Bryn Mawr Classical Review

Emilia Yamuza's new book, Tres verbos griegos que significan "deber" en griego antiguo, provides a detailed analysis of the use and meaning of three ancient Greek modal verbs, ὀφείλω, δέω and χρή, all of them conveying the idea of obligation.

Her diachronic analyses, which extend from Mycenaean Greek to the 2nd century A.D., aim at a unified description of each of these verbs, advancing the claim of an underlying unity along the process of their grammaticalization. Working within the framework of Functional and Cognitive Grammar, but never giving in to the temptation of obscure jargon, Yamuza (henceforth Y.) sets up three main objectives: 1) to examine whether the three verbs display deontic and epistemic meanings, i.e. whether their use conveys, respectively, the idea of obligation (ranging from necessity to a mere wish) or of presupposition (ranging from inferred certainty to a mere possibility or even an assertion of the non-reality of something); 2) to investigate the semantic shifts undergone by these verbs; 3) to suggest, whenever possible, a hypothesis of how these semantic shifts occurred, how the changes in the structures took place.

The results are impressive, mainly because of the sophistication with which Y. sets about her analyses, both descriptively and theoretically. There are few books and articles on this specific subject, and Y.'s work will surely call attention to many nuances that are not so easily perceived, thus rendering one more attuned to the semantic detail in the use of Greek modal constructions. All those interested in Greek modality will profit from the book, but the conclusions drawn by Y. will also have a bearing on modality studies in general.

The book is divided into four chapters, to which is added a brief conclusion. The first chapter is an introduction setting out the methodological patterns of the study, while the three remaining ones, each of which devoted to one of the verbs in question, constitute the bulk of the work.

As for the introduction, Y. makes clear her indebtedness to the outstanding book by J. Coates,¹ whose findings illuminate many of her conclusions. Y. follows in the footsteps of Coates on two major issues. First, the categories to which deontic and epistemic modality belong are not organized in polar opposition, but rather by degrees, i.e. the elements expressing one of the modal meanings are articulated in zones of proximity to, and distance from, the so-called prototype -- the best example. There is, in brief, a 'fuzzy-set' in which the transition from one class to the other is gradual, not abrupt. All this bears on the precise identification of an example as conveying deontic or epistemic meanings, that is, on the correct interpretation of a text.

Chances are that, there being a continuum between epistemic and deontic uses, some instances would reveal an ambiguous meaning, where only the context would be able to tell which one is more appropriate, although sometimes not even the context would be of any help, both meanings being acceptable at the same time ('merger'). This is the second point in which Y. is indebted to Coates, and identifying some examples of this phenomenon in Greek literature constitutes one of the strengths of Y.'s book.

It is generally admitted that deontic meanings display an illocutionary force as indirect orders. The prototypic instance would be one in which the speaker, being hierarchically superior, asserts what he wishes to the agent, who is able to carry out a specific action, most suitably a telic one. However, since the category is structured by degrees, there are in the peripheral zone examples which do not follow exactly this pattern, yet fit in with what is reasonable to understand as deontic in meaning. The speaker may, for example, not be a in a position of command, his wish may boil down to a sheer advice; the situation may concern not an action, but a process without specific agent, or even a state; the agent may not be in the second person, but absent or even inanimate, although states and inanimate subjects might point to an epistemic meaning; negatives and past tense verbs may be found in both categories, in spite of the fact that the epistemic use involves a commitment of the speaker to the truth of the statement and is thus naturally rooted in the moment of speech. In short, there is no clear benchmark with which to gauge mechanically the difference between both modalities: the limits are unsteady, not unusually overlapping, while the meaning of a particular form may not be described in isolation from the broader context in which it appears. And, as Y. hastens to point out, the weight of the literary tradition complicates matters still further, in so far as it blocks clear evolutionary paths: the forms show distinct uses in the first stages, only to be replaced by a uniform pattern, which becomes readily fixed.

Notwithstanding these difficulties, in which the standard handbooks are seldom of any help, Y. argues her case convincingly, providing a picture of these three modal verbs which is as intricate as informative. In what follows I shall provide a brief sketch of each chapter's conclusions, stressing what strikes me as most interesting but leaving aside the minute details Y. goes into to prove her point.

ὀφελ-

The description is similarly structured in the three core chapters: Homeric Greek, post-Homeric Greek, and New Testament Greek, to which is added (when data are available) Mycenaean Greek. In each of these chapters, the major insight is the identification of a pivotal example on which hinges the whole description and the hypothesis put forward by Y. for the semantic and syntactic evolution of the respective form (cf. pp. 45-6, 89-90, 144-5).

As is well-known, the root in question may assume various meanings: apart from conveying the idea of an obligation or of unattainable wishes, it may mean either 'to increase', 'to owe a debt', or even 'to sweep'. According to Y. the most basic meaning, common to all of these, is 'to pile up'. This would account for the relationship between sweeping and increasing, in a metaphorical shift of the type 'to pile up is to increase'. There follows a series of further shifts, from 'increasing' to 'owing a debt', and from 'owing a debt' to an obligation of doing something (i.e. deontic necessity), not to mention the shift from the obligation of doing something to something having to be (or not) the case (i.e. epistemic necessity). In all of these there have been metaphorical 'leaps' linking one domain to the other, yet a metonymical understanding (i.e. favoured by the context) must also have played a role in the path that leads from the more concrete meanings of the root to the more abstract ones.

This is most obvious in the transition from 'owing a debt' to the 'obligation of doing something', where the introduction of a purpose infinitive is conducive to a modal understanding of the verb, fostering a future-oriented scenario and lending a dynamic trait to the structure. This infinitive would then be reanalyzed as the argument of the verb, and not as an expression of circumstance: 'I owe 300 drachmae to pay for the house' (purpose infinitive) > 'I must pay for the house (with 300 drachmae)' (infinitive as the argument of the verb).

All of this is fairly well-known; what is new in Y.'s analysis is her hypothesis on how the forms expressing unattainable wishes developed. They are not, as one might expect, a by-product of the shift from deontic meanings to epistemic ones, but follow a different process of change. They result not only from the use of a past tense verb, but also from the original meaning of the root (the notion of a desirable thing is implicit in the idea of 'increasing'). Thus Y. is able to account for the modal use of past tenses, known from various other languages, and justify why this particular root was chosen to convey, among other things, the idea of unattainable wishes, the finite verbal form of which subsequently became fossilized as a lexical item and was treated as an adverb.

Furthermore, Y. makes here many relevant remarks on the use of past tenses and the aspectual distribution of infinitives, and the same applies to the remaining core chapters.

δέω

The basic meaning of the verb is 'to lack'. 'Lacking' is just a step away from the wish of putting an end to it, to the idea of 'having a need'. Both meanings are fairly common cross-linguistically to convey a sense of obligation, from which ensues their use as modal markers of deontic and epistemic modality.

As a modal verb, Y. selects a significant example on which to base her hypothesis of the origin of the form and its subsequent evolution. In most cases, the verb δεύομαι takes a genitive in Homer. The structure may also appear, although rarely, with an infinitive of purpose, as in Il. XVIII. 99-100: ἐμεῖο δὲ δῆσεν ἀρῆς ἀλκτῆρα γενέσθαι 'and [he] lacked me as the protector of his ruin'. The purpose infinitive, as Y. observes, is not yet an argument of the verb, which still keeps its usual genitive ἐμεῖο. Finite verb and infinitive are not perfectly integrated. This structure is subsequently reanalyzed, whereby the infinitive is understood as the argument of the verb. Later still, there follows formal adjustments such as the omission of the genitive and the impersonal use of the verb, as for instance Il. IX. 337: τί δὲ δεῖ πολεμιζέμεναι Τρώεσσιν Ἀργείους; 'what need is there for the Argives to fight with the Trojans?' Here the verb is already construed with the infinitive, which displays a distinct subject in the accusative.

Deontic and epistemic uses in post-Homeric Greek are analyzed by way of a wealth of examples, including instances where it is impossible to say which meaning is most appropriate ('ambiguity'), sometimes both of them being equally possible ('merger'). Y.'s remarks on the issue are particularly illuminating and may lead to a different interpretation of how specific passages are usually understood. I found the most interesting ones on Herodotus 5.31.15 (p. 115), Euripides' Andromache 252 (p. 129), and Sophocles' Philoctetes 416-8 (p. 128).

Much light is shed also on past tense forms. In both deontic and epistemic meanings they may refer either to the past or to the present, interacting with present and aorist stem infinitives to mark (or not) relative time. The picture is complex, but the comments are on the whole quite elucidating.

As for the development of the form, from its most basic meaning to its proper deontic and epistemic meanings, Y. argues for a path marked by conventionalization of implicature, i.e. metonymical or gradual processes which would be able to account even for the meaning 'to ask' assumed by the verb and for the meaning 'to attempt' or 'to order' acquired by the combination οἴομαι δεῖν.

χρή

Here the first stages of development are not documented, so Y. is wary of advancing a full reconstitution, but argues all the same for an underlying unity of the forms.

As we are dealing with a noun, it is natural to regard as more basic the structure with a genitive construction, not with an infinitive. Even in this construction, we may have an accusative referring to the human affected by the need, as in Il. VII. 109-110: οὐδέ τί σε χρὴ ταύτης ἀφροσύνης 'nor this foolishness befits you'. On this accusative hinges the subsequent development. In the primitive accusative plus infinitive construction, so claims Y., the accusative is an object of χρή and the infinitive is epexegetic, rendering clear what the obligation consists of. It stands to reason that the only human mentioned would be also the agent of the action, i.e. that the accusative refers to the agent of the infinitive as well: τί σε χρὴ ταῦτα λέγεσθαι; 'what need (is there) for you to say such things?' (Il. XIII. 275). But once the accusative is detached from χρή and is linked to the infinitive, there must have occurred a reanalysis whereby the infinitive is viewed as the subject of χρή. Later still, the accusative is construed as the subject of the infinitive and detached from the noun: οὐ χρὴ παννύχιον εὕδειν βουληφόρον ἄνδρα 'a decision-making man is not obliged to sleep all night long' (Il. II. 24).

There are some excellent remarks on the use of past tense forms and on examples halfway between deontic and epistemic meanings, such as Pindar's Nem. 5.49 (p. 165s.) and Euripides' Medea 886 (p. 174s.). Negatives and interrogatives receive their fair share of attention, as well as the incipient use to denote 'unattainable wishes', which is common to the three verbs at issue.

A few minor points:

(p. 44) Y. goes to some length to explain why in Il. XVIII. 364-7 the deontic meaning is coloured by epistemic values. Ingenious though her arguments are, they sound to me a bit far-fetched, as she contrives a complex context in which to assess Hera's statement. Perhaps one might read the passage in terms of a difference in illocutionary force. Whereas the standard deontic meanings are 'performative', our example would be arguably 'descriptive' (the terms are Sweetser's).² It might be analyzed as an instance of what Willmott calls 'objective deontic' modality,³ and this would fit in with the interpretation that Hera is at pains to distance herself from the obligation she mentions. The same would apply to Il. I. 352-54, quoted at page 42, but here the 'descriptive' use of the deontic modality by Achilles would be a subtle way of pointing out Zeus' obligation and at the same time of muting the force of his own assertion, lending it a tint of objectivity. Another possible candidate is Aristophanes' Thesm. 216-17 quoted at p. 76.

(p. 133s.) Y. remarks that, as a rule, there are no future infinitives used in the δέω construction. When the meaning is deontic, the future would be redundant, so claims Y., for the infinitive has a purely dynamic value. In the epistemic meaning, where in principle the future infinitive would be perfectly acceptable for inferences about the future, the future infinitive is not used 'probably under the influence of the overwhelming majority of deontic constructions'. I am not able to come up with another answer, but I think this syntactic peculiarity is in need of a better explanation. All the more so because, in past time reference, it is not rare to find modal verb and infinitive being redundantly used in the past, referring jointly to the past time frame or else focalizing the past opinion of the speaker.

(p. 53ss.) Y. draws heavily on an excellent article by Lamberterie.⁴ At times I found it hard to distinguish in Y.'s description where Lamberterie's opinions end and where hers begin. Compare for instance pp. 53 and 55 with Lamberterie's pp. 209 and 212. One might have expected a more generous employment of quotation marks in some passages.

Beyond the irritating presence in the Greek text of an inverted comma whenever an apostrophe was due (for instance δ̔ οὐδέ instead of δ̓ οὐδέ) I have noticed a few typos and other minor slips.⁵ The lack of an index locorum is to be lamented and will be sorely felt by all those consulting the book.

This is an important book which I earnestly recommend. It comes as a welcome contribution in a field (ancient Greek modal verbs) which tends to receive far less attention than it ought to, given its relevance for the correct interpretation of texts. It will be read with profit both by a broad range of classicists and the general linguist (although not the Greek-less one), and is a suitable match for another recent work on modality, Jo Willmott's The Moods of Homeric Greek. One may only wish that Y. keeps writing on the subject to make us more knowledgeable about it.

1. J. Coates, The Semantics of the Modal Auxiliaries (London/Canberra 1983). 2. E. E. Sweetser, From Etymology to Pragmatics (Cambridge 1990): 65-68. 3. J. Willmott, The Moods of Homeric Greek (Cambridge 2007). 4. Ch. de Lamberterie, "Le problème de l'homonymie: les trois verbes ὀφέλλω en grec ancien", in La langue et les textes en grec ancien: Actes du colloque Pierre Chantraine (Grenoble 5-8 septembre 1989) (Amsterdam 1992): 201-217. 5. (p. 25, note 31) As far as I can see, reference should be made to Eckardt, not to Traugott.