You are on page 1of 6

Homeric : two studies

FT Copenhagen (Sept. 2012) Timothy Barnes, Harvard University 1. Komplexivkomposita in -: , , 1.1. Distribution: 1.1.1. has the widest distribution, cf.: d. Fazit. Appears to be a traditional device in short narratives. This explains why it A 53 | Z 174 M 25 1.1.2. Then, a predilection of : 1.1.4. Od., hymns. Od. poet expands on this use in Odysseus narrative of his wanderings, e.g.: 28 Then, on this model, cpd. in: 1.1.3. Analysis. a. 4 of the 7 instances are in direct speech. This is in part an effect of use in selfcontained/telescoped narratives. Today: Today What is and what is its relation to the formula b. Three instances involve reports of events occurring completely outside the temporal framework of the main narrative: Z 174 appears in the story of Bellerophon; 610 in that of Niobe; and M 25 in a flash-forward to the future destruction by Apollo and Poseidon of all traces of the Achaean siege. These stories are presented as self-contained units. 1.2. Two basic approaches 1.2.1. Wackernagel (1910:14 [= 1953:8336]): univerbation of an adverbial phrase * pluralic seen in . ; further evidence for ? 80 Further hy.Ap. 91 | | | . appears only in reported narratives (or description of a future event) bzw. at the very beginning (A) and end ( ) of the poem as a whole. The internal elaboration of the Iliad story in effect deprived the poet(s) of any reason to frame scenes in terms of the telescoping model for n days, and then on the (n + 1)th. So, despite the distribution in late-ish contexts (speech, probably older) narrative technique. 104 | , the Hesiodic opening of M with its list of rivers) not a recent elaboration of the Iliad tradition, but part of a simpler (and c. In the other cases the information tagged with either gives the background for a further event which takes place on the tenth day, or may be generally understood as telescoping the narration. 664 is more complicated, as it gives Priams answer to the question: how many days ( ) do you plan to mourn Hektor?

610 ( 612 ) 664 | 784 | .

below, but worth pursuing just to be sure] that the preconsonantal syllabification Objections: a. The phrase itself may have a more complicated prehistory. 2.2.1.1. Could ** have been remodelled to as an inverse or false Aeolism? To my knowledge, never been proposed, and with good 1.2.2. Leumann (1950:99-101): is an old Komplexivkompositum aka dvigu (a tatpuru a compound with numeral first member (P.2.1.23; 52), grammatically singular (P.2.4.1)). Plural instead a license based on reanalysis of as consisting of numeral + apparently plural hence the remarkable pseudoarchaisch hy.Ap.91 . 2.2.1.2. Could ** instead have been redone as on the basis of the cardinal? Since the ordinal was never adjusted to Leumann basically right, but details in need of further explanation: - still need to explain the precise form of the compound. - also need to explain how the reanalysis he proposed was motivated. 2.2.2. 2.2.2. Before V-. Expect *h1nun- > *enun-. The original form of the ordinal ought study: 2. First study the form of the compound . 2.1. Hardly possible that we are dealing with a contraction over -- hiatus ** .1 > ** > . For this reason alone cannot be a univerbation.
n. Some WGk. dialects have but for obvious reasons they are irrelevant here
1

*h1n > *en- was generalized everywhere, we might expect ** .

b. The form of cannot derive straightforwardly from such a univerbation.

reason. No credible examples exist of such a process in Homer (e.g. no Homeric ** for , etc.). Would further entail dissociating the geminate -nn- of our form from the famously problematic -nn- of the cardinal, which would be absurd.

match the geminate of the cardinal , it also seems unlikely to suppose such a remodelling in the case of .

to have been *h1nun-o- or *h1nen-o- (: L. nnus), replaced at some point by *h1n-to- (on the process see Rau 2009:2232 after Peters 1991:30414). Possible to extract *enun-V- from such instances, which would explain *enunkonta > (thus e.g. Waanders 1991:376) with an adjustment of the anomalous uvocalism (cf. h).

2.2. Instead, it seems certain that we have in the first member the compositional form in the zero-grade of the numeral 9. Distinguish environments: C2.2.1. Before C-. The clearest example of the reflex of the zero-grade of the numeral 9 in Gk. is found in Homeric < (transposed) *h1n-ets and the ordinal < *h1n-to-. On the assumption [which is not warranted, see H2.2.3. Before H-. Distinguish HC and HV-. HC-: From *h1n-h1konth2 in fact expect *ennkonta. According to Peters 1991, we simply lose the in this unusual environment, yielding (*)ennkonta, and then extract *enn-V- from this. HV-: *h1n-h2eh1m would >**enan-mar. Replaced by *enn-mar, acc. to Peters, after extraction of *enn-V-.

N.b. the majority of such compounds in Greek and IIr. are in fact neuter substantivizations of thematic derivatives (e.g. Gk. Cypr. pe-pa-me-ro-ne (ICS 220b.2), (Xen.), (Arist.); Av. ri.aiiarm ri.xaparm; etc).

2.3. Closer look at Peters 1991. Phonology: *h1n-h1omth2 > *ennkonta > ennkonta. Peters suggests in a footnote lautgesetzlich disappearance of interconsonantal . As examples he adduces the development *(e)t-C >*(e)tr-C >*(e)tr-C, which is problematic, since it ignores the usual explanation, whereby the is dissimilated after , with the variant *(e)tru- playing a key - if difficult to define - rle. His other example is poll < *pol < *pol. But it is possible that the development of poll- illustrates only what occurred in the special environment VlV, and not a general rule pertaining to all cases of interconsonantal . But then again no less likely that the awkward -nn- would develop in a special way too. Morphology: acc. to Peters, then, is the lautgesetzlich continuant of the Proto-Gk *h1n-h1omth2 (itself the product of several small analogies, see Rau 2009:22). (Though, to be sure, the Odyssean hapax ( 174) does not look particularly old, as Peters himself agrees 1991:30829). To get common Gk need to set up template *hebdmkonta, *ogdokonta, *x-konta, where x is solved as disyllabic *enn- made on the basis of monosyllabic *enn-, and *ennthen > enen-. But if 90 was remodelled at a time when the ordinal was still *enuno-, not difficult to get a sequence *hebdmkonta, *ogdokonta, *enunkonta, and invoke a similar vowel assimilation. In all, Peters account has a good chance of being right, but still worth wondering if there might be any other ways to get there. 3. Holzweg: Analogy with 4. 3.1. Possible to get a similar result via a slightly different path through a morphological analogy. In what follows we observe the behavior of 4, and mechanically perform the same operations on 9.

a: *(e)t-C (>*(e)t[]-): -; b: *(e)tur-V/C : Ved. catu -pad-, U. petur-pursus; PN (* , based on ordinal *tur-to- cf. Av. triia-). c: metathesized *(e)tru-C : Av. caru-gaoa-; . (Here too perhaps Hitt. kutruan- witness (e.g. Eichner 1991:802), if < *(e)truon- (vel sim.)). a*h1n- > *en-C : b*h1nun- > *enun-V(/C) : possibly basis for *enun-konta > cmetathesis *h1nun-C > > *h1nnu-C > *ennu-C : so, ex hypothesi, *ennuphaleia, metathesis *ennupedza, sim. 3.2. 3.2. Possible scenario starting from *ennu-. Within Gk we observe a tendency for the archaic compositional form of 4 in *tru- (bzw. *tur-) to be replaced by *(te)tr, which before V- then appears as *-tr-V. Might suggest that hypothesized *ennuremodelled further to *enn- on this model and with help from the variant *en- as well as sequence *hept-C, analogical *okt-C, and *dek-C giving *trpedza, *etrpod- : *ennpedza, *ennpod-. Would create basis, eventually, for both *ennmar > and *enn-konta (in a dialect which had *tetrkonta vel sim.). Sets up the conditions for learners to extend the geminate of the compositional form: *ene > enne. Once this has happened, the compositional forms *enna-C bzw. *enno-C had the appearance of forms of the cardinal missing one syllable, and were updated to ennea- / enneo- (as Myc. e-ne-wo-pe-za). Forms in enn-V did not e e have the same appearance, and hence were never updated. 3.3. 3.3. Obvious weaknesses:

3.3.1. Need to set up intermediate stage with two preconsonantal forms *en-C and *enn-C. Would need to show (a) positive evidence for *enn-C-, (b) reason for generalizing *enn-V and not *en-V-. Ad (a): Worth noting Hesiodic , hapax at op. 436 ,

5.0. 5.0.

is a real compound and is relatively old. The form thus shows that a

nucleus of Komplexivkomposita in - existed which were liable to be reinterpreted by the poets as a sort of univerbated numeral plus plural and hence that, on this point, Leumann was right and Wackernagel wrong.

which looks, on the face of it, like the missing link *enn-C. Hard to trust the form, however. Explanations of dismissed above (viz. i. analogy with cardinal; ii. pseudo-Aeolism) perhaps slightly harder to discount; but more importantly, we cannot rule out a transmissional or redactional error here (early enough, though, to infect the entire tradition; note that erroneous writings like , occur sporadically in medieval mss.). Ad (b): Could be motivated by assumption of cases of synchronically irregular *enun-V-, happily replaced by *enn-V, which looks like a version of *enun- with syncope of the synchronically strange u-vowel. 3.3.2. Why should 4 have exerted such an influence in the first place? : No inherent connection between the numerals, rather relative frequency of type four-legged et sim. vs. nine-legged, which latter occur rarely in the world and are thus likely to be ad hoc creations, for which the model will have been precisely the compounds with 4 in the first member.

5.1. 5.1. Formula

8 X Homer (E 490, X 432,

73, 345, 28, 80,

476, 63), always at verse end, e.g.: E 490 5.2. 5.2. A few things to notice: 5.2.1. Appears with type in three of the five Odyssean instances, but never in the Iliad, e.g.: 28 On this basis, the last Odyssean instance: 63 and thence hy.Ap. . |

We have here before our eyes the Leumannian scenario, whereby the poets begin to 4. Conclusion to first study. Peters explanation makes fewer unverifiable assumptions, and hence is likely to be correct, but a set of analogies cannot be completely ruled out. With this more detailed picture of the prehistory of , it is possible to tackle the issue of the relation of this type to the formula. . 5.2.2. In other positions of the verse the formula has a different shape, namely verse internal 3
4

interpret n- as univerbated numeral plus plural .

( 340, 745) and enjambed #1 2 3 (

186). The internal evidence of the formulaic system shows that the poets were already interpreting as plural in this phrase before the appearance of the Odyssean collocations which show Leumanns scenario.

study: 5. Second study The relation of the Komplexivkomposita and

5.2.3. Leads us to wonder whether the verse-end formula might not be concealing an earlier or rather *nuktas kahi mr (vel sim.) with the archaic neuter plural of the word for day preserved in an adverbial phrase. We could then explain the replacement of *mr by mar by appeal to a kind of first round of the Leumannian scenario. 5.3. But first we need to establish independent reasons for assuming existence of *mr to start with:

inexact, would expect /mr mati/ (and cf. Cypr. a-ma-ti-a-ma-ti (ICS 318:2)); (b) expect a singular in mre itas to start with (see material in Dressler 1968:39f, Klein 2003:773ff.) . b. Another way to do it: take well-formed mre ita *m-m as a phonological word >*m-mar by Garca-Ramns Law (early change of absolute final syllabic to -ar, as in Vedic) > /mor-mar/. 5.3.2. Armenian awr and awr awowr.

5.3.1. Myc. a-mo-ra-ma 2X identical context: KN Am 600+665+8307 .a .b ko-no-si-jo , / KN Am 601 .a e-te , e-so-to , a-mo-ra-ma VIR 9 .b to-so , / a-mi-ni-si-jo - template: e-te es(s)ontoi a-mo-ra-ma tos(s)oi ethnikon VIR n. There will be so many e-te a-mo-ra-ma - a-mo-ra-ma universally assumed to be an mre ita meaning daily, day by day (vid. DMic. I s.v.). [e-te quite unclear] Possible analyses: a. = /mr-amar/, acc. to Peters (1980:243195a and 329) replacing *mr-mr. Supposedly parallel Arm. awr awowr (with dat.-loc. awowr replacing earlier awr) and Ved har-divi, with divi in place of ahar (or ahan?). But (a) parallels are to-so , e-te e-so-to , a-mo-ra-ma VIR 25

a. awr awowr synchronically clear, could be made at any time within Armenian (Tag fr Tag), not really relevant to the issue of the Myc. mre ita. b. But relevant in a general way, as awr < *mr. Might think of reanalysis as a singular, as in mass nouns ( , etc.), where collective reading dominated, and was contrasted at some stage with an externally formed count plural . But day hardly a mass noun. c. In fact, just an inner-Armenian process: old plural *mr >> *mres (> awowrk), whence back-formed singular *mr > awr. Hence, indirect evidence for *mr as a grammaticalized count plural in Balkanindogermanisch. 5.3.3. Adverb by night further indirect evidence for *mr (and precisely where we would like it!).
So Barnes 2006 and (ut mihi aliquantulum postea innotuit!) Leukart 1987:359.

a. For collocation see esp. Hes. op. 1768:

b. Presumably contamination of the type common in such phrases: cf. e.g. L. noct likely after di (though we may have a real u-stem here too: Go. unhtw, etc. but see Pinault 1990:189 for the suggestion that all such u-stems go back to a contamination with *d(i)e of PIE date), Go. nahtam jah dagam (L 2.3 etc.) for *nahtum, OCS notij i dnij (Supr. 292.24), Alb. dit day possibly with -t after nat night. Within Gk. probably after . c. Of course possible to invoke r-locative and various hypostases made thereto. For night, , , L. nocturnus, YAv upa.naxturuu. Difficult to assess, since can be after , nocturnus after diurnus (possibly < *diesino- and hence no -r- at all), and the Avestan might be similarly accounted for after r/n stem time words such as azar- or aiiar-. 5.4. To sum up: - A priori we reconstruct *mr as the inherited plural of the neut. *m < *h2h1m. - *mr certainly existed going into Armenian. - *mr is likely the source of nuktr. - Hence, possible and attractive to reconstruct *nuktas + mr. - Finally, invoke Leumannian scenario for poets to get rid of anomalous verse end *mr with metrically identical *mar given the presence in their repertoire of adverbs of shape *ennmar (and potentially *tetrmar, *heptmar etc.). A few generations later same association would lead to the Odyssean collocations and similar effects of plural in the hymns. 6. Mange tak! tbarnes@fas.harvard.edu

Bibliography: Barnes, T. 2006. Homeric , Paper delivered at the 137th Annual Meeting of the American Philological Society, Montreal, Jan. 2006. Dressler, W. 1968. Ved. dive-dive und die idg. Iterativkomposita, in Studien zur Sprachwissenschaft und Kulturkunde: Gedenkschrift fr Wilhelm Brandenstein, ed. Manfred Mayrhofer (Innsbrucker Beitrage zur Kulturwissenschaft 14), Innsbruck, 947. Eichner, H. 1991. Anatolian, in J.Gvozdanovi (ed) Indo-European Numerals, Berlin 1991, 2996. Klein, J. mreitas and related constellations in the RigVeda, JAOS 123, 773802. Leukart, A. 1987. po-ra-qa-ta-jo, to-sa-pe-mo, a-mo-ra-ma and Others: Further Evidence for Proto-Greek Collective Formations in Mycenean and Early Alphabetic Greek, Minos 20-22 [FS Chadwick], 343ff. Leumann, M. 1950. Homerische Wrter. Basel. Peters, M. 1980. Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen. (SbAW 377) Vienna. id. 1991. Idg. 9 im Armenischen und Griechischen, ZPSK 44, 30110. Pinault, J-G. 1990. Notes sur les manuscris de Maitreyasamiti, TIES 4, 119202. Rau J. 2009. Indo-European Nominal Morphology: The Decads and the Caland System. Innsbruck. Waanders, F.M.J. 1991. Greek, in J.Gvozdanovi (ed) Indo-European Numerals, Berlin 1991, 36988. Wackernagel, J. 1910. Zur griechischen Wortlehre, Glotta 2, 18. id. 1953. Kleine Schriften III. Gttingen.

You might also like