Two Studies in Greek and Homeric Linguistics
 9783666252174, 352525217X, 9783525252178

Citation preview

HYPOMNEMATA 120

V&R

HYPOMNEMATA UNTERSUCHUNGEN ZUR ANTIKE UND ZU IHREM NACHLEBEN

Herausgegeben von Albrecht Dihle/ Siegmar Döpp/Hans-Joachim Gehrke/ Hugh Lloyd-Jones/Günther Patzig/Christoph Riedweg

HEFT 120

V A N D E N H O E C K & R U P R E C H T IN G Ö T T I N G E N

ALAN J. NUSSBAUM

Two Studies in Greek and Homeric Linguistics

VANDENHOECK & RUPRECHT IN GÖTTINGEN

Verantwortlicher Herausgeber: Hugh Lloyd-Jones

Die Deutsche Bibliothek -

CIP-Einheitsaufnahme

Nussbaum, Alan J.: Two Studies in G r e e k and H o m e r i c Linguistics / A l a n J. N u s s b a u m . G ö t t i n g e n : V a n d e n h o e c k u n d R u p r e c h t , 1998 ( H y p o m n e m a t a ; Η. 120) I S B N 3-525-25217-X

© 1998, Vandenhoeck & Ruprecht, Göttingen. Printed in Germany. - Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmung und die Einspeicherung und Verarbeitung in elektronischen Systemen. Druck: Hubert & Co., Göttingen

Table of Contents

All about έ(/:)άα> 1. 2. 3. 4. 5. 6. 7.

Introduction The aorist-future stem(s) and the root final The root initial The root shape The Ionic Aorist and future Three peculiarities of the present in Homer The etymology of ϊάω

9 11 14 39 44 46 52 73

Good for You: The Homeric Genitive ίήος in Second-Person Contexts (with an excursus on ίάων 'good things') 85 1. 2. 3. of

Introduction eoto and e?)oy as variants The question of a different distribution ΐήος and kolo earlier on

87 89 105

4. Excursus: ίάων 5. An account of ef/os

130 146

Bibliographical Abbreviations

160

Index of Words and Forms

165

Index of Homeric Lines and Passages

172

For Suzanne qua uiro indulgentior num. est uxor usquam gentium?

Foreword

The two pieces combined in this volume are both studies in Greek historical and comparative linguistics and in Homeric language, diction, and philology—though to differing degrees. The first of them, "All about etc. If this kind of hyper-Aeolicizing transliteration could produce a nonexistent (~)έασσί(ν) 'are' from ambiguous (-)ΕΑΣΙ(Ν) in a few manuscripts in 12/20 Iliad instances and 19/22 occurrences in the Odyssey, surely it is possible that early written ΕΑΣΕ ( - - - ) etc., simply representing taat etc., was similarly hyper-Aeolicized to an equally non-existent εασσε etc., also in a few manuscripts, and in a mere 7/20 of the Iliadic cases and 6/19 of those in the Odyssey. There is a further indication that we are on the right track in comparing readings of the type έασσα/ο- with the historically impossible reading (-)εασσι(ν) 'are.' As already mentioned,71 of the manuscripts on which Ludwich's Odyssey apparatus is based, only three appear to contain readings of the type εασσα-: W L X. It is presumably not simply a coincidence that the reading (-)ε'ασσι(ν) 'are' in the Odyssey is apparently encountered in only four of the manuscripts on which Ludwich routinely reports—and that three of these are precisely his W (with 14 such readings), L (with 12), and X (with 7).72 2.4.1.4.1 What tells most compellingly against the reality of aor. eässaand fut. eässo- in Homer, however, is not the simple inadequacy of these εασσα- and ίασσο- variants as direct and convincing evidence of them. If εασσα- and ίασσο- had ever actually existed, they could only have been—as described at some length earlier (§§ 2.2-2.3)—the strictly Aeolic correspondents of what would appear in Ionic as έασα- and εα(σ)ο-. And if, realistically speaking, only εασα- and eaao- are found in the epics, then the theory that reconstructs a root *(h)euä- and verbal stems euäs(s)a- and euäs(s)o- thereby requires the further hypothesis that Homeric Ιασα- and έασο- are always and only metrically equivalent substiutes for Aeolic εασσα- and εασσο-.73

71 § 2.4.1.2.2 above, with note 62. 72 The fourth is his D (Allen's P 1 ), which has only 5 instances of (~)ίασσι(ν). 73 So Jacobsohn Hermes 45 (1910), 96 and already Fick. So too Chantraine RPh 57 (1931), who implies (cf. also GH 18) that Horn, ίασα- for εασσα- has something to do with

All about ΐ(/)άω

25

2 . 4 . 1 . 4 . 2 But if Aeolic ίασσα-, ίασσο- are what is ultimately represented both by unanimously transmitted ίασα- or ίασο- and also by those instances of ίασα- or ίασο- for which a variant ίασσα- or ίασσο- is (feebly) attested, then the Homeric situation of these forms becomes suspicious and inexplicable. For it means that in the case of this one verb, the epic tradition used the Aeolic forms ίασσα- and ίασσο- several dozen times altogether, but never once the corresponding Ionic ίασα- and ία(σ)ο-. Note: To be sure, it is securely on the record74 that there are two Homeric forms—one aorist in the Iliad and one future in the Odyssey— that can theoretically scan so as to provide one example each of putative Ionic ίασα- and ίασο-: Ούδ( μίν ούδί Τρώα? άγήνορας d'aaev "Εκτωρ (Κ 299) ουκ ίάσουσιν ίμοϊ δόμοαι β ίου ήδε φαρίτρην (φ 233) But it is equally securely on the record that in neither case is an -ασreally guaranteed. In Κ 299 the variant e tα σ' is found—even if apparently in only one ms., 75 and it would seem possible—at least in the abstract—to read an ίασουσιυ in φ 233 that is to be scanned — with synizesis of the initial ία-.76 And even if ίασα- and ίασο- are insisted upon for any reason in these two isolated passages, they would still not need to be regarded as linguistically real, since their -&σ-, not otherwise found anywhere, could always be the result of a metrically motivated analogical reshaping of actual -δσ-. In Κ 299 the aim would have been to avoid (or replace?) the spondaic line that |«ασ' "Εκτωρ# produced, while the hypothetical rearrangement at φ 233 would have been a way of dealing with the antispastic shape 77 ( w - - w ) of genuine underlying ίάσουσι(ν). The models for providing εάω with metrically advantageous ίασα- and ίασο- were present on the one hand in primary pairs like γ(\άω 'laugh': ye λάσα- and ίράω78 'love': ίρασα-, Attic

eäcra-.

But this (1) makes difficulties understanding why a tiny number of mss. are

resistant to the substitution—and in a tiny number of instances and (2) begs the whole question of how to explain Homeric έασ(σ)α/ο- vs. Attic 74 E.g. Meister HK 105. 75 Allen's P 2 1 . See Meister HK 105 again. 76 So Meister HK 105 yet again. 77 Chantraine 78 Imperfect

GH102.

ΐράασθί

(Π 208).

ΐασα/ο- in the first

place.

Two Studies in Greek and Homeric Linguistics

26

and on the other in cases where, descriptively at least, Homeric grammar offers patterns—also for metrical reasons—like αντιάω 'meet, encounter': αντιασα-, άντιασο-79 or σκ ι ά ω 8 0 'cover with shadow': σκιασα-.81'82 Finally, it should be said that one form from Iliad Κ and one from the Odyssey are just on general principles a suspiciously weak basis for establishing an ίασα-: (άσο- of any real antiquity. As a result, to take the position that Homeric ίασα- and ίασο- are in principle always and only replacements of ίασσα- and ίασσο- is to have to say (§ 2.4.1.4.1) that although both Ionic ίασα-: ία(σ)ο- and Aeolic ίασσα-: ίασσο- were available, the poets for some reason used the Aeolic forms in no fewer than 59 83 lines of the poems as we have them, and never used the Ionic ones at all. This, of course, is unlikely on the face of it in a tradition of which it is a clichi to say that metrical variants have a tendency to be exploited. 2.4.1.4.3.1 Not unexpectedly, a cursory glance at the relevant material suggests that the situation that would have to be envisioned (i.e. short syllable + Vssa/o- 59x but short syllable + Vsa/o- never) would be impossible to parallel—or to explain. If, as seems reasonably assumed, an 79 άντιασας (χ 128). 80 κατίσκίαον (μ 436). 81 σκιάστι (Φ 232). 82 These last two verbs are both certain to be denominative and should therefore

-ασα-/ -ησα- and future -α.σο-1 -ησο-, But in both cases this regular (*άντϊασα/ο-, ~σκϊασα/ο-), and as a result the Iliad and Odyssey have only the aorist and future forms here (άντί&σα/ο-, σκιασα-) that historically could be assigned to άντιάζω (Hdt., Pi., Aesch. +) and σκιάζω (Archil., Anacr., have aorist

morphology would have led to cretic forms

Hdt. +). Those two presents, for their part, were unusable in hexameters for just the same

αντιάω and σκιάω were excluded: the cretic ά,ντίαζω, ~σκω.ζω. The end result in any case was Homeric αντιάω : άντιασα/ο-, σκιάω : σκιασα- and thus both a potential model and partial parallel for a metrically conditioned Ιάω : ίασα/ο- if the Κ 299 and φ 233 instances are to be scanned with an -άσreason that the regular aorist and future of shapes

in the first place. Favoring the idea that these forms—even though analogically metrically

ίασα/ο- are two forms in elegiac poems of Greek Anthology that would seem to have been imitated from Horn, ίασα-: e i'dcre (3.14.4) and αασαν (8.195.2). 83 On the theory that Κ 299 is to be read eϊασ and therefore αασσ\ The count would of course be 60 if φ 233 is scanned ϊάσουσιν and therefore under this assumption ξάσσουσιν. "shortened" in this way—do represent Horn, the

27

All about ϊ(/τ)άω

adequate sample is furnished by just the -s(a)- aorists of roots in Proto-Gk. (C)VCä-, what is found is that at least ten such verbs show both and-ασα- forms in Homer: αγο.σ(σ)α-

'laugh/ ) *auser became *auher and then *äuer, 163 The idea that the *ä oi *auer (< *auher) was prevented from ever fronting to Ion.-Att. χ in the first place by the dissimilatory influence of the following e (e.g.

All about ϊ(/:)άω

47

whence *seuer > Seer all by familiar sound laws, at which point the nom. deer dissimilated see to äe165 (Horn, αήρ) while oblique Seer- remained (Horn. •qepos etc.). Subsequent paradigmatic levelings produced on the one hand αήρ, atpos (Attic) and on the other nom. ήήρ (Hipp.). If, now, this dissimilation were to be (re)formulated to say that Ion.-Att. Se, inherently unstable because it lacked a back correspondent (unlike e [: p] and e [: δ]), whenever as a result of w-loss it came into direct contact with e (its more stable neighbor one step up), was dissimilated down to the next fixed and stable point in the long-vowel system (namely ä); and if it may be hypothesized that this was the regular treatment no matter whether the Se was preceded or followed by the e, there would seem no actual obstacle to supposing that just as seer dissimilated to aer, so did esesa- become eäsa- while, in theory, unaugmented eSesa- and future eSeso- (cf. nom. aer vs. obi. Seer-) would have remained, at least for the time being. 5.4 If, furthermore, as seems conceivable, the augmented aorist *eSesabecame *eäsa- more or less immediately after loss of intervocalic u, the next relevant phonological development could have been the shortening of the augmented aorist's eä- to ea- by a sub-part of the same Ionic-Attic rule as (among other things): 1) s h o r t e n e d e, e, Se to e b e f o r e g: (u)e(u)ptha βασιλέων,

> ΐωθα, basile(u)ön

>

nSe(u)ön > ναών a n d

2) shortened e, e, Se to e before ä (and ö, p); lengthened such an a to α (and changed such an ö or δ to p)—at least when the first vowel was e or Se:

Schwyzer GG 1, 187; Lejeune Phonetique 238f.) is unsatisfactory:

If the dissimilatory

influence of e is supposed to have made itself felt before loss of intervocalic u, and thus over a consonant ("äCe remains"), it becomes more difficult than necessary to explain forms like Ion. κ ρητή pa (Sigeum 6 c. BC DGEEP 731 etc.) < krSter-.

And once u between

vowels has been lost, it is too late for a dissimilatory failure of ä > χ because the fronting had already happened by then, of course. 164 Cf. Nussbaum Head and Horn 236f. with further references. 165 The formulation according to which the S vs. e distinction was lost in Attic and W.,E. Ionic before the dissimilation, and that it was therefore eer that dissimilated to äer (Peters Untersuchungen

302) probably cannot be disproved, but offers no advantage that

would compensate for the (admittedly slight) extra complication of having to assume that e(u)e became äe although e(h)e had become ee (ΐηκα Peters Untersuchungen

256). In

any case, even if this chronology cannot be ruled out, neither can it be decisively supported.

48

Two Studies in Greek and Homeric Linguistics

tele(u)a ovt

> re\ta,

phre(u)ar

> φρίαρ,

onS(u)ar

> oveap

(cf. H o r n ,

ovmap,

ίατα).

The same sound law would have applied to the imperfect e(u)ö-/e(u)ä

(
j... (Π 96) They are presumably just the regular forms of relatively late Ionic introduced into the epics, and they do not require special attention here. 6.1.1.2 In some other cases, it is possible or probable that a trivial modernization has eliminated a form of obsolete structure in the usual way. Such an assumption makes for an actual prosodic improvement in a number of cases: ... ποταμοί

*ίάουσιυ#

... ούδ' *ίάουσι\

...

f o r . . . 7τοταμοϊο ίώσιν#

for ... οϋδε earn| ...

171 But v.l. ea and cf. § 6.2. 172 Cf. Tichy MSS 40 (1981), 219 (note 48).

(δ 805)

(Τ73)

53

All about i(f)au>

ft τον ζύνον

#τώ

δ'

*εάωμ€ΐ>... for

...de

...

ΐώμίν

x\eos ... for #τώ μ£ ea κλεοϊ ...

μ *eae

... μηδ' *eaetyf ... for ... μη& eäv ...

(κ 536)

(σ 420) 1 7 3 (Ρ 16) 174 175

In other instances, it is a simple possibility that requires no violent rearrangements: #ov τίνα μέν κ *eays

...

f o r . . . μέν Kev eas . . .

#αίκ

...

f o r # a t Kev ea ...

*eay

ττρόφρων

(λ 1 4 7 )

(ν 3 5 9 )

Beginning to verge on the radical would be things like: #νω ...

*ίάω

τον

δ'

...

for

*£άωμ(ν#

#νωΐ

for

ίώ ... rovhe

(Θ 428) 1 7 6 δ'

eώμev#

(Β 236) 1 7 7

The situation, in any event, is that Homer has contracted present-stem forms, some of which can or may be resolved and some of which cannot. 6 . 1 . 2 . 1 . 1 The rest of the Homeric forms belonging to the present categories listed above (§6.1.1) all appear in one of two purely epic guises—distracted or contracted and lengthened: 178 Indicative: eaaj (μ 282+), eia (Θ 414), αώσι (Λ 550+) Subjunctive: eta> (Δ 55), Infinitive:

ίάαν (θ

eaas

(λ 110+),

eiώμev



260),€ΐώσι (Τ 139)

509)

It goes without saying that for the distracted forms it is always possible to substitute the original uncontracted shapes (*eaeu or

*eaj]s

for eaay etc.).

But it also happens to be the case that the lengthened contracted forms are without exception versified in such a way {el- in thesis) that here too the expectable uncontracted pre-forms are always substitutable: 179

173 The vv. II. S' ΐώμίν (sic) and δ' άωμίυ (cf. Ludwich ad loc.) d o in fact offer a δ' here, for what it is worth. 174 On which modernization, in turn, might have been based # μή μ( 'ία παρά νηνσι ... (Χ 339), where ... μι ea ... can no longer easily be considered a transformed # μή μ' eae . . . , of course. But it is debatable how good an assumption is the position m a d e by κλ- in the conjectural ... e *eeo- vs. preservation of *e(u)ae- in Ionic. Quite apart from the obvious objection that such a sound law simply cannot be counted on, 184 there is no apparent trace in Ionic of the *£€0-/*«ω- and /or *?δ-/*£ω- that are presupposed. Instead, the Ionic paradigm seems to have been *ϊωσι And explaining the forms that are actually found in Homer therefore entails further complications. One would have to suppose, for example, that Ionic-epic *'ίοσι was spelled 187 —as it would have been—ΕΟΣΙ in an early Attic text, and that the eventual transliteration of this ambiguous spelling as eίώσι was chosen because that could be interpreted as a lengthened form of Attic and Koine ίώσι; and so too for *ϊω —> EO *ϊωμίν

(ίώ,

—» EOMEN —> άώμζν, *Ιωσι —» ΕΟΣΙ —» «ΐώσι. But aside from

arbitrarily discounting the influence of early Ionic texts and the tradition of out-loud recitation, both of which could have disambiguated these archaic Attic spellings for the diorthotes, this means that the eiw- forms as transmitted can take us back only to graphic Old Attic EO- forms, which there is no positive reason whatever to take as in particular guaranteeing *e- from *ee-. In fact the available Ionic evidence suggests instead that Homeric eΙώ- in this paradigm is best seen as ίώ- with its first syllable lengthened. 6 . 1 . 2 . 2 . 2 If eiw etc., as they stand, are unambiguously—from the descriptive point of view—lengthenings of ίώ etc. in metrical structures that demanded (αο-/ΐαω-in

the first instance, all that is actually required

here is the identification of the model that suggested to the later poets of the tradition that εΐω- was a well motivated epic correspondent of their own (ώ-. In this very verb, the forms that would in fact strongly suggest 186 Concordantia 1044. 187 Chantraine GH 5f.

56

Two Studies in Greek and Homeric

Linguistics

that are Ion. imperfect ίων beside epic (ϊων (§§ 3.3.2, 5.6.2); and this model has, to be sure, been invoked to explain εϊώ- in place of *

79

7.5 If it may be argued on this basis that the core meanings of to. ω were 'abandon, leave behind, leave alone,' it would seem at least thinkable to infer a PIE verb to a root *h}ueh2- (§§ 2-4) that meant similar things. But these are precisely the semantics that would be ideal for many of the items that belong by root etymology with evvis, uänus etc. and therefore call for a *h,ueh2- of their own. 7.5.1

Since in Greek itself the verb derived from *h1ueh2-

'leave

behind' very often has a personal object (§§ 7.4.1, 7.4.6), it is no surprise that the corresponding passive verbal adjective *hjuh2-ni- 'abandoned, left behind' > ewis 'bereaved' > 'bereft, deprived' is used of people: δ? μ υιών ττοΚΚών re και ίσθλών evviv ίθηκί

(Χ 44)

'since he has bereaved me of many excellent sons' 7.5.2 Although Greek, at least early on, does not use words for places of any kind as objects of ίάω 'leave behind'—which is why evvts means 'bereaved' but not 'vacated'—there is no doubt that the root of ewts is also that of Latin uänus, Avestan Una- and Armenian ownayn, which all mean 'empty' and thereby practically prove that the proto-language, unlike Greek, did use transitive *72,ue/j2- 'abandon' with objects that were words for places—the result being the meaning 'abandoned, vacated, empty' for the *h1ueh2-no-

reflected in Latin and the *h1uh2-no-

of Avestan and

Armenian. The difference in root ablaut between L. uänus (*h1ueh2-no-) and its nearest correspondents, Av. Una- etc. (*h1uh2-no-) and ewis (*h1uh2-ni-) also attracts attention.

It has been compared 254 to the parallel-looking

distinction observable between Latin again, with plenus 'full' (apparently < *pleh1-no-), and Skt. pürnä-, Olr. Ιάη, Go. fulls etc. 'full' (< *plh1-no-).255

And

the chances that this parallelism is significant are, if anything, increased by the circumstance that plenus 'full' and uänus 'empty' are semantically related in an obvious way. But Latin (-)plere : (-)pleui: (-)pletus 'fill' on the one hand and a *plh1-no- behind even the Olr. correspondent Ιάη on the other make it seem very possible that Latin has remade the *pläno- 'full' that would have resulted from *plh1-no- by analogically giving it the pie- of

254 E.g. Hoffmann Aufsätze 2,467 note 3. 255 Since A v. frana- 'Füllung' is a substantive and the basis of Olr. linaim 'fill' could be one too, these offer no support outside Latin for an adjectival /participial 'full.'

*plehrno-

80

Two Studies in Greek and Homeric Linguistics

'fill'—the special additional factor favoring pleno- over *plano- 'full' being disambiguation with piano- 'flat' < *p\h2-no-. A final point belonging here is the ζ-stem evvis beside the o-stem represented by Av. una- etc., a state of affairs that it may be enough for present purposes simply to parallel by comparing *h1uh2-ni- (ewis): */ijuh2-no- (Una- etc.): *h1ueh2-no- (uänus) with *prk-ni- (Ved. prsni'speckled'): *prk-no- (ττράκυον μέλανα Hsch.): *perbno- {πΐρκνός 'dusky'). 7.5.3 As already noted (§ 7.2) the derivatives of *htueh2- in Latin include not only uänus, which has correspondents elsewhere, but also uästus, with closely comparable relatives (Olr. fas 'empty/ Gmc. *wöstja'deserted') of its own. A *h1ueh2sto- is, of course, what is to be mechanically reconstructed behind this Italic-Celtic-Germanic item; but there remains the question of the further analysis of such a pre-form, and it is in particular the *-s- forming the stem from which the *-to- adjective has been derived that could use some elucidation. On the one hand, a perfectly unobjectionable denominative analysis could be imagined, whereby *h{ueh2- 'abandon' formed an s-stem verbal noun *h1ueh2-(e)s- whose derivative *h1ueh2-s-to- 'having (undergone) abandonment' lies behind uästus and its Celtic and Germanic congeners. In that case the category of derivative in question is that of e.g. *h2eidh-s-to'hot' (*aisto-tät- > L. aestäs, Gmc. *aista- implied by OE äst 'kiln') or *bheh2-s-to- (L.fästus 'lawful') beside *bheh2-s (L.fäs). But this hypothesis has serious weaknesses. First, there is not the slightest independent indication demonstrating or even encouraging an expectation of an assumed nominal s-stem to *h{ueh2-. In addition, the denominative analysis cannot explain in any unarbitrary way why L. uästus can have an insistently passive meaning like 'abandoned, deserted, desolated' or even 'laid waste' while uänus simply means 'empty,' nor why things arrange themselves comparably in Germanic—where OHG wuosti etc. mean potentially passive things like 'deserted, desolate(d)' while Goth, wans, OHG wan etc. are more like essentially intransitive 'lacking.' Both of these defects in the analysis of *h{ueh2sto- 'deserted' can be made good by prefering an explanation that would make it a passive deverbative *-to- adjective derived from a *lz,we/z2-s- that is therefore by definition a verbal stem itself. This not only accounts for the passive value of *hjUeh2s-to-, but at the same time dovetails neatly with a conclusion that

81

All about k(f)au>

emerged earlier (§ 4.3) in the discussion. Namely, once it is ascertained— and strictly on grounds internal to Greek—that the shape of the root of έασα (whence future Ιάσω and surely not the other way around), and that the present 'empty' in Av. Una-: Arm. ownayn) etc. But the likelihood of an important intransitive PIE use of this verb (§ 7.6.1f.) in meanings like 'leave off, give out, fail, be insufficient' gives rise to the expectation that a *ht uh2-no- that is "active" with regard to this intransitive use—and thus means 'failing, deficient'— might also be found. And this is in fact reflected in Indo-Iranian and Germanic. 7.6.3.1.1 Avestan has a merely possible direct trace of the meaning 'deficient' for Una- in the presumably substantivized ünä- 'deficient (offering):' auuarjhsrdzämi janiiöis ünam (Υ 10.15) mairiiaiia duuitö.xraSaiia Ί renounce the deficient (offering) of the ... woman'

256 On väyati 'is deprived' (RV 8.47.6) cf. below. 257 The transitive use of this present (ätram fräuuaiieiti 'puts out the fire' V.5.37) is simply mysterious—at least to me.

All about €y ef)o?# ( Τ 342), listed above in §2.2.4.

If so, there emerges an extra reason for

adopting the reading ίηος of the overwhelming bulk of the ms. evidence against Zenodotus' koto at Τ 342, and an argument in favor of seeing here a s e c o n d instance in w h i c h the m e a n i n g ' g o o d , noble, b r a v e '

is

guaranteed—or nearly so—for the form: 'you are turning your back on a

13 That the ανήρ versions are basic and the φώ? versions conditioned variants (used where a C-initial form is necessary) is shown by the fact that in the one place where they are really interchangeable—namely the beginning of the line—forms of ανήρ appear, but no form of φα>5 does.

94

Two Studies in Greek and Homeric

Linguistics

noble man.' In addition, this simply gives a much better sense than any conceivable interpretation of an avbpos kolo in that passage. In fact, it happens to be the case that in the only two Homeric instances of άνήρ with a personal possessive—namely avhp ψόν at Τ 295 and kov avbpa at τ 209— the context unambiguously demands 'husband' for αυήρ, which raises a serious question of whether avbpos koto with a hypothetical second singular possessive might not have to have had this specialized meaning: 'your husband'—unusable at Τ 342, needless to say. 14 2.3.3 The next link in the chain would then obviously be the expression—yet another of the cases (§2.2.4) in which the tradition would point to ef/os rather than kolo anyway—found at ο 450: iralba

yap ανδρός

£ήος\

evl μίγάροις

άτιτάλλω

To be sure, avbpos ki70s, the adonic of Τ 342, occurs here in a different position—ending at the trochaic caesura. But that is anything but a reason not to see it as simply another occurrence of the same formulaic expression. It is in fact extremely common precisely for demonstrable formulae of adonic shape to be "mobile" between these two line positions:15 e.g. ävbpa ϊκαστον 9x line-final, lx at the trochaic caesura; 'δήϊον avbpa and avepa τούτον lx at each; άγλαός (-όν) or ψαιδιμο? {-ου) vios (-όν) 40x line-final, 3x at the trochaic caesura; άγλαα τίκνα 3x at each; etc. etc. But if, as seems unavoidable, the avbpos ki)os'\ at ο 450 is really a "moved" adonic of this kind, the natural assumption, in the absence of especially compelling reasons to the contrary, would be that it means the same thing as at Τ 342 (cf. £ 505), and the line should therefore go Ί tend the child of a nobleman in (his) hall.' In short, whatever is to be made of the variation of krjos with kolo in the tradition and attendant problems, it seems entirely justifiable to conclude already now that since ||φωτόϊ kijosft, with uncontested krjos, at ζ 505 must mean 'good man,' and since it is hard to see ||άι>δρόί krjostt at Τ 342 as anything other than its vowel-initial 14 In view of these arguments in favor of a eijos that means 'goodly, noble' at Τ 342 (cf. also Hoffmann Aufsätze 2, 602), it is unappealing to suppose (Brugmann Problem 55ff., Schwyzer SP AW 1938, 85 and 88) that eijos is "late epic" (ξ 505, ο 450 + Ω 422) and has replaced an adjectival reflexive possessive f οίο (or an equally possessive reflexive pronominal genitive—Lamberterie Adjectifs 755). This has obvious consequences for the eoio vs. ajos question (and cf. also §§2.5-2.6,3.3,3.5). 15 Hainsworth Flexibility, Table VII (137ff.), where ävbpöi ίήος is itself recorded as an example, provides an extremely instructive set of illustrative cases from which these are selected at random.

95

Good for You: kfjos (and ίάων)

formulaic partner, and since it is in turn unthinkable to separate the instance at Τ 342 from ... avbpos kfjos|' at ο 450 (and vice versa), there is to be recognized an adjectival epithet 'good, noble' whose masc. gen. sg. appears as ίήος.

2.3.4 If such a form can be counted on as not only Homeric but traditional as well, the next step might be to invoke it to solve what is otherwise an intractable problem. As already mentioned (§2.2.4), LI 422— if the variants were left to their own devices—would read: ως TOL κήδονται μάκαρΐς 0eoi|| υίος ER/ostt and this line is one of those (§2.2.4) for which a eoto instead of ίήος is found (directly in one ms. 16 and indirectly as eoto in one more 17 ), even though no reflexive possessive of any person makes sense and the only pronominal possessive of any kind that would fit would be a second sg. non-reflexive, which a kolo could scarcely be. 18 Consequently, the best bet at i l 422 is to believe the tradition, read as just above, and accept the excellent sense 'So, in short, do the blessed gods care for (your) noble son.' This adds to our list an adonic ||uios kfjostt which stands to ||ά^δ/οόϊ ki)ος# (Τ 342), as in §2.3.2, more or less as # χητίϊ τοιοϋδ' νιος)' (Τ 324) does to # χήτίϊ

τοιοϋδ' ανδρός \ (Z 4 6 3 ) .

2.3.5 At this point, it would be thinkable to note that || νιος cijos tt is the reading of the length and breadth of the tradition in two more of the passages grouped in §2.2.4: τώ σ αύ νυν κέλομαι ου yap τι πρήξεις

μΐθίμίν

άκαχήμένος

χό\ον νιος kfjos ( Ο 1 3 8 v.l. νιος ίήος

kolo)

( ί 2 5 5 0 v.l. kolo)

And on that basis nothing seems to stand in the way of arguing that these could merely represent two more occurrences of the very same adonic that is arguably to be read at Ω 422, where it pretty much has to mean (§2.3.4) '(your) noble son.' But although this reading and interpretation are ultimately to be adopted (§§2.5-2.6 below) for the final segment of Ο 138 and Ω 550, it will turn out to be better to argue for that on somewhat different and expanded grounds. This is, first and generally, so that the chain of inferences does not lose credibility as it gains length; but more

16 Allen's P 1 3 . 17 Allen's L 4 .

For what it is worth cf. also ίοιοσ in Allen's Bm 6 Pa Pe, ίηοί in

Ludwich's Ud, and τ(οΐο in the margin of a single papyrus—which may be less a "reading" than an unmetrical gloss that fits the sense perfectly even if redundantly. 18 Cf. §2.2.4 with note 12.

96

Two Studies in Greek and Homeric Linguistics

specifically because it seems less than perfectly ideal to rule against a mos kolo that could be a reflexive '(your) own son' in Ο 138 and Ω 550 precisely on the basis of a case—Ω 422—where a reflexive is excluded: the contrast makes it less than absolutely certain that it is the very same line segment in the first place that must have occurred both in In Ω 422 and in Ο 138 and Ω 550. 2.3.6 The two main results so far, then, are quickly summarized: 1) There are four secure-looking instances (ξ 505; Τ 342, ο 450; Ω 422) of a masc. gen. sg. epithet that has the form ίήος and the meaning 'goodly, noble.' 2) In at least three (T 342, ο 450, Ω 422—if not Ο 138, Ω 550 as well) of the six cases (§2.2.4) in which a—feebly attested, to be sure— variant kolo competes with kfjos, it can be rejected. 2.4.1 Having gleaned a certain amount from one end (§§2.2.4-2.2.5) of the spectrum of competition (§§2.2.1-2.2.5) between kolo and ίήος, the obvious way to work back toward the most difficult cases in the middle (§§2.2.2, 2.2.3, the remnants of §2.2.4) is to examine the opposite end: the externally (textually) and internally (syntactically and semantically) secure instances of the reflexive pronominal possessive ίοϊο. 2.4.2 These were already listed above in §2.2.1: Β 662, Ν 522, Ξ 11, 266, Τ 399, Ψ 360, ξ 177. In these six passages of the Iliad and one of the Odyssey πατρός kolo (6x) or mos kolo (N 522) or muöös kolo (H 266) is transmitted without apparent variant, and in all cases the kolo is furthermore entirely unproblematically parsed as the masc. gen. sg. of the third person sg. reflexive pronominal possessive: 'his/her (own) father, son, child.' There is no reason to think of reading or reconstructing anything else in these lines, and nobody seems to have suggested it. 2.4.3 If, however, || φωτός krjostt is sure at ζ 505 (§2.2.5)—and if I avbpos kfioitt at Τ 342 (§2.3.2) and || mos e?)os# at Ω 422 (§2.3.4) are consequently probable; and if, on the other hand, || -πατρός kolo# (H 11+) and || τταώός eoloU (E 266) are also secure (§§2.2.1, 2.4.2), there may be a plausible economy argument to be considered here: An adonic formula of the form || - 1 - kolott would never be metrically different from the same adonic with ei)os# instead. But one of them is of severely limited utility. The expression || kolott, with its reflexive, can by definition be used only when the topic is recurring as a possessor. In contrast, || — w e?)os# can be used both when the conditions for a kolo would be met and—speaking in descriptive terms about this corner of the epic repertory—virtually

97

Good for You: (f/os (and ίάων)

everywhere else too. For example, it is easy to imagine a Η 11 (cf. §2.2.1 again) that had * ... ασπίδα|| father/ or *||

παώός

ei)ος#

πατρός

ίήος#

... the shield of (his) noble

'

at Ξ 266 etc. The argument would be, in other

words, that if the tradition went to the trouble of creating and maintaining a restrictedly usable || - metrically identical ||

at all beside the much more adaptable and

eolo#

it makes a certain kind of sense to proceed

ίηος#,

on the assumption that the less flexible—but more informative— alternative, namely ||

eoloti,

would have been used, in principle,

whenever syntactic and semantic conditions permitted. This certainly would apply to: ως ίίττων σάκος ei\e πτυγμένου

νιος eolo

'Αντίλογος

πατρός eolo

But if the variant

δ' ί'πποισιν eolo

(κύκλΐτο

(Ε 9) a n d (Ψ 4 0 2 )

in these two passages gains any support from such

considerations, all that has been supplied in this way is an additional and independent reason for favoring what is transmitted almost unanimously by the ms. tradition anyway. 2.4.4.1 This leads directly to a consideration of the two lines in which a choice between the variants

eolo

and

begins to pose a serious

ίηος

question (§2.2.3): ό£ν δε κωκύσασα ως αρα φωνήσασα

κάρη Kaße πάλιν

παίΐός

eolo

( Σ 7 1 v.l.

τράττίθ' νιος eolo

ίηος)

( Σ 1 3 8 v.l.

ίηος)

It is obvious, as already mentioned (§2.2.3), that a third sg. reflexive

eolo

would be instantly admissible in both cases, giving lines meaning ' ... she clasped her (own) child's head' (Σ 71) and '... she took leave of her (own) son' (Σ 138). And applying the reasoning sketched out just above (§2.4.3) in connection with || eolo

νιος eolo

# in Η 9 and ||

πατρός eolo it

in Ψ 402, whereby

would be read wherever it is syntactically admissible, the choice

seemingly indicated would be for the

eolo

that most of the tradition

supports in both these lines of Σ in any event. 2.4.4.2 What is less immediately obvious, however, is that no account concluding that

eolo

is the better reading not only in Ξ 9 and Ψ 402 (§2.4.3)

but also in Σ 71 and 138 can be considered sufficient and convincing if it fails to explain why the variant

ίηος

is so much better represented in the

two Σ passages than in the other two. And an explanation does seem to be available. As already noted (§2.2.3), both instances in Σ of this surprisingly well supported

ίηος

happen to occur in one and the same scene. At Σ 71 Thetis

98

Two Studies in Greek and Homeric Linguistics

has just arrived, having heard Achilles mourning Patroclus, at the ships; and just before asking him what the matter is she takes her son's head in her arms:... κάρη Xäße παώός koio. At Σ 138 the meeting between Achilles and his mother has just ended, and promising to come back in the morning with some new armor, Thetis turns away from him: ... ττάΚιν TpaTieff vios

kolo.

But it is presumably not a simple coincidence that the Σ 71 variant which is so much more strongly transmitted than || νιος er/os# at Η 9 or || πατρός er?oi# at Ψ 402, occurs—or has come to occur 19 — "for real" (to anticipate what will be established on grounds of its own in §§2.5-2.6) once and only once elsewhere in the Iliad·, and this is precisely in an earlier scene between Achilles and his mother, where he says: άλλα συ ei δύνασαι ye irepiayeo παιδόί efjos (A 393) 'But you, if you can, should support (your) noble son' Furthermore, this is a particularly striking instance because—for reasons having to do with the demonstrable existence (§2.5.2.3) of a system of formulaic line segments here—it puts the essentially ornamental epithet er)os 'noble' into Achilles' own mouth in reference to himself.20 It is this memorable feature of the A 393 passage, it can be supposed, that has helped introduce || τταώος ef/ostt, to the extent that has happened, in place of expectable (§2.4.4.1) || muöös eolott at Σ 71—again denoting Achilles in the course of a meeting with Thetis. It was then, naturally, this || παιδόί £Τ)ΟΪ# option, once established at Σ 71, that led directly to a || vios cqostt competing with "original" | vios eoiott later on in the very same passage at Σ 138.21 Ι ττο,ώός efjosft,

19 Whether this earlier instance of || παώος erjostt is "original" epic or represents a replacement of a putatively second-person eolo 'your (own)' by eijos (§3.3) affects the point not at all: no matter which history irjos itself or any particular instance of it is given, A 393 is a primary or secondary i-ηος passage (§3.3) while Σ 71 is a eolo passage with an unusually heavy representation of eijor as a v.l. 20 Schwyzer SPAW 1938, 85f. gives two excellent parallel cases of self-referent ornamental epithets, but only in connection with his judgment that || παιδός ojo?# at A 393 is "unpassend" as a third example of the phenomenon. This judgment, however, is hard to accept once it emerges that || παιδόϊ iyostt is on the one hand a by-form (§2.6) of Ι παίδόs eoiott (Ξ 266+) and on the other (§§2.6, 5.1) may be considered the genitive version of «ϋ? 7rcus|| (Β 819+). 21 Conversely, this scenario also immediately explains why || -rraiSos iolott is a far more frequent v.l. at A 393 than is e.g. || vios iolott at Ο 138.

Good for You: cijos (and

99

ίάων)

2.4.5 At this point it seems useful to summarize again. In addition to a confirmation of the view that a masc. gen. sg. epithet kr\o$ 'good, noble' is beyond question at ξ 505 (§2.2.5) and i2 422 (§2.3.4), highly probable at Τ 342 (§2.3.2) and ο 450 (§2.3.3), and very possible at Ο 138 and £1 550 (§2.3.5); there is no more reason than there ever was to doubt unanimously transmitted (§2.4.2) kolo at Β 662, Ν 522, E l l and 266, Τ 399, Ψ 360, ξ 177; it can be argued (§2.4.3), unsurprisingly, that the tradition's kolo at Ξ 9 and Ψ 402 is perfectly trustworthy; and the relatively authoritative appearance of the variant k-ηος at Σ 71 and 138 (compared to Ξ 9 and Ψ 402) has a special explanation (§2.4.4). There thus seem to be grounds for choosing between kolo and ΐηος in all the cases mentioned above in §§2.2.1-2.2.3 and §2.2.5 as well as in three (T 342, i l 422, ο 450) of the six cases grouped under §2.2.4—of which A 393, 22 Ο 138, and £1 550 therefore remain. 2.5.1.1 The result of the discussion in §2.4, in any case, is that for various reasons and with (only slightly) varying degrees of confidence the following formulaic line segments are attributable to the Homeric tradition: # vlos

ίοϊο

ft πατρός

(N 522) koto

(£ 177)

νιος •πατρός

kolo | ( B 6 6 2 )

kolott

πατρός

( E 9, Σ

138)

kolo ft ( Ξ 1 1 ; Τ 3 9 9 ; Ψ

360, 402; [υ 289?]) παώός

koto it ( Ξ 2 6 6 , Σ

71)

More particularly this little system appears to consist of three parallel adonics, one of which—1| νιος kolott—is mobile between the end and the beginning of the line, 2 3 while another—1| ττατρός kolott—has been transferred for use once at the line initial and once to end at the trochaic caesura. 24 2.5.1.2 All instances of these parallel formulas, furthermore, occur with a third-person topic and consequently mean '(of) h i s / h e r (own) father, son, child.' This in turn means, however, that the system of

22 Cf., however, §2.4.4.2 above. 23 Cf. §2.3.3 and again Hainsworth Flexibility 137ff., noting as exactly analogous e.g. Ipyov del««: 5x line final, lx line initial. 24 Cf. again H a i n s w o r t h Flexibility as in the p r e v i o u s note and e.g. γαία μέλαινα:

line final, lx line initial, lx ending at the trochaic caesura.

5x

100

Two Studies in Greek and Homeric

Linguistics

traditional line-segments just listed is made up of nothing more than the third-person correspondents of the first-person expressions: # πατρός

ίμοΐο

(Η 1 1 8 , •πατρός τ

ϊμοΐο

| (£"308,

I πατρός

ίμοΐο#

παώός

£μοϊο#

(α 4 1 3 , £"290,

υ 339)

ο 417)

180)

(λ 4 5 8 )

It is potentially noteworthy, to be sure, that the third-person set is confined almost entirely to the Iliad and is, of course, always reflexive, while the first-person set includes only one instance outside the Odyssey and only one reflexive (υ 339); and those distinctions will certainly have to be borne in mind. Nevertheless, there can be no serious doubt that from the purely technical point of view of composing verses with traditional, reusable, movable elements of particular metrical contours, these two sets of expressions are the first- vs. third-person versions of one and the same segment that is thus "inflectable." 2.5.2.1 The identification of first- and third-person forms of a unitary type of expression (metrical, positional, and semantic), however, insistently raises the further question of what the poets used when a second-person version was called for.

And the question becomes

somewhat interesting with the realization that there is no second person pronominal possessive with the metrical characteristics of third person eoio and first person Ιμοΐο: σοϊο ( - - ) 2 5 and σου (**), of course, do not even have the same number of syllables, and reoto 26 could never be used to form a second-person form of the expressions in question either, since its consonant initial would always produce an inadmissible

shape

(*7τατρος τΐοΐο , *υίος re οίο etc.) which—even if remedied in the usual way (i.e. by metrical lengthening)—would not be usable in parallel with Ι νιος

kolo,

Ι πατρός

ίμοϊο

etc.

25 I πατρός ϊμοΐο and || πατρός eo!o# could have a 2nd person *|| πατρός aotoff (and cf. #

πατρός σοϊο'\ Ω 486) only at the expense of a spondaic line. That the poets went

out of their way to avoid paying this price even if it meant "formulaic heteroclisy" is demonstrated by the fact that the nominative corresponding to acc. || πατρίδα 140+ and dat. || πατρίδι

yairjtt Γ 244+ (and gen. || πατρίδος

γαϊαν # Β

αϊης# Β 162+) is not spondaic—

and otherwise unobjectionable—*|| πατρίς γαία ff, but rather || πατρίς apovpatt a 407+. 26 Cf. note 17 on this reading in the margin of one papyrus at Ω 422.

101

Good for You: ίήος (and (άων)

Consequently the poets had to use some item altogether different from σοΐοΙ σου and reolo in order to work the analogue of mos / τταιδός/ πατρός kolo and παώός/

πατρός Ιμοϊο into a passage of direct speech in

which it was a question of the son, father, or child of the person spoken to. In purely descriptive terms there can be no uncertainty whatever about what item it was that the poets hit upon to fill this function.

It was

precisely ίήος. 2.5.2.2 We have already had occasion (§§2.2.4, 2.3.4) to look at one passage in which || υίος ίήος# appears where a second-person version of | υίος

eoloit (Ξ

9, Σ 138)—and cf. || παώός

(μοΐο#

(λ 458)—might

theoretically be expected. This was at I I 422 (the disguised Hermes to Priam): ως TOL κήδονται μάκαρες öeol νιος έήος 'That is how the blessed gods are caring for (your) noble son' But it will naturally be no accident that wherever the written tradition favors er}os and a possessive can be accommodated at all, it is always a second person possessive that would have been called for. 27 In addition to 12 422 itself, this is true of: 28 άλλα συ ei δύνασαί ye irepio^eo παώός ίηος (A 393 v.l. ίοΐο) 'But if you are able, look out for (your) noble son' τώ σ αϋ νυν κίλομαι μίθέμεν χόλον υίος ίηος (Ο 138 v.l. ίοΐο) 'So I bid you give up (your) anger over (your) noble son' 29 ού yap τι ιτρή^ις άκa^χf|μevoς νιος ίηος (Ω 550 ίοΐο Zen., schol. A) 'You will not accomplish anything by grieving over (your) noble son' 27 At ο 450 a possessive would have to be first person, but even that can be worked in only on condition that äv&pos there means 'master'—a particularly weak assumption (cf. e.g. the schematized semantic account of άνηρ in LFGE 829 col. 1). 28 Boiling's treatment (Language 23 (1947), 31f.) of this class of cases explicitly forswears consideration of any factor but whether a kolo variant is found somewhere or other—an approach that seeks to substitute policy for philology and cannot, for example, even explain why kolo is so much better represented at A 393 than in the other passages grouped here (cf.§2.4.4.2 with note 21). Cf. also §3.4. 29 The possessive that would make sense here could even be considered a reflexive possessive, since ... σ ... κίλομαι μίθίμίν is for all intents and purposes the equivalent of imperative μέθα.

But Ω 422 beside A 393 already show that the use of έήο9 illustrated

here goes with second persons whether reflexive or not.

102

Two Studies in Greek and Homeric

Linguistics

The possessive—impossibly strained, to be sure (§2.3.2)—that would be demanded in the final such instance is, in any case, also second person: τίκνον ίμόν, δη πάμπαν άποίχεαι ävbpos ίηος (Τ 342 ΐοϊο Zen., schol. A) 'My child, you are indeed turning your back completely on a noble man (of yours?)' But it is hard enough to read a possessive of any kind into this that it seems justified to conclude that this Τ 342 instance is on the one hand, as already proposed (§2.3.2), simply one of the two examples of formulaic I ανδρός erjostt 'a noble man'/ ανδρός evjoy'f 'of a noble man,' the partner of once-attested || φωτός ef)os#; and is simultaneously an illustration of the poets' way of using ίηος in the second-person versions of expressions that have e μ ο ΐ ο and eolo in their first- and third-person forms. 2.5.2.3 The argument is easy to encapsulate. It is no accident that aside from the passages in which a ίηος 'goodly, noble'—and no possessive at all—must or should be supposed for good and sufficient reasons in any event (£ 505 plus Τ 342 and ο 450, il 422), the direct and indirect authority for ίηος in preference to εοϊο is compelling precisely where a possessive is semantically and syntactically admissible, and where a second person singular possessive in particular would have been the one called for. That is to say that the full picture of the systematic interplay of ίήος with (μοίο and eolo in this set of line segments can be schematized as: A. Possessive certain or at least admissible 1. Explicit first sg. possessive (reflexive30 and not): ϊμοΐο 'my (own)' # πατρός

ίμοΐο

(Η 1 1 8 + )

•πατρός ΐμοΐο

| ({"308+)

Ι ττατρός ίμοϊο#

(α 4 1 3 + )

Ι τταώός ίμοΐο#

(λ 458)

30 O n l y at υ 339 m i g h t || πατρός ίμοΐο It be t a k e n as s h o w i n g a reflexive possessive.

Good for You: efjos (and ίάων)

103

2. Implicit second sg. possessive (reflexive and not 31 ) admissible: (fjos '(your) / (your own) noble'

παώός ίήος# (A 393) νιος ίήος# (Ο 138, il 422, 550) 3. Explicit third sg. possessive (reflexive): kolo 'his/her own'

νιος kolo ft (Ξ 9+) ft νιος kolo (Ν 522) # πατρός kolo (ζ 177)πατρός koto | (Β 662)πατρός kolott (Η 11+) παώός kolott (Η 266+) Β. No possessive admissible: ίηος 'noble'

άνδρός £ήος\ (ο 450) ανδρός ifjostt (Τ 342) φωτός ίήος# (£ 505) 2.6 The last few instances of competition between ίήος and kolo (A 393, Ο 138, Ω 550: §2.4.5) can be settled in favor of ίηος, in other words, on the basis of a distributional hypothesis that is itself based on the assumption that the correlation between heavy representation of ίηος in the tradition and the admissibility of an implicit second person possessive (whenever any possessive at all is admissible) is not coincidental. One of these last three instances, furthermore, presents an aspect of particular interest in the context of this discussion—namely A 393:

άλλα σίι et hvvaaai yt πβρίσγίο παώός ίήος Here Achilles, talking to Thetis, seems to refer to himself as '(your) noble son;' and this, as mentioned above (§2.4.4.2), may make a peculiar impression. But if it is really the case, as just now proposed (§2.5.2), that one of the two functions of ίηος was to act as a substitute for an actual, but metrically unusable, second sg. possessive in expressions that had firstand third-person versions using εμοΐο and kolo respectively, then what is seen in A 393 is simply the result of the application by some poet of one of the innumerable rules of his craft: The second person form of || --εμοΐο#/ I - w eoto# is made by substituting ίήος, where the second-person function, 31 An implicit possessive would be a non-reflexive one only at Ω 422.

104

Two Studies in Greek and Homeric Linguistics

however, is entirely implicit and the meaning 'noble/ though certainly not inappropriate, is relatively colorless. Finally, the recognition of a traditional || παώός kijostt 'noble son' at A 393 gains considerable support from repeated ew? πάϊς\\ 'noble son' at Β 819, Μ 98, and Ρ 491; 32 and it also allows || uioy er}os# at Ο 138 and Ω 550 to be regarded simply as the vowelinitial correspondent.

32 Always em 7rcus 'kyyiaaoti—i.e. of Aeneas. Cf. Hoffmann Aufsätze 2, 602; Beekes o-o-pe-ro-si 366f. But the relationship between eijos and Ιύί/ηύί based (§5.1).

is even more broadly

3. The question of a different distribution of efjos and kolo earlier on

3.1.1

The system set out in §2.5.2.3 is strictly descriptive; and, if

acceptable, is valid only synchronically and for the end of the Homeric tradition.

From the historical point of view, not one but two much

discussed questions—related but different—arise. The first is about the etymology of ίήος 'noble' (traditionally considered together with that of gen. pi. ίάων 'good(s), good things' Ω 528+). The other amounts to the question of the history within the Homeric tradition, if it can be internally reconstructed, of the modest system of expressions given in §2.5.2.3. Since the ramifications of this second question in some previous discussions have been such as to make it doubtful whether the first one, concerning the history of ίήος 'good,' should even be asked in so simple a form, the next step must be that of seeing whether the set of formulas in §2.5.2.3 have a recoverable history. 3.1.2

At least since Brugmann, 3 3 part—but only part—of this

inflectable repertory of formulas (§2.5.2.3) has several times been seen as a relatively last-minute rearrangement of what the oral epic tradition had had earlier in these expressions.

The indispensable premise for such

analyses is the one-time availability to Homeric language of a Greek reflexive pronominal possessive htfo-lhfo- that could refer back (cf. §2.2.4 above), like its Vedic correspondent svä-, to topics of all three persons— and numbers too, for that matter. The actual claim, in its most modest form, is that the four examples we have (§2.5.2.3) of second-person versions of the formulas in question are modernizations.

To take one

instance—Ο 138—to illustrate, the idea would be to accept at a suitably remote stage the possibility of a reflexive possessive htfo- that was not yet restricted exclusively to third sg. reference, to suppose that the epic tradition created a formulaic | huiios heuoiiott 'own son' ('my, your,' etc. indifferently) at that stage, and to make the kolo read there by Zenodotus (and thus schol. A and a few mss.) a genuine archaism preserved from 33 Brugmann Problem 52-62.

106

Two Studies in Greek and Homeric Linguistics

that time: * ...

methemen

kholon

huiios

heuoiiolt

'... to give up (your) anger

over (your) own son.' The eijos that most branches of the tradition show instead at Ο 138—whatever its origin—is then an innovation, evidently brought about by the synchronic unsuitability of «oto after possessive ko(and ο-) had been limited to use with third sg. topics. 3.2.1

The cogency of this scheme mostly depends, obviously, on

whether a "free" reflexive during the epics' formative period is a good hypothesis, and that in turn on whether the direct evidence for it is believable. That evidence has been collected and discussed more than once, 34 and it would not serve any worthwhile purpose to set it all out yet again here. 3.2.2

What might not be so much a matter of repetition and

duplication, however, would be to single out from the various indications that have been used as epic evidence for an all-person reflexive

heρό-

j hfi-, which—it can be said right now—Greek may well have had, two sets of cases whose probative value has almost certainly been overestimated: 3.2.2.1 Whether or not it is legitimate to claim in a general way that φίλο?, -η, -ov can simply have the meaning '(my, your, his etc.) own' in Homer, 35 there is no denying the descriptive existence in the epics of what is, in effect, a unitary formulaic expression that is found both in a version with

φίλη ν

been duly

and in a version with recorded 36

ίήν

'(his) own.' Beyond this, it has also

that there is a suggestive distribution to be noted

(the only question being what it actually suggests). In third singular reflexive contexts, to be precise, is found the formula (of modest extension): ...]' eijv h πατρίδα

γαΐαν#

(e 42=115, ι 533, ν 52) 37

'... to his (own) homeland.'

34 Cf. Brugmann Problem 17-112, especially 41-78; Leaf Iliad 1, 559-65 (Appendix A); Monro GHD 219f. (§253), 220-224 (§§254-255); Chantraine GH 273f. 35 In favor of this are Leaf Iliad 1, 563f.; Landfester Philos 13-20; Hooker Glotta 65 (1987) 64f. Against it, most recently, is Robinson Owls to Athens 97ff. (especially 97-101). 36 Brugmann Problem 69-72. 37 To which cf., in turn, J καΐ σην is πατρίδα γαΐαυΚ ( δ 476+) and #σ~ην h

πατρίδα

yatav] (Ω 557), though this last cannot come directly into question here since it can never be metrically equivalent to the ίήν (and φίΚην) versions of the expression.

107

Good for You: ίήος (and (ίων)

This is in turn obviously related to: kfi ίν πατρίδί ην πατρίδα

γαίτ]#

γαΐαι>~--#

(Χ 404) 38 and (Ο 505+), 39

which are both also third sg. reflexive, but that is another matter. What is to be brought to the fore right now is the much noted circumstance that in situations where an appropriate possessive would be reflexive in a person other than the third singular, this formula (e 42+ as above) takes the form: φίληυ

ei πατρίδα

γαϊαν#

'...

to (my) own/dear homeland' (E 687, I

414, Σ 101=Ψ 150, Υ 145, β 221, ο 65) '... to (your [sg.]) own/dear ...' (Π 832, α 290, £ 204, λ 455) '... to (our) own/dear...' (Β 140 = I 27) '... to (your [pi.]) own/dear ...' (B 174, κ 562) '... to (their) own/dear ...' (B 158, 454 = A 14, Η 460 = Ο 499,1 47) At first glance, this distribution 40 seems to open the door to the idea, original with Brugmann, 4 1 that in all these passages there originally figured a unitary form of the expression in question: namely ... )' ίην es πατρίδα

γαΐαι>#

(or its phonological predecessor) with a possessive hifo-

that could be reflexive to a topic of any person. Later on—and in fact fairly recently, after this reflexive possessive had been restricted to third singular function (§3.1.2)—all of its instances in this formula that did not make reflexive reference to a third singular were replaced by

φίλην

(whatever it really meant), while ΐήν was retained where the context was in fact third singular. There are decisive objections to this hypothesis.

Aside from the

absence of serious manuscript support for it, the facts on which it is based 38 Cf. note 45 below. 39 For this expression there is a metrically equivalent | σην ττατρίία γα lav the Odyssey (y 117+). 40 Cf. also X 404 (3 sg.): Γ 244 (3 pi.). Cf. note 45. 41 Brugmann Problem 70ff.

in

Two Studies in Greek and Homeric Linguistics

108

form only part of the complete picture here. The missing piece, which proves to be crucial, is that instances of the expression ...]' - - e s -πατρίδα γαΐαν# that go with a third sg. verb take not only the form that begins with ίήν (e 42+ as above), but also—and more frequently, as it happens—the form with φίλην (Δ 180, e 37, ξ 333 = τ 290, σ 148, τ 258, τ 298, ψ 340). The full version of this formula system, in other words, is: Third person sg. subject:

... ίήν es πατρίδα γαΐαυ# and

Non-third person sg. subject:

... φίλην

es πατρίδα γαϊαν#

... φίληυ es πατρίδα γαΐαυ #

This immediately raises the question of how the two third singular alternants distribute themselves. Given that φίλο- is consonant-initial and eo- (for these purposes) is not, it is not surprising that the choice between the alternants is strictly based on what immediately precedes the trochaic caesura: ... öe'f φίλην

es πατρίδα γαΐαν

Δ 180

... και f έηι> es πατρίδα γαΐαν

e 42=115, ι 53 ... νηι I φίλην ... πίμπουσι

... e 3 7 , ψ 3 4 0 ]' φίλην

... νοστησίΐί'ΐ ... νοστήσαντα

φίλην

... ττέμπωμβν \ erjv ... ν 5 2

... ζ 3 3 3 = τ 2 9 0 ... σ 1 4 8 , τ 2 9 8

\ φίλην

... τ 2 5 8

and cf. also: ... αίσα\ φίλην

es πατρι'δ'

ίκέσθαι

ψ 315 etc. Obviously, φίλην would be unmetrical where the ίήν version appears. What is somewhat more interesting is to note that although ίήν would always be technically preceded by a hiatus were it to be used in place of φίλην in the lines just listed, hiatus at the trochaic caesura hardly even seems to be avoided under normal circumstances; 42 and yet such a banal hiatus has been obviated by the use of φίλην in every instance. From the point of view of the internal history of the epics, there are really only two ways of accounting for this (certainly non-accidental) state of affairs. Either the poets of the tradition were themselves sooner or later 42 Chantraine GH 90 (§39a).

Good for You: kf/os (and kauiv)

109

in the habit of avoiding hiatus in the third singular version of the formula—since the means were at hand—by using φίλην in place of ίήν (whether always or mostly or sometimes); or else the oral tradition basically had just the ίήν form of the expression—usually with (an unremarkable) hiatus, and it was only afterwards that φίλη ν was consistently substituted for ίήν after a short vowel. Whatever the relative merits of these alternative scenarios, the distribution of φίλην (after -v\) vs. ίήν (after -V\ or -VC |) as the adjectival component in the third singular version of this formula is itself a very strong argument against Brugmann's hypothesis. For the fact is that in all seventeen instances of f φίλην ... γαΐαι;# in non-third singular situations, there is—as there must be—a word-final short vowel ending the preceding line segment. 4 3

If, now, the rule found governing the

distribution of φίλην vs. ίήν with third singular subjects was the creation and the practice of the poets of the tradition themselves, then the presumption must be very strong that the same rule also governed the choice between φίλην

and (a hypothetical all-person reflexive) ίήν

wherever the subject was a person and /or number other singular.

than third

But that means that in the seventeen non-third singular

occurrences of the formula '( ~-...γαΐαν#

that are actually found, it is

precisely φίλην that would have been chosen in the first place by the application of the distributional rule. In other words, as long as it is assumed that this way of deciding between φίλην and ίήν was the poets' own, it is impossible to point to any actual instance of φίλην that could be a replacement for ίήν—no matter what the person and number of the reflexive reference, in fact.

43 Only I 414 calls for comment.

φίλην ...

The general run of the tradition is ... ϊκωμαι'ί

c'qv with a 1 sg. ίήν vs. φίλην. But that emendation πατρίδ' iμην Ε 213 without apparent variant.

This is evidently faulty. One way to proceed is to emend to f

reflexive that follows the usual distributional rule for is in no obvious way superior to (.μην—cf. #

It is hard to tell what Boiling (Language 23 [1947], 33) has in mind when he says the first of these two possibilities is "simplest and best." In any case it is also thinkable to accept the

v.l.

(Venetus

A)... ϊκωμίϊ φίλην ... with,

e.g., Allen. The difficulty is that

ϊκωμι

cannot be a

form of ϊ κ ω (which regularly seems to take a plain terminal accusative to boot), and therefore, as a form of deponent

ϊκόμην,

would have to be an artificial form "metrically

shortened," in effect, by analogical active inflection so as to be fitted in at | before formulaic

φίλην.,.γαϊαν.

110

Two Studies in Greek and Homeric Linguistics

If, however, only a real free reflexive was "genuine epic," and if either a substitution of φίλην for non-third singular ίην or the practice of using φίλην after -V but ίην after - V and -VC was an innovation in some sense (or if both were), there arise problems that, although different, are just as intractable: Is it to be supposed that at the time of the innovation(s) every instance of | ίην ... γαίαv# that was reflexive to a first person, second person, or a third dual/plural luckily happened to follow a short vowel with (admittedly permissible) hiatus, and could therefore be neatly replaced by φίλην?

A n d / o r were lines that had been constructed with

non-third singular j' ίην... γαΐαν# somehow banished without trace from the epics if and always if φίλην

could not be substituted?

Neither,

obviously, is an appealing thing to have to believe. In fact, these distributional observations arrange themselves instantly and without remainder only around a reconstruction that is exactly opposite.

Namely, Homeric language—if only for the purposes of the

creation and deployment of the formula under discussion—had a reflexive possessive εό?, -ή, -όυ that was used only as a third singular. In addition, naturally, there was φίλος,-η, -ov ' o w n / d e a r , ' which could, of course, be used with subjects of all persons and numbers, 4 4 and was—in part to provide third singular (only) f ίην ... yalavU itself with a consonant-initial form that would expand this formula's domain of usability, and in part precisely to create a version of the formula that could be used where the subject was something other than third singular—the metrical constraints on the use of the f φίλην ... γαΐαν# alternative being everywhere uniform, of course. Nothing would even stand in the way of going so far as to conjecture that the poets originally—out of motives of e c o n o m y — u s e d

only

...V'\ φίλην ... γαΐα.ν# and used it for all persons and numbers, and that the addition to their repertoire of a metrically different, third-singular-only f ίην ... yalavU was a refinement that was first made later—a scenario that fits neatly, in fact, with the observation that f ίην ... γαϊαυ# occurs in only three lines (e 42=115, ι 533, ν 52), two of which (e 42=115 and ι 533) are

44 The use of masc. acc. φίλον along with 3 sg. possessive ίόυ/'άυ is fairly common (ΙΟχ II.) in Homer. Very often the context calls for an actual 'his/her own dear' (E 314, 318, Ζ 474, Π 447). But there is also "colorless" formulaic || ov φίλον viövtt (Φ 330, 378), which seems to be a Streckform of kov/ov υίόν that was created to give genitive || mo? iolott a metrically identical accusative.

111

Good for You: eijos (and ίάων)

extremely similar to one another, and occurs only in the Odyssey.45 But that, to repeat, would only be a conjecture. 3.2.2.2.1

The other phenomenon that may have been too readily

accepted 46 as indirect evidence of the one-time availability to epic language of a Greek reflexive possesive that was usable in all persons and numbers is a usage of the demonstrative 6, ή, τό in the epics as transmitted that has been labeled "difficult" and accordingly supposed to result from the replacement of non-third person forms of *huö-/*huä-

by metrically

equivalent forms of the "article" that are, however, functionally eccentric as such. The argument takes as its point of departure sets of expressions that contain a form—or various forms—of the third singular reflexive possessive. It might start, for example, with a set like: . ου παιδό? ϊ Ζ 4 6 6 , 1 2 6 9 , Ψ 2 2 2 , Ω, 85, ο 3 5 8 , τr 4 1 1 ,

I ού παώός

... Π 522

I ού πατρός

... Α 404

ω 56

.ού πατρός\

η 3

Of these expressions, in their earlier line position, a second singular correspondent is found in: ... σον πατρός]

LI 504, β 271, 47 ρ 106

And, similarly, a third plural parallel is: ... σφοΰ πατρός

| A 534

To the question of how these phrases are rendered in contexts where the person of the topic has no associated possessive that could offer a masculine genitive singular that is metrically equivalent to ού/ σου/

σφοϋ

the answers are: 45 If so, this new f cqv (ς πατρίδα on unique but parallel f I f ) iv πατρίδι du. and pi. f ψι'λϊ; tv ττατρί&ί yaiytt

yo.io.vti (Od.) might look as if it had been modeled yaigtt

(X 404), which acts as the 3 sg. version of 3

(Γ 244). On the other hand, it is notable that Ιυ πατρίδι

yaLyff is almost always preceded either by an unshortened long vowel or diphthong

(| ήδη ev...ff ρ 157; ]' ky ev...tt X 404; f φιλτ; (| φθίμίνος

Γ 244, ω 266) or else by a brevis in longo

iv...tt Θ 359), the only exception being the Od. formula Ϊ tow ev πατρΐδί

(Θ 461=f 143). And just as this could mean that | ψθίμίυος

ίυ.,.ϋ

yalrjtt

(II.) has something to do

with )' ίων ev... (Od.), it could also suggest that, contrary to first impressions, ϊ φίλτ]

iv...#

(IL, Od.) and j' ky Ιν.,.Ά (11.) simply adopted the structure of ]' φίΚην es...ff (II., Od.) and ϊ (ήν «...# (Od.). 46 Brugmann Problem

45ff.; Leaf Iliad 1,563.

47 The σον is non-reflexive in this instance, reflexive in the other two.

112

Two Studies in Greek and Homeric

Linguistics

... του vreuöos'i48 'of (my) son' λ 492 49 ... τον 7τατρός'ί 'of (my) father' Τ 322, β 134 (?), 50 π 149 and 'of (your [du.]) father' Λ 142 The next step is to conclude from the apparent use of o, η, τ6 as a (reflexive, as it happens) pronominal possessive ('my, your [du.]') that all is not what it seems in these passages, and then to remedy their faulty language by assuming that a ov παώός 'of (my) own son,' with a free reflexive h(f)6-, is what is genuinely epic at λ 492, and so too a ov πατρός 'of (my) own father' at Τ 322, β 134 (?), π 149 and a ov πατρός 'of (your [du]) own father' at A 142. Finally, for good measure, it is proposed that even σον πατρός 'of your (own) father' (Ω 504+ 51 ) and σφον πατρός 'of their (own) father' (A 534) are likewise modernizations of *hfo πατρός 'of (my, your, etc.) own father.' ... τον πατρός]

3.2.2.2.2 For two kinds of reasons, however, this line of analysis is probably illegitimate. First of all, if the use of the demonstrative το- in circumstances that admit an implicit reflexive possessive is really a reliable indication that a non-third person Μ/)ό- has been supplanted, then an instance that has surely had this history—and in fact has figured as a prime example in this regard52—is Φ 412: οϋτω κΐν της μητρός

kpivvας

ίζαποτίνοις

'Thus would you fully satisfy the Erinyes of (your own) mother' But if this line, before a putative clumsy modernization, had ... Η/)ής μητρός'of (your) own mother' 53 and if this is reconstructed precisely because της is somehow out of place here, there is no conceivable explanation of why the passage was modernized with an unsuitable της 48 More properly μοι τον iraMs.

Cf. §3.2.2.2.2 below.

49 Cf. also Ιμον υίόν\ Ω 227 andφίλον υ'ών\ '(my) own/dear son' Ε 377. 50 It is actually far from clear that Telemachus is talking about his own father here and not Penelope's. But if there is any implicit possessive, roO is odd on either reading, since both της 'her' and ϊμον 'my' would fit in its place with no dislocation. 51 Boiling Language 23 (1947), 32. 52 Brugmann Problem 47,49; Leaf Iliad 1,563. 53

ήί μητρός is even said to be confirmed by the reading και μητρός of a single

papyrus (cf. Allen ad Φ 412). So Boiling Language 23 (1947), 32 with references. But there is still no explanation of why a της, if it were illegitimate, would have been chosen instead of σης to replace it. Cf. in any case further down in this paragraph, where it is suggested that της is perfectly in order as an attributive antecedent of the relative that follows in the next line.

Good for You: ef/os (and ίάωυ)

113

and not with a perfectly unremarkable

σης,

'of your (own) mother' exactly parallel to ... above 5 4 ).

resulting in a *... σης σον πατρός]

(il

504,

μητρός] ρ

106 as

Paradoxically, in other words, if there were something

objectionable about this (Φ 412) use of της, it presumably would not be here in the first place—especially since the alternative σης is not only perfectly admissible but well paralleled too.

But if there is nothing

objectionable about της after all, there is no longer any real reason to suspect that it does not itself represent a legitimate usage. Analogous conclusions are at least consistent with the circumstances of the cases in which

του

τταώος\

(λ 492) and

του πατρός|

(Τ 322+), as

above, occur where it is an implicit first singular possessive that would be admissible.

In particular, if there were something defective about

του 7τατρός]

'of (my own) father' in its three occurrences, and if this

indicated that an obsolete first singular

*h(f)5

-πατρός]

underlies it, it is

difficult to imagine why the modernization did not instead make use of the alternative offered by the tradition in: 55 ... ίπά

μοι πατρός]

άμύμονος

ϊγγύθι

vaUi

(η 2 9 )

'... since it is near my noble father's' This use of

μοι

simply as an adnominal possessive is of the type ... μοι

(Π 517) 'my arm,' 5 6 ... μητίρι

μοι

... (β

TOL γούνατ

χΐίρ

... (X 335) 'your knees,' and especially

50) 'my mother;' 57 and that the first singular personal

pronoun (like those of the other numbers and persons) can be used reflexively as well as not in Homer is made clear by such cases as ...

Ιγωυ

ίμΐ λύσομαί... (Κ 378) Ί will ransom myself.' 58 The open but unexercised option of a *... πατρός])

for a case like Φ 412, or of *... μοι

πατρός]

σης

(cf.

μητρός] η

(cf. ...

σοϋ

29) for Τ 322, may

therefore cast some doubt on the hypothesis that reflexive possessive ης and ού, with second- and first-person value, have been replaced by της and τοϋ; and this may consequently argue at the same time that, whatever its precise nuance, the demonstrative ό/το- in a possessive-implicit context is a legitimate usage. Any such argument would, of course, be strictly a negative one; and that raises the question of whether there is not 54 Cf. note 47. 55 On the other hand there is an obvious reason w h y it is not μοι παιδό?]' at λ 492: a μοι already occurs before the του itself. 56 Also μοι αίμα ' m y blood' and μοι ώμο? ' m y shoulder' in the same passage. 57 Chantraine GH 2, 71f. (§90) for these and other examples. 58 Chantraine GH 2 , 1 5 4 (§228); Boiling Language 23 (1947), 32.

114

Two Studies in Greek and Homeric

Linguistics

something positive to be added in the form of parallels for the use(s) of the "article" in question here. In some cases, such support is not very difficult to supply—so, for example, in the case of Τ 331: cos αν μοι τον παΐδα θοη ein νηϊ μελαά>η ... Ιζαγάγοις ... '...to take my son in (your) swift black ship ...' The notion of "restoring" 5 9 a *ov παΐδα]' here is especially unappealing— not only because the absence of a plausible first singular topic makes a reflexive questionable, but also because a possessive of any kind after μοι is redundant at best and, finally, because μοι τον παΐδα f 'my son' is so clearly of a piece 6 0 with την γαστέρ'|' 'my belly' (σ 380), μοι τον

oveipov]

'my dream' 6 1 (r 535), and more generally # σοι το γέρας 'your prize' (A 167). 6 2

But these are also, in turn, parallel to another of the alleged 6 3

demonstrative-for-reflexive possessive cases collected above (§3.2.2.2.1): μοι τον παιδός]' 'of my son' (λ 492), with the consequence that a reflexive ov '(my) own' (itself very forced in any event) in place of τοΰ is completely unconvincing in this passage too. At Λ 142 (§3.2.2.2.1), where *ov πατρός] 'of (your [du.]) own father' has been seen to lurk behind the transmitted τ ο ϋ πατρός)',64 the demonstrative—though this time with a different value—could perhaps be accepted as an instance of an zsie-like use: 'of that father (of yours [du.]).'

Such a hostile deictic, not too uncommon in H o m e r , 6 5 is

comparable, for instance, to that of σ 333: η άλύβις 'ότι TIρον ίνίκησας τον αλητην 'Or are you beside yourself because you beat Irus—that vagrant'? The same might possibly apply to the της μητρός

of Φ 412—'of that

mother (of yours)'—whose demonstrative is in any case corroborated by the failure of σης to appear in its place. But it seems better to explain both the της μητρός .../ η ... of Φ 412 f. and the τοΰ πατρός...! ös ... of Τ 322f. as 59 Brugmann Problem 49, 109; Boiling Language 23 (1947), 32. No variant suggests this in any case. 60 Monro GHD 230f. 61 Or easily so interpretable. 62 To which cf. A 120:... ö μοι ye/>a?|| ... 63 Brugmann Problem 48,49,109. 64 Boiling Language 23 (1947), 32, with further reference to Wackernagel (SU 150), who considers both ού and του "schlecht," preferring σφοΰ. 65 Monro GHD 229f. (§261.2).

115

Good for You: ίη os (and ίάων)

simply showing, in an attributive form, the kind of demonstrative that, as a substantive pronoun, so commonly anticipates a relative66—e.g. (T 228): α λ λ ά χ/377

τον

καταθάπτίΐν

ος κε

Oavyai

'... him who ...,... the one who ...' If so, Φ 412f. and Τ 322f. offer nothing more startling than '... of the mother who ...' (i.e. 'yours') and'... of the father who ...' (i.e. 'mine'). As potential instances of non-third singular reflexives that have hypothetically been replaced by problematical forms of the article in this particular set of parallel expressions (§3.2.2.2.1), that leaves only two Odyssey instances (β 134, π 149) of τοϋ πατρός] 'of (my own) father.' And without being able to explain why a theoretical first singular reflexive *οϋ πατρός]—if it had to be made to look more first singular—was not redone as μοι πατρός] (as in η 29), it might even be safest to assume that τοϋ πατρός] at β 134 and π 149 is simply already an example—in late epic,67 to be sure—of the "possessive article" 68 that is so common later on—e.g.: 6 πατήρ, 'my father' (Ar. Wasps 139), ανιψιός τοϋ πατρός 'my father's first cousin' (Andoc. 1.47), τ ω πατρί 'your father' (Ar. Clouds 860), τον πατίρ" 'their father' (Theogn. 277), την θυγατέρα 'his daughter' (Hdt. 1.2.3) etc. 3.2.2.3 Since the object here is not to reconsider every example of every Homeric phenomenon that has been invoked as evidence of the modernization of an all-person reflexive possessive h(e)fo- out of the text, but rather to suggest—partly through a re-examination of the details—that at least two very finite classes of the "restored" instances of such a h{(.)focan be questioned, and since that has now been sufficiently done, it seems useful to stop here and say in summary that there are tangible advantages (§§3.2.2.1-3.2.2.2) to the view that, in the two modest formula systems that have been looked at, the poets used somewhat heterogeneous elements to inflect them for person while maintaining metrical and thus positional uniformity: 6 6 Chantraine GH 2 , 2 3 6 (§345). 67

την χαρ' at Ψ 75—especially since the "possessive" article is c o m m o n e s t with

w o r d s for (relatives and) body parts—might even indicate that this use of the "article" is not even necessarily confined to late epic. But cf. Chantraine GH 2,164

(§243 Remarque 2).

In any case M o n r o GHD 231 (§261, 3b), where ψ 75 is labeled " a n o m a l o u s , " seems a bit too pessimistic. 68 Gildersleeve Syntax 2,227f.

(§534).

116

Two Studies in Greek and Homeric Linguistics

Possessive sg.

1. I

φίΚην...γαΐαν#

2.

"

3.

"

du. pi. 1.

"

2.

"

3.

"

\£ην...γαϊαν#

2. Possessive sg.

1. μοί πατρός \ 2 . σοϋ

πατρός]'

Λ

Demonstrative ,



του πατρός | 69

τον πατρός f ( Τ 3 2 2 ) της μητρός | ( Φ 4 1 2 )

3 . ον παιδός f

du. pi. 1. 2.

του πατρόςϊ

(Λ 1 4 2 )

3. σφοϋ -πατρός \

3.2.3 Although it may not in the end be convincing to assume that I φίληυ ίς πατρίδα γαΐαν# 'to (my) own/dear homeland' (I 414+) etc. is a superficial modernization of a version with f Ιην ..., or that του πατρός ] 'of (my) own father' (T 322+) simply replaced a *h{f)o πατρός|' etc. at some point, it remains possible that Greek—and the language of the epic tradition in particular—did have a reflexive possessive *hifo-l *hfo- that was not (yet) restricted to third singular reference. There are indications of three kinds favoring this possibility. 3.2.3.1 First, the comparative evidence of Vedic sva- (§2.2.4), itself backed up by the indications offered by Slavic and—up to a point—by Italic, 70 go far toward establishing a PIE reflexive possessive *s(e)uo- that could make reference to a topic of any person or number.

69 If the Odyssey really has two "possessive article" instances—§3.2.2.2 above. 70 Cf. e.g. Brugmann GVGIS 2,395ff. (§§391ff.).

Good for You: ΐηοί (and ϊάων)

117

3.2.3.2 In addition, the direct evidence for such an all-person reflexive *hef0-/*h/r0- offered by the Homeric tradition cannot easily be simply dismissed. There are first of all four well known 7 1 instances of a reflexive possessive o-l η- that could make reference to a first-person topic, and where the reading is unassailable: # θάρσίΐ ω 'in my (own) boldness' (H 153) # oicriv evl μζγάροισι 'in our (own) hall' (δ 192) # Tjs γα LT] ς 'my (own) country' (ι 28) # ... φρβσϊν flaw] 'in my (own) mind' (v 320) It has to be admitted that the first of these is a weak example. At Η 153 θάφσίί ω could very easily be said of the θυμό; in the line before and mean 'in its boldness.' 72 A less questionable example, but still not a compelling one, is the second, since the οίσιν of δ 192 could conceivably mean 'his (own).' 73 This amounts then to three instances (δ 192, ι 28, ν 320)—even if one of them (δ 192) cannot be insisted upon—of a free reflexive convincingly transmitted in Homer. As it happens, all three of these phrases recur in circumstances where the reflexive possessive o-l fj- has an ordinary third singular reference: To δ 192 t 28 ν 320

cf. # οίσιν evl μεγάροισί 'in his (own) hall' (a 269) # ήΐ γαίης 'his (own) country' (a 59) ) *eeu- to *eeu- followed by hyphaeresis of *eeü- to eii- (i.e. ηύ ) really convincing: In no category of forms showing Homeric "hyphaeresis"—and this explicitly includes cases of consecutive identical vowels—do the epics show the hyphaeresized form to the exclusion of the older, non-hyphaeresized form. Typical, in other words, is 2 sg. μυθίαι (< -«'eat, which can actually still be "restored" in this case) at β 202 only in addition to μυθΛαι at θ 180. Highly atypical would be eii- < *eeu- with *eeü- nowhere found. 126 It seems worth noting that if Greek ίάων—and έήος—do not themselves confirm *hjuesu- for the *Huesu- supported by Indo-Iranian evidence (Ved. purüväsu- 'with many goods' [< *-u-Huasu-], visv&vasu- 'having all goods': GAv. vispä.vohu- [Mayrhofer EWAia 2, 533f.]), then the initial H- can be specified as h,- only on the relatively problematical and thus tenuous assumption that Go. iusiza 'better,' which goes back to a comparative stem *eus-is~, more specifically reflects a *ft,eus- that represents a schwebeablauting alternative to what would consequently be *htues- (as in Ved. vasu- etc.). If *h,ues- (rather than 'Hues-) is nevertheless set up in what follows here, it is not to be insisted upon.

136

Two Studies in Greek and Homeric

Linguistics

make it a neut. acrostatic because of the direct testimony of Hitt. äsSu, -uw'good thing' and the indirect testimony of Ved. väsu, väsv- 'id.;' and, taking a cue from Av. varjhuuam (< *h1uesu-öm), one would most particularly make the required gen. pi. the *hjesu-0m > Gk. *ehuön which Hoffmann,127 citing Szemerenyi, also admits as one possibility (§4.2 e). 4.5.2 If the hypothesis is now recalled (§4.3.3) according to which the subsequent survival of such an archaic substantive is attributable to its having gotten into a fixed collocation, formed early and kept late, with *dötor- 'bestower,' it is advisable to suppose that that collocation was first formed at least as early as the stage at which the Greek form of the substantive was this *ehuön—if not, in fact, even earlier. Recalling in addition that the primary form that this all-important collocation took was evidently the vocative one that is eventually seen in 'ύώτορ ίάων (since the other traditional-looking form, δωτ%>εϊ ίάων, shows unmistakable signs of being a nominative-pluralized modification: §4.3.2), everything would point to a voc. *dötor ihuön as the fixed expression that allowed the old genitive plural to survive. 4.5.3 At the same time, however, this reconstruction immediately suggests a reason for the failure of this impressively pedigreed piece of diction to appear in the epic right away. It presented a cretic sequence (*dötör ehu-12S); and this cretic could not be gotten rid of in the usual way 129 —namely by lengthening its medial short. Α *dötör ihuön as the metrical lengthening—by way of analogical remodeling—of a vocative was excluded for the obvious reason that it would have had to be taken as a nominative and precisely not as a vocative. Nor can it simply be assumed that a *dötSr ehuön, with a hybrid agent noun comparable to what is eventually seen in δωτη/Οίί ίάων, would have been a possible vocative either at any early stage. To be sure, the agent nouns in -τηρ did eventually generalize that (nominative) suffix allomorph to the vocative 127 Hoffmann Aufsätze 2,603. 128 The numerous Homeric indications of position made by initial, and therefore tautosyllabic, *hu- (type θυγατέρα ην It cf. Chantraine GH 146) make it a fortiori likely that internal, and therefore potentially heterosyllabic, -hu- (as in our *ehuön)

also made

position—and presumably continued to do so up until the sequence became either -uwith compensatory lengthening or -uu- (e.g. *nasuo- 'temple' > *näuo-, whence Horn, νηός etc., or > *nauuo- > Lesb. vavos. Cf. Burger REIE 1, 447ff., Kiparsky Language 43 [1967], 619ff.). 129 Chantraine GH lOlf.

13 7

Good for You: (T)OS (and ίάων)

(as they had to the other cases, sg. and pi.) along with its oxytonesis. But the presence of what is evidently a relic form in Attic

'savior' makes

σώπρ

it unclear just how early an innovation this was. 130 Perhaps significantly, the one Homeric instance of a voc. sg. to this kind of noun is indeed transmitted with a Κωβητηρj

-τήρ,

but not a metrically guaranteed one:

ά^λαε|| ··· (Λ 385). 131 If, in any event, neither

κέρα

τοζότα

*dötör

nor

*dötir

was workable as a substitute for the metrically impossible vocative

*dötor

for some considerable time, then the collocation

*dötor

ehuön

that is

practically indispensible to explaining the survival itself of the isolated u-stem genitive plural will simply have been an item which, though appropriate to epic, was unusable there; and it must therefore have led its traditional—presumably cultic—life outside the epic repertory for an indefinitely long part of the time during which the epics as we have them were being formed. 4.5.4 That is not to say, of course, that this

would not have

*ehu5n

been liable to undergo any further developments during that period. There are in fact two sound changes that come into question. 4.5.4.1

The first of them is aspirate anticipation. In vowel-initial

words 132 an internal -h- from pre-Greek s or i133 occurring in the environments

#V(R)h(R)V

(these environments perhaps being necessary

but not sufficient) could be anticipated as a secondary aspiration of the word-initial vowel, producing a fthV(R)h(R)V,134 iepoy 'sacred'
*heuo- (Kiparsky Language

43 [1967], 623ff.).

This

expectable present, however, has been analogically refashioned to show the -eu- that was preserved, phonologically regularly, in the aorist *eus-s-, as also happened in

γεύομαι

'taste,' with analogical γεν- from the aorist ycvrra- < *geus-s- (Kiparsky Language 43 [1967], 628). Since aor. ίΰσα- (Hom.+) also has an initial h-, it would seem that (ϋω, at the *hehuoor *heuo- stage (cf. above), had its initial aspirate generalized to the aorist, whence *heusa-, at which point the present, in turn, was analogically remodeled after the aorist as described. If, in any case, the aspirate originated in the present, (ϋω may serve as an example of h- anticipation despite the remodeling. 145 See earlier in this section and cf. note 130. 146 This seems to hold true also for the other category (or two categories?) in which a secondary initial aspirate has been said to result specifically from the anticipation of an internal -h-: Namely— 1) The aspiration of initial u- to u'1- in cases like ίαυόί 'robe' (//.+) < uehano- (Myc. we-a2-no) and άμα 'garment' (//.+) < *uehma (cf. Sommer Lautstudien 115-119 and Lejeune Phonitique 176f.) 2) The supposed aspiration of unvoiced stops in the environment tt(s)_(r)Vh in a few instances (Sommer Lautstudien 53ff.) like θρΐναξ 'three-pronged fork' (Ar. +) < *trihnak- ? (DELG 441; GEW 1, 683f.) or Ορύου 'reed, rush' (II.+) < *truho- V. (DELG 443; GEW 1, 688). This must be considered a sporadic and even doubtful phenomenon, however, both because few of the etymologies marshaled in its support (Sommer Lautstudien 54-68, 6873, 73-74) are really compelling and because of the lack of plausible explanations (analogical or otherwise) for the failure of this proposed sound law to apply in cases where it might have been expected (Sommer Lautstudien 78ff.): α) τpäs < *trehes < 'freies or κρίων 'lord(ly)' < *krehont~.

140

Two Studies in Greek and Homeric Linguistics

in w h i c h *euihuön

d e v e l o p s r e g u l a r l y t o *heiön

is όίομαι / A t t . οΐμαι


eu^er^ksa-

> Att.

άρξαι and future euer^kso- > Heracl. αφ-, ϊφ-ίρξοντι; or ίορτή 'holiday' < *uJ1eu^lor^lt- < *ueuor^t- (so also ϊρση 'dew' < eueres-, on which cf. Peters Untersuchungen 316f. note 262 with another example and further reference to Salvaneschi SILTA 4.1 [1975], 80). 2) A M'1 may also be produced by a -sC- or -s# one syllable later in the word (the socalled "ϊστωρ rule"—cf. Sommer Lautstudien 119-24, Lejeune Phonetique

176f.).

In the

most familiar examples the affected u is initial: μστωρ (so in Boe. DGEEP 491.18 etc.) > {fftiaTuip

'expert, judge;' *uespero- (cf. Locr. fea-napiov DGEEP

362.10-11) > ea-ncpos

'vespertine.' But this rule, like the one above, could also apparently produce an internal -u'1- that could end up anticipated by initial aspiration: 'tauos > *tau^os > *t^au^os > Aeol. 6as (vs. Att.-Ion. τίωs) 'so long' (Schmidt Glotta 53 [1975], 41f.; Peters Untersuchungen 316f. note 262). The question that then arises is whether primary -h-, like -r^- and -sC- or -s#, may be expected to have effected the aspiration of a non-initial -u- that could then itself have been anticipated by a secondary word-initial aspirate—i.e. a hypothetical development in the present case like: *euehuön > "eu^ehuön >*heu)lehuön,

whence ultimately ίάων.

The

answer seems to be that this is neither excluded nor guaranteed. The reason it is not guaranteed is that the sound law anticipating primary -h- clearly cannot simply be collapsed with the other two, if only because primary -h- can be anticipated either by a preceding vowel (iepos, tvo> etc.) or—as far as we have seen so far—by a u one syllabic nucleus to the left (αίμα < * u^ehms), while the aspirates conditioned by -r' 1 - and -sC-/-s# are apparently always cases of iJ1 (άρξat, ϊστωρ) and never hV (άρκτος, αρσην etc. etc. and not *hap-; Ιστι—whence

άμι, άσι etc.—and not

*ht(TTL etc.). In fact, it is possible to go still further. Since the hypothesis by which internal -hcould aspirate an unvoiced stop one syllable earlier (above in this note) is required by practically no convincing equations, it can be maintained that internal -h- conditioned the anticipatory aspiration only of preceding vowels and u. But it also turns out that the only good examples of uh by assimilation to a following -h- (since ήλος 'nail,' ίνα 'sinews'/

141

Good for You: efjos (and ΐάων)

4.5.4.2 The next sound law undergone by *hihuon—either soon after h- anticipation 148 or perhaps even concomitantly with it—would have

wiov 'occipital b o n e / and Theran htapa. '?' [Sommer Lautstudien 116-119] are uniformly unconvincing—cf. DE LG, GEW s. vv. ήλος, is, Ιαρόν) are all to be found among the forms of a single verb and its derivatives (cf. DELG 350f.; GEW 1, 521f.)—namely pres. -μαι (Hom.+)/ (ϊνυμι,

ΐυυυμι,

-μαί; aor. ϊσ(σ)α, -άμην (Hom.+); pf. είμαι, άμένος etc. 'dress, get

dressed, be dressed' along with άμα (Hom.+), eVöo? (Hom.+), ΐσθής (Hom.+) 'clothing,' we-a2~no (PY) and iavoi (Horn.) 'robe,' ίφίστρίς

'cloak' (Xen.) etc.

To be noted quickly before proceeding is only that it has long been seen (cf. DE LG 350, GEW 1, 522 again) that what is synchronically the deponent perfect of this verb (είμαι, ΐσσαι,

ϊσται; plpf. ϊστο; ptcpl. άμίνος

etc.) is a restructured middle root present

corresponding to Ved. vdste 'dons' and Hitt. westa 'wears' (cf. Oettinger Stammbildung 303 with reference to Eichner). Assuming now anticipation of -h- by u, the idea would be that initial H'1- was regular in pres. *u^ehnu- (> dvv-,

later remade on the aor. as evvv), in the pres. (—> perf.) forms

*uheh -mai and *uheh -meno- (before the advent of pf. ptcpl. accentuation -meno- if Sommer is right about the accent's role—cf. note 140), in *u^ehma (> άμα etc.), and possibly in *u^ehano- (if it may be assumed that tavos, which is essentially epic only, is a genuine contribution of a non-psilotic dialect to the Homeric repertory). From these forms initial uh- (> h-) would have been spread throughout the verb (aor. ϊσ(σ)α-, fut. ϊσ(σ)ο- etc.) and to all the nominal derivatives (e.g. (φ(στρίς) where it was not actually disfavored by the dissimilatory presence of another aspirate (έ'σθο?, Ισθής). It is obvious, however, that a number of the forms in question would also be liable to undergo the ϊστωρ

rule. Chief among them would be the 3rd sgs. of the old root

present (—> perf.) *uis-tai > 'u^estai and imperfect (—> plupf.) "ues-to > 'u^esto, along with the predecessors of such forms as du. ϊσθην (Σ 517). But it is conceivable in addition that the instances of -sC- that conditioned this kind of uh included -ss-, in which case aor. *uessa- and fut. *uesso- could have become *u^essa- and *u^esso~; and this even in Attic and Ionic if the ϊστωρ rule can be dated earlier than the -ss- > -s- simplification that produced ίσα- and έ'σο- here (along with τρέσα- to τρίω, 'ίσομαι, γίνεσι, where that happened.

μίσος etc.) in the dialects

However that may be, the ϊστωρ rule will have produced a

different set of forms (whether έσται etc. only, or those plus aor. ('σ(σ)α-, fut. 'έσ(σ)ο-) from which uh- could in principle have been generalized to other verb forms. Among nominal forms, the ϊστωρ

rule could perhaps be directly responsible (by way of *uestr-id-

*u^estrid-) for ϊφ-ίστρίί, Hsch. [Latte HAL] Πστρα

>

but this could itself be derived from a *u^istro- < *uestro- (cf. (= /:ίστρα)· ϊν&υσις, στολή, ιμάτια and Ved. västra-, Αν. vastra-

'garment'), or the (u)h- could have been generalized from the verb. The question then becomes that of whether all these t A forms in verbal and nominal derivatives of this root are to be attributed to -h- anticipation, the ϊστωρ rule, or both. There is one substantial advantage to assuming that the ϊστωρ rule is responsible and that nominal άμα, ίανός, pres. üvv-/ acquired (ufi-

ϊννυ-, and—if necessary—aor. έ'σ(σ)α- and fut. ('σ(σ)ο-

from pres. —> perf. (uftestai

etc. and nominal (ufrestro-.

Namely, this

would make it possible to maintain that -h- anticipation resulted only in the aspiration of

142

Two Studies in Greek and Homeric

Linguistics

been the elimination of internal -h- with compensatory lengthening in most dialects, and certainly in the two—namely Ionic and Attic—whose features would be most likely to characterize and/or influence a late epic form like ίάων. become

*/iguön149

In all such dialects, at any rate, *hehuon

would have

and the traditional vocative phrase being envisioned

here would accordingly have become *dötor hiuön. Since h- in instances of word-initial hV- does not make position with a final -C of a preceding word in Homer, the further inference would have to be that the expression

*dötor

hiuön,

by the prosodic

rules

contemporaneous with it, would have contained just as cretic a sequence as its pre-fonm *dStor ihuon had always had.

The expression would

consequently still have been unusable. Furthermore, a vocative *dötör was certainly still out of the question, and there is no reason to suppose that a vocative *dötir could have became admissible any time soon after the *d$tor hiuön stage had been reached. 4.5.5. At a significantly later date, however, a solution to the metrical problem attaching to this phrase would have presented itself. And if it was this solution that was adopted, as it apparently was, the relatively late versification of what should be an archaic expression is no longer puzzling: The stratagem finally employed amounts to a kind of diektasis that presupposes loss of intervocalic u and the like-vowel contraction that followed it, and the closest single parallel is also strictly late epic. 4.5.5.1 Homer has a certain number of cases in which forms in initial eeCC- have by-forms in eCC-.150

In most such instances it seems that

a preceding vowel (Upos, eiito), and incidentally that -sC-/-s# (Γστωρ etc.) and -r^1- (ϊρση etc.), at least as their immediate effect, triggered only the aspiration of a u one syllabic nucleus earlier in the word. And that, in turn, allows one to regard the lack of any trace of aspiration in tap 'spring' < *uehdr < *«es-r (cf. Av. varjri 'in spring,' L. uer 'spring' etc.) as regular. In any case, however, as long as (ίνυ-Ιΐυυυ-,

είμαι, άμα.,

and ίανός

do not

unconditionally demand u- > u'1- by -h- anticipation—which they do not—there is no actual parallel for a hypothetical development of *euehuön to *eu^ehuon

in the case at

hand, and thus no guarantee that it is assumable without further ado. 147 Sommer Lautstudien 13. Note also αάρω < *auehro- with no anticipation. 148 Cf. note 136. 149 The timing and dialectology of psilosis can be ignored for these purposes, since ίάων evidently was not the contribution of a psilotic dialect—or if it was, its Attic correspondent had h-. 150 Chantraine GH 181f.

143

Good for You: tijos (and ίάων)

original *eueCC-

lost its - « - , contracted to eCC-, law 151

underwent a secondary Osthoff's 'enclose' {ttpya

*euergo-

c o d d . Ψ 7 2 ) > ergo-

and then—in Ionic—

that produced a shortened eCC-:

A 617+ etc.) > ergo-

(Att. el'ργω,

ίίργουσι

Ρ 571 etc.)

(Ιργομίνη

'long for' (εε'λδομαt Ξ 276+ etc.) > *eldo- > eldo- (tkberai

*eueldo-

Ε

481 etc.) *euersx 'dew' (ίίρση Ψ 598+) > *ersS- > *ersS- (έρσαι ι 222) Once both ee- (eepy- etc.) and e- (ίργ- etc.) were present, side by side, in the epic repertory and available for use where each would do the most metrical good, it could easily—or perhaps only—appear to the poets of the tradition that the ee- form was a distracted version of the e- that was, after all, their contemporaneous form. That this was the analysis given to ίίρση

vs. ϊρση,

suggested by the markedly different case of hbva tbva}51

Here an etymological aebva

bride price' I 146+ < *n-h2 αποινα#

A 13+) and *μυρι

poets themselves as άπepiaia/

'wedding gifts' vs. cf. avathvoi

in formulaic *α·π(.ρίσι

uedno-153)

άττΐράσι

(< *h2ued-no-

for example, is

αεδι>α# (Π 178 cf.

(Π 190+) was re-analyzed by the

αώνα

εδνα.154

μυρία

'without

This re-analysis was so

successful that what is descriptively a distracted form of this strictly epicanalogical ebva—namely original aehva,

eehva—seems

to have been substituted for

where that was metrically required, and/or to have been

put to use in subsequently composed material: I και άρτυνίουσίν I αττοΰωσιν I και (ζώφίλλΐν ου τοι f eebvivrai

277=/3 196)

eebva

(a

eebva

(θ 318)

ίώυα

(ο 1 8 ) e t c . a n d cf. a l s o :

κακοί

ΐίμεν

(Ν 3 8 2 )

4.5.5.2 For an account of the hypothetically very old vocative phrase that eventually turns up as late epic δώτορ ίάωυ,

it may be that the

segment seen in θ 318, as above, is particularly important. For it seems plausible that it was something like \ άποδωσιν analyzed as a distracted ibva—for bihovrestt

with an

which cf. also especially } καΐ

(λ 117 = ν 378) = *| και äebva

151 Peters Untersuchungen

eebvati,

bibovrestt—that

eebva

ebva

suggested to some

316f.

152 Wyatt Prothetic 37f. with references to Beekes Development 58f. 153 Peters Untersuchungen

317 η. 262.

154 On the eventual initial aspirate of tbva Untersuchungen

see Wyatt Prothetic

as in notes 151 and 152 above, but also cf. §4.5.5.3 below.

and Peters

144

Two Studies in Greek and Homeric

Linguistics

poet fairly late in the tradition that the way to versify the hitherto intractable set expression

*δωτορ

segment by providing

'goods, blessings' with a distracted by-form of

the kind that in

Ιιίων

*heewv

expression of the shape δώτορ 4.5.5.3

with strictly analogical distraction and an *ΐίων#.

A few points remain, of which the first is that this

hypothesized diektasis of exact

was to turn it into an adonic

'bridal gifts/ not too semantically distant a model, had

el>va

The result was a

eebva.

Μων

analogue 155

*ϊων

to

*ΐίων

finds what is perhaps its most

in eeis, the irreducible

(οφθαλμοί]

eeis

ίνίκαιτο)

distraction of ds 'one' 156 found only in Hesiod (Th. 145)—and therefore presumably a product of the same later epic stage as saw the creation of *ίίων.

Secondly, if an association of this kind with helps with a putative

*ϊωυ

edva

'bridal gifts'

: eebva

: *e (ων 'goods, blessings,' it is the same

association that might provide an explanation of the otherwise mysterious initial aspiration with which difficulty here is that if assumed by-form

*kioiv

ϊ'όνα

ebva

is so emphatically transmitted. The only

really has its aspiration after

This, in turn, means either that readings like εώυωταί *'ίων,

4.5.6

a contrast

ebva : tebva

was favored by

'ιώνa

as well.

(Venetus Λ at Ν 382)

should be taken very seriously indeed or that while initial to

then the

*ϊων,

would very much lead one to expect

ϊρση

ebva :

does owe its

έί'ρση.157

Whatever the truth may be about the aspiration of

however, the next development took place when

*ίίωυ,

ϊ'όνα,

still clearly

analyzable as a genitive plural, but an isolate without paradigm membership, was remodeled, evidently by poets who were unaware or no longer aware of extra-epic *6ώτορ number

'ίων,

to

ίαων

in assimilation to the large

of -άων genitives (in particular those shaped - — )

characteristically occupied the very same place in the verse that

*ΐίωυ

that did.

If it was precisely a *ίίων that was refashioned in this way, the remodelers may have imagined that they were dealing with a metrically lengthened form of their own Ionic α-stem genitive plural -ίων, and as a result the usual way of working a long penultimate into a form of that category

155 I.e. ie- for etymological e- in both cases. 156 e'ets as a form of ds w a s itself perhaps most directly suggested by (χίκοσι (for h-) beside ακοσι. 157 Cf. Peters Untersuchungen

317 η. 262, where it is suggested, however, that ϊδνο.

vs. Ιώυα is directly imitated from ϊρση vs. ίίρση in the first place.

145

Good for You: eijos (and ίάων)

(namely -αων) was simply substituted for what looked like a highly unusual one

(-ίων).

As, finally, to } Ιιωτήρας eaowtt



325+) and unique independent ίάων

(Ω 528), it is naturally impossible to tell whether they go back to expressions that arose from the core vocative usage ||

'ύώτορ

before

or only after *ί(ων was redone as ίάων.

4.6 The tracing of ίάων ultimately back to an acrostatic substantive *hjis-u-

can in this way directly or indirectly explain.the substantival value

of the form, its troublesome initial h-, and even the apparent incongruity between the likely antiquity and late epic debut of δώτορ ίάων

while

invoking only a nominal stem (cf. §§4.4, 4.5.1 above·, 5.1 below) that Greek is known to have had in any event. 158

158 Just as this last consideration disfavors seeing a *h,uesu- behind ίάων, it also recommends against other hypotheses that might in other respects have something in their favor. It might otherwise be attractive, for example, to posit a substantive

*h,es-eh2-

'good(s), possession(s)' whose Greek gen. pi. *ehahon could lie directly behind ίάων, (cf. §4.5.4.1 for the h-), and whose possessive derivative *h1esh2-oproprietor' (cf. *roteh2- 'wheel,' —> *roth2-o-

'the one with goods,

'chariot') would be reflected in Latin erus

'master'—if not also, perhaps as *hjSh2-0-, in Hitt. ishä- 'lord' (cf. Puhvel HED 385ff. and Tischler HEG 372ί(., both with many further references). But since the *h1es-eh2- that this presupposes is not otherwise demonstrable in Greek, and since adoption of this explanation of ίάων would inevitably mean dissociating it from a70s, the analysis of ίάων just presented here seems preferable.

5. An account of efjos

5.1 If there are reasons for deriving ίάων from a *h1es-u-, and if more particularly one of those reasons is that that stem is assured for Greek in any event by kv- and ηύ- < "7z,(e)s-tt-, there is all the more reason to start with the same stem in accounting for er)oy: That this epithet is at least descriptively nothing more than the gen. sg. of ίύς/ήύς is the practically unavoidable conclusion to be drawn from the observation that all three 159 of the masc. gen. substantives to which ίήος is applied also occur as nominatives or accusatives with ίνς/ήύς or evv/-ηΰν:ίω παιδόί ΐήος (A 393)

evs Trais (Β 819+)

V'LOS KI)os

υίόν ίύν

άνΰρός kfjos



138+)

(Τ 342+)

(Θ 303)

νΐον ...

ηύν

άνηρ ήύς

(Ζ 8)

(Γ 167+)

The ideal account, therefore, would make tfjos an actual gen. sg. of ew/i)w if possible. The first point to be made about the reconstruction of krjos, therefore, is that just as was the case with—and partly because of—ίάων (cf. §4.4), a *h1(e)s-uis a less problematical assumption than *hjues-u-. 5.2 While the reconstruction of an acrostatic genitive pi. substantive *hjesu-öm—though formulated on its own grounds altogether (§4.5.1)— turns out to allow for a reasonable account of ίάων, a precisely analogous reconstruction of the pre-form of irjos—namely *htesu-os (> *esu-os —> *e+ehu-os) would probably not be chosen for its own sake. This is because it is doubtful that an acrostatic paradigm—or the Greek outcome of one, at any rate—can be assumed without further ado to have had the adjectival value that ßapvs, Skt. guru- etc.) was already an innovation of the proto-language.

150

Two Studies in Greek and Homeric Linguistics

M o r e e l a b o r a t e l y , a n d a m o n g f o r m s m o r e i m m e d i a t e l y at i s s u e h e r e , 1 7 6 Celtic reflects n o t o n l y a *hjU0sit- in OIr. fo ' g o o d / 1 7 7 b u t also a in OIr. flu ' w o r t h y ; w o r t h / W . gwiw humored.'178

' w o r t h y / Br. gwiou

'good-

T h i s m a k e s t w o s e p a r a t e w-stem a d j e c t i v e s w i t h r o o t

v o c a l i s m s that w e r e both theoretically restricted to acrostatic, substantival, p a r a d i g m s . 1 7 9

The hypothesized acrostatic

and

substantive

176 For a discussion of these forms and problems from a different point of view cf. Bader ttudes 13ff. 177 Olr./d in the meaning 'lord/ if simply a substantivization of fo 'good, excellent, noble/ would obviously be reminiscent of avhpos ίήος 'nobleman' at ο 450. 178 OIr. α-stem gen. feibe/ dat. feibl nom. pi. feba ("uesua-) in the meaning 'excellence' can be analyzed in more than one way. It could theoretically be an abstract ^juesue-h}- to a *h,uesuo- 'good' that would stand to *htuesu- 'good' (> Ved. vasu- etc.) as does e.g. *h,lengwhuo- (> Lith. lengvas 'light') to *h1lngwku- (> ίλαχύί 'light'). On the other hand, 'good' could itself have made an abstract *h1uesu-h2- (type ϊθόί [*-u-h2 ] 'direction, course' to ιθϋί [*-u-] 'straight'), and if such an abstract inflected *-u-h2l *-u-eh2~, oblique *ft,uesu-eft2- could be the direct source of the fl-stem forms feibe etc. This second account may in fact be slightly preferable: In the first place *htuesuo- 'good' is not otherwise reflected (though cf. Luv. wasui- 'good' [Laroche DLL 110]). In addition, δ-stem feibe I feib means not only 'excellence' but also 'wealth, fortune' (DIL 295)—two meanings that are different enough to raise a question. This semantic problem finds a neat solution, however, under the assumption of a *hjuesu-h2/ *htuesu-eh2- for 'excellence:' Since the nom. of this derivative was completely homonymous with the collective neut. nom.-acc. pi. *hjuesu-h2 'goods, wealth' (> Ved. väsü, Pal. wasü) to the neut. sg. substantive *h,uisu 'a good, possession' (see what immediately follows in the main text and the next note), *hjuesu-h2 'wealth' (collective substantive) adopted in this scenario the inflection of 'excellence' (adjectival abstract)—or rather the two things would have become a single lexical item with both meanings. 179 In my view (which is only a modification of what can be found in Rix MSS 18 [1965], 79ff. plus Schindler Flexion und Wortbildung 262 and BSL 70 [1975], 4-8) there were two acrostatic types (see now also Rasmussen Morphophonemik 255 and Schindler In Honorem Holger Pedersen 398): A. nom.-acc.M)-S(z)

oblique R(ö)-$(z)-D(z)

locative sg, RM-$(e)

*iikw-r (ήπαρ)

*i0kw-n- (L. iocin[er]-)

*iekw-m (L. iec[ur])

%ig -i- (Arm. iz)

%0g -i- (οφ«)

B. nom.-acc. R(6)-S(z)

oblique Rfe)-SCz)-D(z)

mh

*nökw-t- (L. nox etc.) *gon-u (γόυν)

locative sg. R(z)-S(e) 'night'

*nekw-t- (Hitt. nekuz) *gen-u- (L. genu etc.)

'liver' 'snake'

wh

*gn-eu (cf. RV sno[h})

'knee'

Good for You: if/os (and έάων)

(nom.-acc. *hjuisu/ the aberrant

151

obi. *h1uosu-1 loc. *h2ueseu) that provides a source for

root vocalisms

of these Celtic u -stem adjectives

simultaneously provides a substantival pre-form for the inherited substantive seen not only in Ved. väsu / väsv- 'wealth' and Av. varjhuuam '(of) good things' as above (§5.2.1), but also in Palaic wasü 'good things' 180 a n d L u v i a n waSu 'bien,

salut!.'m

If, in turn, adjectival *h1uesu- (OIr. flu, W. gwiw) and *hjuosu- (Olr./o) 'good' may be explained essentially as outcomes of proterokinetic *hju(e)sii- that have analogically adopted features of an acrostatic substantival *hjuisu-/

*h1uosu- in the way described earlier, it becomes

possible to give the same explanation both to the *Ji,esw- that is called for, on the face of it, by Homeric ήύ- 'goodly, well favored, noble' 182 and also to *hjOsu- as the pre-form of Hitt. aSsu- 'good.' For ήύ-, an acrostatic neuter substantive *h1esu > *esu > *ehu 'good thing' that was reanalyzed at some stage as a substantivized neuter adjective would make a purely adjectival neuter ή(Ιι)ύ 'good' immediately possible—whence the usage eventually seen in repeated Homeric

Since both proterokinetic and amphikinetic paradigms had only e/z root ablaut, and hysterokinetic ones had ζ in the root throughout, e and ο root vocalism are both immediately indicative of an acrostatic pre-form, e pointing automatically to type A and ο ambiguous between the two acrostatic types in the absence of further information. 180 Watkins Investigationes Philologicae et Comparativae (Gedenkschrift H. Kronasser) 256,261. 181 Laroche DLL 110. 182 This hypothesis can provide an explicit and tenable basis for the approach taken by Beekes Development 287-9 and Ruijgh (apud Beekes Development 289 note 8). It is only that the neuter to which the lengthened grade of ήύ- is ultimately to be traced was a substantive in the scheme given here, and not in the first instance simply the adjectival neuter somehow belonging with masculine adjectival ίύ- < *h,sü-. The view taken more recently by Beekes (o-o-pe-ro-si

367 note 4) is therefore less attractive: If the notion

expressed there that "long vowels in Proto-Indo-European arose only in monosyllables and when a vowel in a final syllable stood before a resonant (except m)" were tenable, and if it really necessarily followed therefrom that ήύ(ς) is metrically lengthened (cf. Leumann Horn, Wörter 317 note 107), the same logic would presumably demand that Celtic *uehu- (Oh. flu, W. gwiw), formerly mentioned in this connection by Beekes himself (Development 289), must also be the result of metrical lengthening.

But since that is

impossible, and Celtic *uehu- consequently has a legitimate «-grade root, an equally legitimate *ehu- in Greek becomes an entirely open possibility too.

152

Two Studies in Greek and Homeric Linguistics

μίνος ήύ#

(Ρ 456+). And that could naturally lead in turn to masc. adjectival ήύί—as in the formula | ήύς re μίγας re# (Β 653+).183 A ^JOSM-, in addition to being a plausible pre-form for Hittite ässu 'a good' (§5.2.1 no. 4), may even be indirectly attested—by a derivative—in Greek: Hypothetical *hjosu-ro-, a possessive denominative derivative in -ro- that descriptively shows the oblique stem of the acrostatic basis *htSsu / *hjösu- 'good thing, blessing/ and which would mean something like 'bringing prosperity/ could provide an entirely satisfactory etymon for the otherwise mysterious ovpos 'fair, favorable wind' (Horn. +).184 And unless it is supposed for some reason that the o-grade root of this formation was introduced as part of the derivational process (*h1(e)su*hj0su-r0- ?), there is a good chance that the substantive basis of the derivative itself featured a stem allomorph \osu-. The point to be emphasized, however, is only that it is possible to explain the root vocalisms of Gk. ηύ- 'noble, well favored' (if < *h-,es-u~) and Hittite äSSu- 'good' (< *hjOs-u-) as transferred to these adjectives from the acrostatic neuter substantive from which they could have been derived (originally with proterokinetic apophonic characteristics) in the first place. 183 There is no telling, of course, whether masc. r/vs is anything more than a device of the epic poets (based on the neut. ηύ) intended to extend *(ύί re μ-iyos re one mora "backward" so that it would not violate Hermann's bridge. Cf. Beekes o-o-pe-ro-si 367 note 4. 184 Phonology: ovpos is descriptively vowel-initial as far as Homeric prosody is concerned—i.e. with virtually no convincing evidence (including δ 520: # α ψ δε Oeotj ονρον στρίψαν) for f - . But this does not mean much one way or the other in this case because of the frequent and well known versification of etymological /ro- and /τω- as *hj0su-r0'bringing prosperity' is a derivation comparable to cases like RV amhü 'distress' —» RV amhu-rä- 'distressed,' AV pämsü- 'dust' RV pämsu-rä- 'dusty,' or RV madhu 'sweetness' RV madhu-ΐά- 'sweet'—not to speak of äsu- 'life' —> asu-ra- 'lively, active.' For another etymology of ovpos altogether see Peters Untersuchungen 56, where an association with L. auere 'be eager' and Skt. avas- 'aid' is tentatively suggested. Since that root would seem to be *h2euH- (comparing also Skt. Uti- 'aid' etc.), ovpos 'favorable wind' would have to reflect */i2ou(H)-ro- in this analysis, with an unambiguous root ο grade, the morphological status of which is not quite clear in this -ro- formation. In the end, however, this alternative explanation of ovpos cannot be ruled out.

Good for You: (ijos (and ίάωυ)

153

5.2.3 Returning to ίήος with all this in mind, it would now appear to be evident that this epithet, which is strictly and unambiguously an adjective, could reflect *e+ehu-os (< *ht(e)su-os with e- eventually transferred from nom.-acc. *ehii-) only on condition that it is plausible to suppose that Greek, like Indo-Iranian, effaced the putatively inherited opposition between acrostatic substantive and derived proterokinetic possessive adjective (§5.2.1) partly and/or sometimes by transferring acrostatic-looking oblique R-u- from substantive to adjective. The indications are, however, that in Greek the breakdown of the original derivational pattern saw only the introduction of acrostatic root vocalisms from the substantives into the proterokinetic adjectival paradigms: ·πο\ύand ηύ- have just been mentioned (§5.2.2.2) in this connection, to which may possibly be added the more complex (R-S1-S2-) οξύ- 'sharp,' as if < *fa2ok-s-u-.185 But neither -u-/-u- inflection nor the -u-/-uu- that has sometimes replaced it in Greek ever seem to have been generalized as replacements of -u-/-eu- in inherited proterokinetic paradigms—which very conspicuously include the u-stem adjectives: 5.2.3.1 Far from being generalized beyond the items in which it was inherited, the -u-/-u- that reflects acrostatic inflection is only rarely preserved even where original in Greek. Some inherited aerostatics do prove to be conservative: yowl yovvos 'knee' < *gonu-/ *gonu-, for example, or vaüs/ νηόί 'ship' < *neh2u-/ *neh2u-.186 Others, however, have replaced -u- /-u- inflection in one of two ways: 1) The neuter μίθυ 'wine,' to give just one example, whose pre-form may well have inflected acrostatically (cf. Ved. madhu/ madhv- and §5.2.2.1) has remade expectable *mithu/ *methu- to -u-/-uu- (μίθυ, -vos). 2) The neuter 7τώυ 'flock,' on the other hand, may be reasonably seen as the reflex of a ^oh^u (root *peh2- of Hitt. pah-S- 'protect' etc.). That this allomorph originally belonged to an acrostatic paradigm is suggested already by its root o-grade. But it has become proterokinetic in Greek by 185 The possibility of an o-grade root here—and the problems that this raises—have been made explicit relatively recently by Narten o-o-pe-ro-si 213f.

186 Cf. Peters Untersuchungen 253 note 210. The contention of Lamberterie Adjectifs 788 that -U-/-U- inflection is simply not as old as -u-/-eu- inflection takes account of neither the functional nor the apophonic nor the accentual contrasts that are to be observed in cases, among others, like I-Ir. *krätu-/ *krätu- 'power' vs. *krtii-/ *krteu- (> κρατΰΊ κρατεί/:)-) 'strong' in Greek. See, in any case, Hoffmann Aufsätze 2, 597ff. with further references (as in note 109 above) and §5.2.1.

154

Two Studies in Greek and Homeric

Linguistics

the time of its earliest attestation: Horn., Hes. pi. ττώία points to ττώυ / πάκο? etc. 5.2.3.2 Among the items, conversely, that are likely to have been inherited with proterokinetic paradigms, there seems to be no clear case of what would descriptively amount to a replacement of -u-/-eu- by -u-/-u-. This applies even to substantives, where Greek u-stems appear to have kept proterokinetic inflection wherever there is comparative evidence to suggest that that was the original pattern: e.g. αστυ, -eos/ -ecoy 'town': RV västu, -oh; πέΚίκυς, -eo?/-€ooj 'axe': RV parasü-, -oh;1S7 πήγυς, o? 'forearm': RV bähü-, nom. pi. bähdvah and Av. gen. sg. bäzaus. But if the object of the exercise is to estimate the plausibility of *e+ehu-os (§5.2.3) as the pre-form of efjos, then the crucial question is whether it is safe to assume that Greek, like Vedic and Avestan (§5.2.2.1), ever occasionally introduced oblique -u- into the originally proterokinetic (-ü-/-iu-) paradigm of an inherited u-stem adjective. And as already intimated (§5.2.3), this does not seem to have happened. The category established by familiar examples like βαρύ-, βαριοί: Skt. gurü-, guröh 'heavy;' ίΰρύ-, evptos : urü-, uroh 'wide;' παχύ-, παχέος : bahü-, bahoh 'thick' etc. contains over forty adjectives in Greek,188 many clearly inherited, that make it quite clear that the original -u-/-eu- inflection in this category was consistently preserved. 189 The assumption of *h1(e)s-u-/ *h1(e)s-u{—> *e+ehu-os) as the ultimate starting point for ίηος 'goodly, noble' is consequently a weak and risky one. 5.2.4 What this implies for an interpretation of er/o? is clear. It now seems best assumed not only that erjos is a genitive of em (§5.1), but also— since it is strictly adjectival—that it somehow more particularly represents the usual Greek treatment of an inherited proterokinetic genitive. The *hjSti-/ *h]Seu- 'good' thus implied would result regularly in *ehü-s / *eheu-os etc. And if masc. gen. *ehiu-os is ultimately represented by ef)o -κAeos

(Brugmann IF 9 [1898], 158ff.; Meister Η Κ 153f„ 159) and a simple analogy (after contraction of the nom. sg. from -K\C[/:]J7S to -κλης) like -yevqs : -yeyeor = -κλτ/? : X. 196 Werner η und a 30ff„ 41ff. 197 Werner ij und a as in the previous note. Also see note 125 above. 198 Meister HK 173f., Chantraine GH 95, Risch Wortbildung 157, Wyatt Lengthening 26 with notes 10 and 11 and further references—especially to Witte.

Good for You: ifjos (and (ίων)

II ήνιόχοιο#

|| (άνδρο)φόνοιο #

.i

|| Άντιφάταο#

ι

|| ηνιοχηα# Ι

157

.i

Ι (πατρο)φονήα#

|| Άντιφατήαϋ

ήνιοχή($#

But as soon as localized formations like || ήι>ίοχηα# and || ττατροφουηαϋ had been created, they were at the same time automatically classifiable as lengthened ( — w - ~ ) line-end versions of shorter ( - " " - ) forms that were at home elsewhere in the verse:199 #



j and —

# Αιθιόπων]

(Ψ 206)

|| || ΑΙΘιοπήαςΧ

(A 4 2 3 )

Ι Άντιφατήα#

(κ 1 1 4 )

#Αΐ0ιοτταίί (α 2 3 + ) Atöt'o7ras| (α 2 2 ) Αιθιόπων]

(e 2 8 2 )

# Άντιφάτην]

(Μ 191)

# Άντιφάτης]

(ο 2 4 3 )

Άντιφάτην

I (ο 2 4 2 )

# ηνίοχοι]

(Σ 2 2 5 )

#ήνιόχους]

(Λ 161)

# ήνιόχω]

Ι ήνιοχηα# || ήνιοχηα#

(Θ 312+) (Ε

505)

(Λ 4 7 = Μ 8 4 )

ήνίοχον]

(Θ 1 2 6 + )

ηνίοχος]

(Λ 2 8 0 + )

άνδροφόνονI

(α 2 6 1 )

πατροφόνος]

([I 4 6 1 ] )

φασσοφόνψ]

(Ο 238)

|| πατροφονηα#

(α 2 9 9 + )

What it is most essential to stress here, however, is that from the point of view of the position and inflection of line segments whose content is an NP, there is no distinction to be drawn between the pairs given above and:

199 O'Neill YCS 8 (1942), 144 (Table 15).

158

Two Studies in Greek and Homeric

# υίον

ίύν\

(Θ 303)

IviosefjosU

(O

Linguistics

138+)

5.3.2 If erjos is ultimately an analogical, stretched, line-end-localized form like πατροφονήα etc. that was created in this way in order to give Ι --εμοΐο# and || -~eoio# a correspondent that was usable in secondperson contexts, a trivial consequence is that avbpos ei/ofj' (o 450) must be considered a "moved" adonic; 200 but a more important consequence is that although erjos is based on f i r 'goodly, noble/ it does not even indirectly represent an actual genitive *eeos of that adjective. This is an advantage, since it entirely does away with the embarrassing need to assume such a genitive, which would have to have existed in relatively recent Ionic (§5.2.4.2: 1, 2), but in spite of that—like every other form of this adjective other than the nominative and accusative sg. 201 —never manages to be attested as such. The explanation of erjos as strictly an artificial epicism allows ev(s)/ ήύ(ς) to be in fact the defective (nom. and acc. sg. masc. and neut. only 202 ) adjective that it so clearly seems to be. 5.4 As, finally, to the initial aspiration with which ίήος is heavily transmitted in the manuscript tradition, there seems no better explanation available than a version of the traditional one: 203 namely that expressions of the type || --er)os# systematically alternated with || --eolott, and— partly for this very reason, and partly because of Zenodotus' theories (§§2.2.4, 3.4)—practically all the instances of || )o?# acquired an actual variant that showed iolo# instead. The opportunities for contamination

200 Cf. §2.3.3 above. 201 The neut. pi. Hesychian gloss ηία· αγαθά (so Latte HAL without comment) need only show—as is clear anyway—that there existed an adjectival singular ήϋ to which this plural could have been made. If, however, ψα is read (so Schmidt HAL, but with the comment "Malim ή{α, rji'a..."), and if this reading is correct, it could theoretically point, of course, to a neut. nom.-acc. sg. *ηυ, which is potentially analysable as a trace of an acrostatic substantive *h1is-u 'good(s)/ as proposed in §5.2.2.2. And if such a substantival *ηυ has innovated in making a plural ije(/)a (with full-grade suffix), it would only be a second example of what is already seen in -πών : ποκα (§5.2.3. lb). See also Lamberterie Adjectifs 790ff. 2 0 2 Cf. μέγαί, μέγα2 0 3 E.g. DELG 388.

Good for You: tijos (and ίάων)

159

under these circumstances—especially since evjo? in this scenario existed only in epic language—would obviously have been extensive, and one result was that efjoy (which is, after all, itself found as a variant) was frequently turned into erjos.204

204 Conversely, even so obviously wrong a form as lolo exists as a variant too (e.g. at Ω 422). Still more dramatic is the creation of both (ήο and dolos as contaminated variants by essentially the same process (§3.5.2.a.l). The success of contaminated CTJOS may perhaps be explained by reasoning that if—as is implicit in the assumption of a contamination in the first place—there was eventually a tendency in the written transmission to give eήος and eolo the same initial, it is not surprising if artificial—and eventually arcane—eijos was assimilated to much more familiar eolo rather than the reverse.

Bibliographical Abbreviations

Allen, T.W. (ed.): Homeri Ilias 1-3 (Oxford 1931). —: Homeri Opera 3-4: Odyssea (Oxford 1958-1962). Bader, F. Etudes: Etudes de composition nominale en mycinien, I: Les prefixes milioratifs du grec (Rome 1969). Bechtel, F. GD: Die griechischen Dialekte 1-3 (Berlin 1921-1924). — Lex.: F.B., Lexilogus zu Homer (Halle 1914). — Vokalkontraction: F.B., Die Vocalcontraction bei Homer (Halle 1908). Beekes, R.S.P. Development: The Development of the Proto-Indo-European Laryngeals in Greek (The Hague-Paris 1969). BVKSG: Berichte über die Verhandlungen der Königlich Sächsischen Gesellschaft zu Leipzig. Philologisch-historische Klasse (Leipzig 1846-1917). Blümel, W. Aiolisch: Die aiolischen Dialekte (Göttingen 1982). Bowie, A.M. Poetic Dialect: The Poetic Dialect of Sappho and Alcaeus (Salem, N.H. 1981). Brugmann, K. GVGIS: Grundriss der vergleichenden Grammatik der indogermanischen Sprachen 1-2 (Strassburg 1897-1916). — Problem: Ein Problem der Homerischen Textkritik und der vergleichenden Sprachwissenschaft (Leipzig 1876). Buck, C.D. GD: The Greek Dialects (Chicago-London 1955). Chantraine, P. GH: Grammaire homerique l 3 -2 (Paris 1958-1963). Concordantia: G. Maloney and W. Frohn, Concordantia in corpus Hippocraticum (New York, 1986). Delbrück, B. SF: Syntaktische Forschungen 1-5 (Halle 1871-89). DELG: P. Chantraine et al, Dictionnaire itymologique de la langue grecque. Histoire des mots (Paris, 1968-80). DELL: A. Ernout and A. Meillet, Dictionnaire itymologique de la langue latine. Histoire des mots (Paris 19594). DGEEP: E. Schwyzer (ed.), Dialectorum Graecarum exempla epigraphica potiora (Leipzig 1923, repr. Hildesheim 1960). DIL: Dictionary of the Irish Language (Dublin 1983).

Bibliographical

Abbreviations

161

Dubois, L. Recherches: Recherches sur le dialecte arcadien 1-3 (Louvain-laNeuve 1986). Erbse, Η. (ed.) Scholia: Scholia Graeca in Homeri lliadem 1-6 (Berlin 19691983). Festschrift Vasmer: M. Woltner and H. Bräuer (eds.), Festschrift für Max Vasmer zum 70. Geburtstag am 28. Februar 1956 (Wiesbaden 1956). Flexion und Wortbildung: Η. Rix (ed.), Akten der V. Fachtagung der Indogermanischen Gesellschaft (Wiesbaden 1975). Garcia-Ramon, J. Origines: Les origines postmyceniennes du groupe dialectal eolien: Stüde linguistique (Salamanca 1975). GEW: H. Frisk, Griechisches etymologisches Wörterbuch 1-2 (Heidelberg, 1960-72). Gildersleeve, B.L. Syntax: Syntax of Classical Greek 1-2 (New YorkCincinnati-Chicago 1900-1911). Gunnerson, W. u-Stems: History ofu-Stems in Greek. University of Chicago Dissertation (Chicago 1905). Hainsworth, J.B. Flexibility: The Flexibility of the Homeric Formula (Oxford 1968). Hansen, P. LGVI: A List of Greek Verse Inscriptions down to 400 B.C.: An analytical survey (Copenhagen 1975). Hoffmann, K. Aufsätze: Aufsätze zur Indo-Iranistik 1-2 (Wiesbaden 19751976). Hoffmann, O. GO: Die griechischen Dialekte 1-3 (Göttingen 1891-1898). JEG: M.L. West (ed.), Iambi et elegi Graeci ante Alexandrum cantati2 1-2 (Oxford 1989-1992). IF: Indogermanische Forschungen (Strassburg, Berlin 1891-). Insler, S. Gathas: The Gathas of Zarathustra (Leiden 1975). Kellens, J. Verbe: Le verbe avestique (Wiesbaden 1984). Kock,T. (ed.) CAF: Comicorum Atticorum Fragmenta 1-3 (Leipzig 18801888). Landfester, M. Philos: Das griechische Nomen φίλος und seine Ableitungen (Hildesheim 1966). Laroche, E. DLL: Dictionnaire de la langue louvite (Paris 1959). La Roche, I. (ed.): Homeri Ilias 1-2 (Leipzig 1873-1876). Latte, K. (ed.) HAL: Hesychii Alexandrini Lexicon (Copenhagen 1953-). Leaf, W (ed.) Iliad: The Iliad2 1-2 (London 1900-1902). Lejeune, M. Adv. -ΘΕΝ: Les adverbes grecs en -6tv (Bordeaux 1939).

162

Two Studies in Greek and Homeric Linguistics

— Phonitique: Phonetique historique du mycenien et du grec ancien (Paris 1972). Leumann, M. Horn. Wörter: Homerische Wörter (Basel 1950). LEW: Α. Walde, Lateinisches etymologisches Wörterbuch. 3. Auflage von J. B. Hofmann (Heidelberg 1938-1954). LFGE: B. Snell et al. (eds.), Lexicon des Frühgriechischen Epos (Göttingen 1979-). Ludwich, A. (ed.): Homeri Carmina 1-2 (Leipzig 1889-1907). Macdonell, A.A. VG: Vedic Grammar (Strassburg 1910). Mayrhofer, M. EWAia: Etymologisches Wörterbuch des Altindoarischen (Heidelberg 1986-1997). — KEWAi: Kurzgefasstes etymologisches Wörterbuch des Altindischen (Heidelberg 1953-1980). Meister, Κ. HK: Die homerische Kunstsprache (Leipzig 1921). Mendez Dosuna, J. Noroeste: Los dialectos dorios del noroeste. Gramatica y estudio dialectal (Salamanca 1985). Miller, D.G. Ionian: Homer and the Ionian Epic Tradition (Innsbruck 1982). ΜΝΗΜΗΣ ΧΑΡΙΝ: Η. Kronasser (ed.), Gedenkschrift Paul Kretschmer 1-2 (Vienna 1956-1957). Monro, D. GHD: A Grammar of the Homeric Dialect (Oxford 18912). MSS: Münchener Studien zur Sprachwissenschaft (Munich 1951-). Narten, J. Aoriste: Die sigmatischen Aoriste im Veda (Wiesbaden 1964). Oettinger, N. Stammbildung: Die Stammbildung des hethitischen Verbums (Nuremberg 1979). o-o-pe-ro-si: A. Etter (ed.), Festschrift für Ernst Risch zum 75. Geburtstag (Berlin-New York 1986). Owls to Athens: E.M. Craik (ed.), Essays on classical subjects presented to Sir Kenneth Dover (Oxford-New York 1990). Peters, M. Untersuchungen: Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen (Vienna 1980). PLF: E. Lobel and D. Page (eds.), Poetarum Lesbiorum Fragmenta (Oxford 1955). PMG: D.L. Page (ed.), Poetae Melici Graeci (Oxford 1962). Powell, J. (ed.) Collectanea: Collectanea Alexandrina (Oxford 1925). Puhvel, J. HED: Hittite Etymological Dictionary 1-3 (Berlin- New YorkAmsterdam 1984-1991). REG: Revue des etudes grecques (Paris 1888-).

Bibliographical

Abbreviations

163

REIE: Revue des itudes indo-indoeuropeennes (Bucharest 1938). Risch, E. Wortbildung: Wortbildung der homerischen Sprache. Zweite, völlig überarbeitete Auflage (Berlin- New York 1974). Rix, H. Grammatik: Historische Grammatik des Griechischen (Darmstadt 1976). RPh: Revue de philologie, de litterature et d' histoire anciennes (Paris 3. serie 1927-). Schmidt, J. Pluralbildungen: Die Pluralbildungen der indogermanischen Neutra (Weimar 1889). Schmitt, R. Dichtung: Dichtung und Dichtersprache in indogermanischer Zeit (Wiesbaden 1967). Schulze, W. QE: Quaestiones Epicae (Gütersloh 1892). Schwyzer, E. GG: Griechische Grammatik 1-2 (Munich 1939-1950). SGDI: H.Collitz et al. (eds.), Sammlung der griechischen Dialekt-Inschriften 14 (Göttingen 1884-1915). Shipp, G. Studies: Studies in the Language of Homer (Cambridge 1972 2 ). SILTA: Studi Italiani di Linguistica teorica ed applicata (Padua 1972-). Solmsen, F. Untersuchungen: Untersuchungen zur griechischen Laut- und Verslehre (Strassburg 1901). Sommer, F. Lautstudien: Griechische Lautstudien (Strassburg 1905). SPAW: Sitzungsberichte der Preussischen Akademie der Wissenschaften. Philosophisch-historische

Klasse (Berlin 1922-1938).

Stork, P. Index: Index of Verb Forms in Herodotus (Groningen 1987). Thumb, A.-Scherer, A. HGD: Handbuch der griechischen Dialekte. 2. Auflage, 2. Teil (Heidelberg 1959). Tischler, J. HEG: Hethitisches etymologisches Glossar 1-10 (Innsbruck 19771994). Verdier, C. Eolismes: Les Solismes non-ipiques de la langue de Pindare (Innsbruck 1972). Voigt, E-M. (ed.) SA: Sappho et Alcaeus (Amsterdam 1971). Wackernagel, J. and A. Debrunner AiGr.: Altindische Grammatik 1-3 and Nachträge (Göttingen 1896-1957). Wackernagel, J. SU: Sprachliche Untersuchungen zu Homer (Göttingen 1916). — Vorlesungen:

Vorlesungen über Syntax2 1-2 (Basel 1926).

Wathelet, P. Traits: Les traits ioliens dans la langue de Γ ipopee grecque (Rome 1970).

164

Two Studies in Greek and Homeric Linguistics

Werner, R. η und et: η und ei vor Vokal bei Homer. Zürich Dissertation (Freiburg i. d. Schweiz 1948). Wyatt, W. Prothetik The Greek Prothetic Vowel (Cleveland 1972). — Lengthening: Metrical Lengthening in Homer (Rome 1969). YCS: Yale Classical Studies (London [etc.] 1928-).

Index of Words and Forms

Mycenaean Greek a-mo, a-mo-ta 138136. we-a2-no 139146, 141146.

Alphabetic Greek άασί-φρων 29. άάω: Lac. ά/:αταται, Horn, αάται etc., δασα-, άάσθη- 27iL, 70 230 ; Cret. αταμίνοί, Hsch. άγατάσθαι· βλάπτεσθαι and αγάτημαι- βέβλαμμαι 2893. αγαμαι: aor. άγασ(σ)α- , fut. αγασσο-1414,15,18,19, 27, 1942

34114.

äyye'AAco: Att. fut. άγγΐλώ 1627. άγίνω: αγίνεσκον (beside άγινίω)

άρκίω: aor. άρκεσα-, fut. άρκεσο1736. αρμα 138,139. άρόω 'plow': άροσα- 36. &τη/άνάτα: 28f. άφίημι 13. βαίνω: aor. subj. /3ε ιω (Ζ 113) 43, 56. βάλλω: fut. βαλΐο-1521. βαρύ-, ßapios 154. βασιλεύς: βασιλέων 47. βοάω\ βοά α, βοόωσιν etc. 70229,230 γαμίω: fut. γαμέσσεται (I 394)/ γαμεϊται 1841. γίλάω 12: aor. γίλασ(σ)α-, fut.

>>ελασο-14,16, 27, 34» 4 , 35. yevos: γίνεσι 141 146 .

ö/toj? (Argive) 138.

Γεστρα (= /τεστρα)· ει>δυσ (φ 2 6 0 ) , εΐώσι ( Τ 139); i n f i n . έάαί> (0 5 0 9 ) 53ff.,

56ff. Are. earoi,

Boe.

4 5 , 69. T h e s s . earov

45

63.

Att. eip^ai,

ίρζουτι

H e r a c l . αφ-,

1 4 0 1 4 6 ; Ιίργω/

έργω

,69.

e t i / e e t s 144.

I m p f . ίϊων/ίωι> etc. 1 2 , 3 9 f f . ,

e'uos (Γ 2 9 1 + ) 43.

4 1 1 4 4 , 4 6 f f . , 50f., 55f. Iterative

e\avva>: aor. έ λ α σ ( σ ) α - 1 4 , 2 7 ,

i m p f . Ίασκου

64ff., ίϊασκον

70ff.

I m p e r a t i v e εΰ'α·... ea l l f .

Ιφ143.

εί'ρω: f u t . e p e o - 1 5 2 1 .

taovrvs 161

139146,14114