Bryn Mawr Classical Review 1999.04.26
Alan J. Nussbaum, Two Studies in Greek and Homeric Linguistics. Hypomnemata 120. Göttingen: Vandenhoeck & Rupprecht, 1998. Pp. 177. ISBN 3-525-25217-X. $49.00.
Reviewed by William F. Wyatt, Brown University (William_Wyatt_Jr@brown.edu)
Word count: 916 words
The two studies concern: (1) the verb ἐάω "leave, allow" and (2) the Homeric Genitive ἑῆος "good, noble" (with an excursus on ἑάων 'good things'). A detailed study of (1) reveals to N. that the verb derives from a Proto-Indo-European (PIE) root *h1weh2- cognate with Latin vânus "empty" and Greek εὖνις "bereft of," and meaning something like (p. 75) "abandon, leave behind, leave alone." (2) ἑῆος is the genitive of ἐύς (< ehéwos) with the second /e/ lengthened (by analogy) and the aspiration brought over from the possessive of the third singular ἑοῖο. Though utilized where one might expect a second singular possessive, it is not itself a possessive.
N. is thorough and informed in his discussions. He canvasses almost all possibilities, and his conclusions are based on very solid research and thinking. They also -- inevitably -- rest on a number of assumptions about language, about Greek, about epic, about PIE. For his etymology ἐάω to be accepted, for instance, one must accept that all PIE roots began with a consonant, in this case the e-coloring laryngeal; and that H1 will be vocalized in Greek initially. The theory of PIE laryngeals is widely accepted, but the only bit of evidence for H1 in this word is in fact the initial vowel in Greek.1
According to A. Meillet -- I think it was -- the only good etymology is an obvious etymology. I would hold in addition to this that a good etymology will require few additional hypotheses: the plausibility of any etymology is diminished 50% or so by any additional hypothesis required to support it. N.'s conclusions fail on the first criterion and are diminished by the second. On the other hand the value of his work rests less in his conclusions (which may, after all, be correct) than in his method and his discussions.
(1) Thus with ἐάω Earlier etymologists gave up on the word, and furthermore held that the original root contained a short [a]. This N., by a thorough examination of Homeric aorists and futures, convincingly proves not to be the case. Initial [s] or [w] had been suspected because of Attic imperfect and aorist in εἰ-, but N. argues (plausibly) that the Attic augment is secondary, and therefore that the word never had an initial consonant: root shape, therefore ewa:-. That shape, laryngeally rewritten, yields the proposed etymological connections. The semantics of the equation appear convincing to N, and are certainly not impossible. On the other hand, it would have been very useful to have had a thorough discussion of the semantics of the comparandum vanus. Is the semantic fit as satisfactory as N. maintains?2
N. does not stop here, however. He discusses (very real) problems with the Ionic aorist -- why is it in long α and not η? -- and with certain presents in Homer. He also handles the odd distribution of iteratives from this root. It is a measure of his intelligence and method that he refuses to leave any uncertainty unclarified. A less thorough scholar might have been content simply to write off peculiarities as peculiarities, and not endeavor to fit everything into a single, neat package. N.'s arguments are interesting, and in varying degrees convincing.3
(2) The case of purely Homeric ἑῆος is even more complicated and more detailed, and contains four parts: (a) the distribution and semantics of ἑοῖο and ἑῆος; (b) a discussion of an all-person reflexive possessive pronoun, together with a (correct) dismissal of Schwyzer's accounting of (c); (c) the history and etymology of ἑῆος; (d) an excursus on ἑάων good things'. I pass over the first with the comment that N. is thorough and convincing. The second issue has been debated long and hard. N. comes down on the side of the possibility of such, but his Greek evidence would indicate (to me) that Greek in its earliest stages possessed only reflexives differentiated as to person.
(c) The discussion here is detailed and thorough, a model of how linguistic investigations should take place. I feel that N. has not made his case, but cannot offer any hypothesis as a competing one -- I have no idea about the etymology of these forms. N. connects both (c) and (d) with a PIE root *h1su seen in ἐύς "good(ly), brave". (c) ἑῆος derives from the root shape *h1seu- which in the genitive with vocalization of the initial laryngeal, yields *ehéwos. The middle vowel is lengthened by metrical lengthening (of one sort or another), and the initial aspiration is brought over by analogy from ἑοῖο.
(d) ἑάων is even more complicated. N. begins with the assumption of an aboriginal phrase "giver of good things (vocative)" *dôtor eswôn, which, though ancient and traditional, cannot fit into epic meter. eswôn > ehwôn (by regular passage of s > h); ehwôn > hehwôn (by anticipation of interior aspiration); hehwôn > hêwôn (by regular and well-attested phonological rule); hêwôn > heêwôn (by diektasis); /w/ of course disappears, and heêôn > ἑάων by analogy with other genitives plural feminine. It is at this point that the archaic phrase can enter (late) into hexameter verse.
This is altogether too much work to account for two forms, but linguists are unhappy with the unexplained. N. knows -- or believes -- that these words are connected with ἐύς but cannot leave things there. He has attempted to show how they attain their Homeric form, and has laid everything out clearly and challengingly. His work will have to be considered definitive on these matters.
1. Additional hypotheses include: metacharacterismos of the Homeric text (p. 55), (more or less unrestrained) metrical lengthening (p.31), Aeolic forms in Homer (p. 68-9), the PIE triliteral root (p.40).
2. N. observes that imperatives of this verb had "a secondary function as interjections of a sort" (p. 49). So ἔα in Aeschylus and later. This seems unlikely, since ἔα is usually followed by a question. But invocation of an exclamation does suggest Modern Greek ας, said to be derived from ἀφές, which means "let" or "let's" -- e.g., ας πάμε "let's go" and (perhaps ας τ' αστεία "cut out (leave off) the jokes." This latter, though, is felt to be the imperative of ἀφήνω (< ãse). It is at least suggestive that Modern Greek has a set of modal particles να ψα ας), two of which (να, ας) can be used as deictic particles as well.
3. Given that some at least of his explanations explain but one form, his resolutions of problems amount simply in fact to acknowledging that (a) there are problems and (b) a solution to these problems can be found. Were I seriously to criticize N. on this work, it would be that he does not lay out at the very beginning the allomorphy of the verb, displaying to the reader all the shapes that the verb in question can show, together with their morphological distribution. Such a display might indicate better than his arrangement has done the (putative) attitude of Homer to this root.