Bryn Mawr Classical Review 1999.11.25
Karin Kulneff-Eriksson, On 'have' in Ancient Greek. An investigation on E)/XW and the construction EI)=NAI with a dative as expressions for 'have'. Lund: Lund University Press, 1999. Pp. xxii, 192. ISBN 91-7966-564-0.
Reviewed by Helma Dik, University of Chicago (email@example.com)
Word count: 1503 words
"'To have' is used in a variety of ways. In the first place it is used with reference to habit or disposition or some other quality, for we are said to 'have' a piece of knowledge or a virtue. Then, again, it has reference to quantity, such as someone's height; for one is said to 'have' a height of three or four cubits. It is used, moreover, with regard to the body, such as 'have' a coat or tunic; or part of it, as a ring on the hand; or a body part, like a hand or a foot. Or as in a container, as in the case of a vessel [having] wheat; or of a jar, wine; for a jar is said to 'have' wine, and a medimnos, wheat. All that is said to be 'had' as in a container. Or it refers to a possession; we are said to 'have' a house or a field. We are also said to 'have' a wife, and a wife a husband, and this appears to be the most remote meaning of 'have', for by it we mean nothing but 'live together'. There may be other uses of 'have', but the usual ones have been enumerated just about completely."
This quotation does not come from the book under review -- what need for a book if everything can be listed in one paragraph? As readers will have suspected, this is a quote from Aristotle (Categories 15b18-end). Does modern linguistics have something to add to Aristotle's insights? We have certainly waited a long time for a modern treatment of 'have' since the appearance of Kahn's μέγα βιβλίον on the verb 'be' in 1973.1
This book, the author's Lund University (Sweden) dissertation, focuses on ἔχω 'have' in ancient Greek, and to a lesser extent on its rival construction ἔστι μοι. Kulneff-Eriksson starts out from a typological perspective: The world's languages can be divided into 'have' and 'be' languages, on the basis of their expression of possession. 'Be' languages are in the majority, i.e., most languages do not have the option of a transitive verb like 'have' to express possession. K. goes on to say that the ramifications of a 'have' construction are considerable. 'Have' lends itself to adoption as auxiliary, and this tendency is evident in Greek as well: its uses as modal (ἔχω + infinitive) and temporal, or rather aspectual, auxiliary are well known.2 But this study is not primarily concerned with auxiliary uses of ἔχω. It is the transitive verb denoting possession that forms the focus of the inquiry.3
Since Indo-European languages do not have a 'have' verb in common, the conclusion imposes itself that all 'have' verbs are language-internal developments. What tends to happen is that a verb meaning 'hold' or 'keep' develops a less concrete use as 'have'. The result is that possession can be expressed in two different ways: the standard 'be' plus dative, and the new transitive construction, with the latter becoming dominant over time.4
Naturally, it is difficult to delimit precisely where 'hold' or 'keep' stop and 'have' begins. Moreover, focusing on the correct translation equivalent in another language may compound the difficulties. No matter how popular the practice once was, it cannot be assumed that English verbs in caps or quotes can provide a universal, basic semantics that is in no need of further analysis. K.'s quandary over distinguishing 'have' and 'get' (she seems surprised that (p. 3) "there is no particular verb for expressing the notion of 'get'" and that (p. 4) "Modern Greek still has not developed an equivalent of 'get'") seems to arise from a need to have things work the way they work in some Western European languages, rather than from an appreciation of the aspectual system, which of course is very much alive in Greek to this day. (The importance of aspect and its interplay with lexical meaning is largely ignored, cp. the observations on the distribution of present and aorist forms over verb meaning like 'keep' and 'stop', p. 41, which are doubtless correct but are left unexplained.5)
Given that the determination of verb meaning in particular instances is so complicated, criteria should be given that distinguish 'have' from 'keep' and 'hold' on the one hand, and 'get' on the other. The principal differences seem clear enough semantically: 'keep' and 'hold' imply control on the part of the possessor which is absent in the case of 'have' (and 'get'), 'get' implies a change of state which is absent in the case of 'have'6 (and 'keep' and 'hold'). Unfortunately, there will not always be sufficient evidence whether 'control' is involved or not. Does a position imply control, in other words, is it "I keep my valuables in the safe" and not "I have"? How do we know that a Greek would draw the line where an English speaker does?7
A look at the OED, where the entry for 'have' amusingly does refer to Aristotle's ancient wisdom, will suffice to illustrate the complexity of English 'have'. Where does that leave us when we start looking for Greek equivalents? K. never quite makes up her mind whether to start solely from form (all instances of transitive ἔχω and ἔστι μοι in her corpus) or function (predication of possession). Rather, she considers the translation equivalents of transitive ἔχω and discards those instances where she interprets the verb as something other than 'have'. The few general rules she goes by are not all unproblematic. K., I think rightly, decides to discard as 'hold' instances of ἔχω where hands occur as an instrument (e.g., Il. 20.420 ἔντερα χερσίν -- presumably conscious containment) or with an adverb like 'firmly' (p. 31)). But why also discard locative ἐν χερσί ('had/held wreaths in his hands'), when other locatives that suggest more strenuous effort (had/held his shield in front of his chest) are classified as 'have'? It seems that K.'s method of relying on translation equivalents is shaky.
There are alternatives: When one investigates the two rival expressions for 'possession', it seems that clauses that actually express possession rather than location should be the focus of attention. This would imply that all clauses with location expressions should be discarded, leaving 'Aeneas has a shield', 'Those weapons belong to Ajax', and the like.8 Expressions of the sort 'Wrath has Achilles' would not enter into this.9 The principal question would be what factor or factors determine the choice between ἔχω and ἔστι μοι when it comes to expression of possession.
Another option would be to take all instances of transitive ἔχω and focus on the kinds of objects that can undergo ἔχειν, on the part of what kinds of agents. The result of this kind of lexical study typically is a description in terms of prototypical agents and patients, with a set of meanings that are not subject to rigid classification that leads students to the question 'is this ἔχω IIIa2 or ἔχω IIa3', but rather acknowledges that the range of meanings can be placed in a (multi-dimensional) continuum.10 In this case, ἔχω 'hold' would simply be a part of that bigger picture (which eventually would also include non-transitive uses of the verb) and a boundary zone could be identified where 'hold' meets 'have'.
Fortunately, the methodological problems do not mean this study has nothing to offer students of Greek. Chapters 5-11 on 'have' in seven different samples ranging from Linear B e-ke to Isocrates at least begin to give an answer to the question raised above. K. classifies the possessors and possessions in ἔχω and ἔστι μοι 'have' constructions in terms of a hierarchy that ranges from female (sic: the pressure of distinctive features took its toll), human, to animate, to concrete and finally non-concrete referents. This will be familiar to some as resembling an 'animacy' hierarchy.11 In addition, definiteness of the possession is taken into account.12
The picture that emerges from K.'s examination of the evidence is one that those familiar with issues of grammaticalization will have come to expect. The rise to prominence of ἔχω starts with concrete objects that can be conceived of as 'held' or 'kept', and ἔστι μοι survives longest with non-concrete, indefinite possessions. It is perhaps too much to ask for more than conclusions that go beyond statements of statistical preference for one construction over the other given particular types of possessor and possession. However, it does come as something of a surprise and a disappointment that the author shows some awareness of pragmatics as a possible determinant (p. 18-20 and again p. 144 and following) but does not follow up on this beyond the observation that there is a correlation between non-definite and focus (the word topic only occurs in a header on p. 18). The conclusion suggests itself that it is up to the next person who takes on this issue to delve deeper into the factors that determine the choice between ἔχω and ἔστι μοι. The present work makes clear that there is still a lot that we don't know, despite Aristotle's and Kulneff's efforts.13
1. Kahn (1973), The verb "be" in ancient Greek (and specifically his section on the possessive construction of 'be', pp. 265f.), is only one of a number of relevant studies not cited. See below.
2. However, when she states (p. 3): "There are examples of ἔχω as a temporal [or rather aspectual: resultative -HD] auxiliary already in Greek of the classical period", with due reference to W.J. Aerts' Periphrastica (1965) in a footnote, it should be borne in mind that any suggestion of continuity is misplaced. Rather, ἔχω + aorist participle disappears from the scene and much later, different constructions develop, again with ἔχω (Aerts, 128ff., esp. pp. 159-160).
3. Chapter 3 is devoted to "ἔχω not meaning 'have'" -- Section A sets out K.'s criteria for distinguishing the other transitive uses from 'have'; section B discusses active intransitive uses. K. notes that intransitive ἔχω is usually found combined with adverbs. She notes separately (p. 34 n.8) the instances where it is combined with adverbial prepositional phrases, and finally includes Od. 9.6 as "no adverb" -- this in fact should be treated as a case of κατέχω with tmesis (LfgrE, s.v. ἔχω II7aγ). Sections C and D discuss intransitive and transitive middle-passive forms of ἔχω, respectively. In section E, 'Auxiliary ἔχω', criteria are again unclear. Most importantly, K. does not indicate where she departs from Aerts in her treatment of ἔχω plus participle (see also (p. 16) her treatment of εἶναι plus participle at Hdt 8.46.1, which later leads to the claim that the ἔστι μοι construction is not used with ships as possessions in Hdt. 8). Secondly, we read that ἔχω as auxiliary occurs three times in her sample from the Odyssey, but citations are not given in the text or in the (otherwise full) index locorum (LfgrE, s.v. ἔχω I1aγbb gives Od. 11.584 and 12.433, among others). The auxiliary use occurs in the Iliad as well; in the case of such rare uses of a verb reference should be made to mitigate statements, based on the limited sample, that auxiliary ἔχω has not been found in the Iliad.
4. I see no references to typological studies of possession, such as Hansjakob Seiler (1983), Possession as an operational dimension of language and especially Bernd Heine (1997), Possession. Cognitive sources, forces, and grammaticalization. Heine objects to the characterization of PIE as a 'be' language for the reason that no 'have' has survived in the later languages. He argues that 'have' can have a very short life cycle, going from esse to habere to tener/ter in the documented history of Spanish and Portuguese. For Latin, see B. Löfstedt (1963), 'Zum lateinischen possessiven Dativ', ZVS 78: 64-83; A.M. Bolkestein, 'Dative and genitive possessors in Latin', in S.C. Dik (ed.) (1983), Advances in Functional Grammar, 55-91. H. Pinkster notes (Sintaxis y semantica del Latín: (1995), Appendix, p. 339) that in Latin the two constructions are not interchangeable: "En el periodo clásico predominan los sujetos abstractos en la construcción de dativo possessivo, siendo relativamente poco frecuentes los accusativos abstractos en la otra construcción. En el latín plautino son normales los sujetos concretos".
5. Chapter 4 deals with 'have' in the different tense stems and suffers most from the confusion over aspect, and the unclear boundaries of English 'have' and 'get', which should not enter into the argument. Greater familiarity with the narrative uses of the different stems would have been helpful in the interpretation of, e.g., Hdt 4.42.4 (p. 16), 8.138.3 (p. 43). In the discussion of the future, Symp. 205c5 is quoted, where Burnet (cited on p. xv as the edition used) in fact prints a (generic) present.
6. The OED does of course recognize an inchoative 'have', as in 'have a baby'. On reflection it becomes clear that 'get' in English does seem to allow for some measure of control. This is yet another reason to avoid describing one language in terms of another. For classifications of verb meanings, Z. Vendler (1967), Linguistics in Philosophy would be one place to start. See in addition A. Rijksbaron (1989), Aristotle, verb meaning and functional grammar.
7. In English, at least, 'keep' has an additional suggestion of permanence and principle: "I try to keep an umbrella in my office, but I have it in the car right now". And this last example brings us to an additional problem. "I have an umbrella in the car" does not really express a relation of possession -- this statement is not equivalent to "I possess an umbrella, which is located in my car", but more likely to be used when you find yourself not having an umbrella when you actually need one, or when you are rejecting the advice to take one.
8. Strictly speaking, indefinite objects and locative expressions may still denote possession: 'She has a cottage in Wisconsin' denotes possession, not temporary position, but this group should be assessed on a case-by-case basis (see above n. 7, umbrellas in cars). Definite objects will denote temporary position rather than possession: 'Achilles has his shield in front of him/to his side'.
9. Since F. Mawet (1979) other studies on these expressions of emotion have appeared: see for a more recent study and references, A. Rijksbaron (1992), "D'où viennent les 'algea'", in F. Létoublon, La Langue et les textes en grec ancien: actes du colloque Pierre Chantraine, pp. 181-191, and by the same author, 'Further observations on expressions of sorrow and related expressions in Homer', In: E. Banfi (ed.), Atti del secondo rencontro internazionale di linguistica greca, Trento 1997, pp. 215-242.
10. For this approach, see for instance R. Langacker (1987), Foundations of Cognitive Grammar. B. Heine (above, note 4) adopts a prototypical approach for the notion of possession, using five criteria (p. 39): (1) Possessor is a human being (2) Possessee is a concrete item (3) Possessor has the right to make use of the possessee (4) Possessor and possessee are in spatial proximity (5) Possession has no conceivable temporal limit. The various types of possession can be characterized as conforming to some or all of these criteria.
11. Some familiarity with the literature on animacy/empathy/topic hierarchies would have paid off, starting with, e.g., B. Comrie (1989), Language Universals and Linguistic Typology. See above note 4 for recent typological studies.
12. Definiteness and genericity (not mentioned) are not distinguished, however, i.e. definite articles are simply taken to denote definiteness, e.g. the citation (p. 19, n.43) of Hdt. 4.192.1 ὄνοι, οὐκ οἱ τὰ κέρεα ἔχοντες). The distinction of alienable vs. inalienable possessions does not enter the discussion either.
13. It is merely symptomatic that no reference is made to works dealing with pragmatics, no definitions of topic or focus are given by K. herself, and all but one of the examples used in the section on topic and focus (pp. 18-20) are made up. The work has an annotated index locorum, a bibliography that is strongest on the historical-comparative aspects, weak on general linguistics, specifically semantics, syntax, and pragmatics, and linguistic typology (see the bibliography in B. Heine (1997) for up-to-date references). I have not noticed misprints, but the frequent oddities in the English text could have done with a sterner editorial hand.