autosufficiente
del participio 115
Forschungsberichte zu den Modi im
Indogermanischen,
zu Hermogenes von Tarsos und zu Avien
Lustrum Band 53 – 2011
2012. 298 Seiten, kartoniert
€ 99,99 D
ISBN 978-3-525-80207-6
E-Book ISBN 978-3-647-80207-7
Einem übergreifenden Beitrag zu den Modi im Griechischen und
Lateinischen folgen kritische Übersichten der Forschung zum griechischen Rhetor Hermogenes von Tarsos und zum lateinischen Dichter Avien.
Gualtiero Calboli untersucht die Forschungsergebnisse der Jahre
1966–2010 zu den Modi des griechischen und lateinischen Verbums.
Pierre Chiron bietet mit »Hermogène 1913–2009« eine kritische
Übersicht über die in fast hundert Jahren publizierte Literatur zu
Hermogène de Tarse, ergänzt durch einen Index. Martin Fiedlers
Beitrag zu Avien, dem lateinischern Dichter der 2.Hälfte des 4.
Jahrhunderts n. Chr., schließt den Band ab.
ipabo_66.249.78.36

Glotta — Zeitschrift für Griechische und lateinische Sprache

Transcription

Similar documents

Anatomically shaped foil for hair coloring

Swiss Food Tech Day, Photonics 4 Food

Johann Sebastian Bach`s Performance Environment

1512-1594 Gerardus Mercator, or Gerard Kremer as he was called

mancher Wortarten (Adverbien, Präpositionen), die Syntax

LOS Sekundar Issue No. 4 - German

Safer Internet DTatort Schulhof - Mobbing, Drogen, Erpressung

(Das) Werk Band (Jahr): 61 (1974) - E

magazine by Alfabet students • June, 2014

ForAllYouveDone - Happenings At Grace Chapel