Introduction
This chapter discusses and evaluates the evidence for the regular outcome of *r̥ in the alphabetic Greek dialects other than Ionic-Attic.1
The first aim is to determine under which conditions and in which dialects o-colored reflexes are regular. There is currently no consensus on this matter. The case for a-colored reflexes in dialects like Arcadian and Cypriot has been overstated: Morpurgo Davies (1968), though an influential treatment, suffers from a lack of clarity about Pan-Greek developments that yielded âαÏâ at an early date, such as those involving laryngeals. This issue has been clarified by previous scholars (GarcÃa Ramón 1985, Haug 2002) and will not be discussed in all its details here.
The second main aim is to determine, for each dialect, the regular place of the anaptyctic vowel. Surprisingly, only few previous discussions have paid attention to this issue: the main focus is usually on determining the vowel color of the regular reflex. There is a broadly-shared presupposition that the apparent hesitation between âαÏâ and âÏαâ in Ionic-Attic was also characteristic of other Greek dialects. In fact, some scholars suppose that in most lexemes the place of the anaptyctic shwa was fixed already in Proto-Greek; this shwa would have merged with /a/ or /o/ later, depending on the dialect and in some cases on further phonetic conditioning factors. However, in the previous chapter we encountered examples where the dialects have a diverging vowel slot in the same etymon, e.g. Hom. Ïá½³ÏÏαÏοÏ, Thess. ÏεÏÏοÏÎ¿Ï âfourthâ as opposed to Ion.-Att. Ïá½³ÏαÏÏοÏ, Arc. ÏεÏοÏÏοÏ. Moreover, we have seen that the regular Mycenaean reflex of *rÌ¥ was either âorâ or preserved ârÌ¥â, and that there is no need to assume a separate outcome *rÌ¥ > âroâ for that dialect. The evidence for the vowel slot in the other dialect groups (West Greek dialects, Aeolic, Arcado-Cyprian) must be reconsidered in this light.
3.1 The Alleged Cretan Liquid Metathesis
The West Greek reflex of *rÌ¥ is normally assumed to be identical to that of Ionic-Attic: âÏαâ is regular, âαÏâ arose by analogy with related forms. On Crete, however, we find a number of forms with âαÏâ for which an analogical explanation is difficult to find. They are the following (for places of attestation, see Bile 1988):2
-
δαÏκμα âdrachmâ (Knossos), also δαÏκνα (Ion.-Att. δÏαÏμή);
-
καÏÏεÏÎ¿Ï âprevalentâ (cf. Ion.-Att. καÏÏεÏá½¹Ï âstrongâ, Hom. κÏαÏεÏá½¹Ï) and the following related words:
-
καÏÏÎ¿Ï âforceâ (cf. Ion.-Att. κÏá½±ÏοÏ);
-
PNâ¯s with âκαÏÏÎ·Ï and ÎαÏÏαιâ (cf. Ion.-Att. âκÏαÏηÏ, ÎÏαÏαιâ);
-
καÏÏαιÏοδ- âcattleâ (cf. Pi. κÏαÏαίÏοδâ);
-
ÏÏοÏεÏαÏÏον adv. âon the fourth day beforeâ (Lex Gortyn XI.53);
-
ÏÏαÏÏÎ¿Ï âband; clanâ, also in proper names (cf. Ion.-Att. ÏÏÏαÏá½¹Ï âarmy; campâ).
Since Hirt (1901: 232â238), most scholars have accepted that âÏαâ, âÏοâ was metathesized to âαÏâ, âοÏâ in Cretan.3 At first sight, this claim seems reasonable because of the Cretan forms ÏοÏÏι âtowards, againstâ (cf. Ion.-Att. ÏÏá½¹Ï, Hom. ÏÏοÏá½·) and ÎÏοÏδιÏα (Ion.-Att. á¼ÏÏοδίÏη).4 If metathesis must be assumed for these forms anyway, Hirtâs reasoning goes, it follows that âαÏâ < *rÌ¥ may have developed through âÏαâ. The argument presupposes, however, that the o-vocalism of ÏÏοÏá½· and á¼ÏÏοδίÏη is old and did not develop from *rÌ¥. As I will show below, this is not evident at all.
The major problem with the assumption of liquid metathesis in Cretan is the existence of forms with âÏαâ and âÏοâ, such as the following:
-
pres. inf. αÏοÏÏαÏεν âto run awayâ (Olous, 3rd c., cf. Class. ÏÏá½³ÏÏ âto runâ);
-
pres. opt. ÏÏαÏοι̣ (Eleutherna, 6th c.), inf. ÏÏαÏεν (Lex Gortyn III.49), impv. 3pl. ÏÏαÏον̣ÏÌ£Ïν (3rd c.), all from ÏÏá½±ÏÏ âto feedâ (cf. Class. ÏÏá½³ÏÏ âid.â);
-
pres. γÏαÏÏ âto writeâ (= Class. γÏá½±ÏÏ âid.â);
-
κÏÎ¿Î½Î¿Ï âtimeâ (Class. ÏÏá½¹Î½Î¿Ï âid.â);
-
ÏεÏÏαâ âfourâ in compounds such as ÏεÏÏαÏοδ- âcattleâ (IC IV 41, III 8â9);
-
ÏεÏÏαδ- âfourth dayâ (= Class. ÏεÏÏá½±Ï);
-
δÏÎ¿Î¼Î¿Ï âcourse; race trackâ (= Class. δÏόμοÏ) and δÏÎ¿Î¼ÎµÏ Ï âyoung adultâ.
Hirt (1901: 235) discusses some of these examples. He notes that ÏÏαÏÏ may have been influenced by the full grade ÏÏεÏâ, and that γÏαÏÏ, as a technical term, does not carry much weight. Furthermore, he does away with κÏÎ¿Î½Î¿Ï and δÏÎ¿Î¼Î¿Ï with the remark that liquid metathesis never operates on a fully regular basis, and makes the ad hoc suggestion that they were borrowed from another dialect. This is clearly unsatisfactory. Bile (1988: 125) proposed that âÏαâ and âÏοâ were metathesized in open syllables, but preserved as such in closed syllables. This idea is contradicted by the forms just listed, as well as by δαÏκμα / δαÏκνα.
Thus, there is no satisfactory explanation of the fluctuations.5 It is true that the present stems ÏÏαÏÏ and ÏÏαÏÏ can be explained as secondary (see below), and that γÏαÏÏ is a problematic form on any account.6 However, the forms κÏονοÏ, ÏεÏÏαÏοδâ, ÏεÏÏαδâ, and δÏÎ¿Î¼Î¿Ï cannot all be done away with as mere âexceptionsâ (Bile 1988: 125); they strongly militate against the idea of liquid metathesis. Moreover, if some form of liquid metathesis was operative, one would expect to also find examples of Ïε, Ïι, ÏÏ appearing as εÏ, ιÏ, Ï Ï in Cretan. There is no apparent reason why the metathesis would have been restricted to back vowels.
In view of these problems, let us now investigate whether âαÏâ and âοÏâ in Cretan can be viewed as the regular outcomes of *rÌ¥.
3.1.1 Cretan âαÏâ < *rÌ¥: Evidence and Counterevidence
Positing a regular Cretan development *rÌ¥ > âαÏâ immediately explains the following forms: καÏÏεÏÎ¿Ï < *krÌ¥teróâ, ÏÏαÏÏÎ¿Ï < *strÌ¥tóâ, and ÏÏοÏεÏαÏÏον < *âkwetrÌ¥toâ.7 The retention of δÏÎ¿Î¼Î¿Ï âtrackâ and κÏÎ¿Î½Î¿Ï âtimeâ, forms with an original o-vowel for which an analogical explanation seems out of reach, also finds a natural explanation. It remains to account for the Cretan forms with âÏαâ.
The reflex seen in ÏεÏÏα-Ïοδα âcattleâ and other compounds seems to contradict that of the ordinal Ïá½³ÏαÏÏοÏ. However, just as in Ionic-Attic, the linking vowel of ÏεÏÏαâ may have been taken over from the higher numerals á¼ÏÏαâ, á¼Î½Î½ÎµÎ±â, δεκαâ (see section 2.7). The collective numeral ÏεÏÏαδâ âquartetâ (which also exists in Ionic-Attic) has a suffix âάδâ that originated in δεκάδâ âgroup of tenâ and derives from a pre-form with syllabic nasal, PIE *deḱmÌ¥-tâ.8
The remaining counterevidence concerns the so-called âDoric presentsâ of the type ÏÏá½±ÏÏ, corresponding to class. ÏÏá½³ÏÏ. There are four such verbs in Cretan (see Bile 1988: 124). A zero grade thematic present stem is directly attested in forms of ÏÏαÏÏ and ÏÏαÏÏ (see above). In addition, the formation seems reflected in the PN ΣÏÏαÏÎ¹Î¼ÎµÎ½Î·Ï (Pylorus, 2nd c.; cf. class. ÏÏÏá½³ÏÏ) and the future [ε]ÏιÏÏαÏÎ¹Ï (Lyttos; cf. Class. ÏÏá½³ÏÏ âwill turnâ).9 The corresponding Ionic-Attic verbs have an e-grade present stem and sigmatic aorist, as opposed to a-vocalism in the η-aorist. A possible scenario would be reconstruct a zero grade root for the thematic root present, i.e. pres. *dhr̥ǵh-e/oâ, beside an e-grade root in the aor. *dhreǵh-sâ. If so, Cretan and other West Greek dialects then generalized the a-colored reflex throughout the verbal paradigm, while Ionic-Attic extended the e-grade root to the present stem.10
In sum, the Cretan evidence for âÏαâ < *rÌ¥ is easily reconciled with a regular development to âαÏâ in that dialect. It is unnecessary to assume that âÏαâ underwent liquid metathesis on an irregular basis. In order to further strengthen this conclusion, let us now consider the three forms with âοÏâ for which liquid metathesis has been assumed.
3.1.2 Cretan âοÏâ < *rÌ¥ after a Labial Consonant
Must Cretan ÏοÏÏι âtowards, againstâ and ÎÏοÏδιÏα really have developed by metathesis from the forms ÏÏοÏá½· and á¼ÏÏοδίÏá¾± as attested elsewhere, or might they directly reflect forms with *rÌ¥? As we have just seen, Cretan δÏÎ¿Î¼Î¿Ï (δÏÎ¿Î¼ÎµÏ Ï) and κÏÎ¿Î½Î¿Ï never contained *rÌ¥, and therefore speak against the assumption of metathesis. Therefore, even if only a plausible case can be made that ÏοÏÏι and ÎÏοÏδιÏα may have a pre-form with *rÌ¥, it is attractive to think that *rÌ¥ became Cretan âοÏâ after labial consonants, but âαÏâ in all other positions.
The evidence for the alleged pre-form PGr. *proti consists of Ion.-Att. and Lesb. ÏÏá½¹Ï, Hom. ÏÏοÏá½·, and Central Cretan ÏοÏÏι.11 On the other hand, PGr. *poti is reflected in Thessalian and Boeotian, perhaps in Arc. ÏÎ¿Ï and Myc. po-si,12 as well as all in West Greek dialects other than Central Cretan. Wyatt suggested that Ion.-Att. ÏÏá½¹Ï might reflect the prevocalic outcome of *poti contaminated with the ârâ of ÏÏá½¹. In Wyattâs view Homeric ÏÏοÏá½· arose in the same way; he also shows that ÏοÏÏι only occurs in Central Cretan, whereas the rest of Crete (like West Greek generally) has ÏοÏι. He accounts for Cretan ÏοÏÏι by assuming that it represents a contamination of ÏοÏι with ÏεÏá½· (Wyatt 1978: 121 n. 78), and concludes that Proto-Greek had only *poti.
At first sight, Wyattâs scenario offers an attractive reduction of the West Greek situation. However, it is unlikely that Proto-Greek had only *poti because in Homer ÏÏοÏá½· cannot be secondary beside ÏοÏá½·. A fuller treatment of the Homeric evidence (including muta cum liquida scansions in ÏÏá½¹Ï and related forms) will be given in section 7.2.5. Anticipating the conclusions to be reached there, the Homeric evidence favors a reconstruction *prÌ¥ti, and the precursors of ÏÏá½¹ÏÏ âforwardâ and ÏÏá½¹ÏÏÏον âfaceâ also continue a form starting with *prÌ¥ti̯â. This means that Proto-Greek had both *poti and *prÌ¥ti,13 and that Central Cretan ÏοÏÏι can be a retained archaism. The reconstruction *prÌ¥ti also accounts for the scansion of ÏÏá½¹Ï and related forms in Homer, while the form ÏÏá½¹Ï in Ionic and Attic prose may be ascribed to the influence of another local adverb, either ÏÏá½¹ or *poti, on the vocalization of *prÌ¥ti. Furthermore, a zero grade *prÌ¥ti could also account for a related adverbial element in Hittite, parza â-wardsâ.14
The assumption that Aphroditeâs name contained *rÌ¥ may come as a surprise, as it has no established etymology. However, positing a pre-form with *rÌ¥ would be the only reasonable way of accounting for the structural muta cum liquida scansion of á¼ÏÏοδίÏη in Epic Greek, since that metrical license is extremely rare in word-internal position, especially when plosive plus liquid does not follow a synchronic morpheme boundary.15
A third potential instance of a Cretan development âοÏâ < *rÌ¥ after labials are the proper names in âμοÏÏοÏ, which are especially frequent on Crete, but also appear in Lesbian and two West Greek dialects (Theran, Aetolian). Nothing crucially depends on including or excluding this example, but the evidence is suggestive. The simplex is attested only in post-classical sources:
-
μόÏÏοÏÎ á¼Î½Î¸ÏÏÏοÏ. θνηÏá½¹Ï. μέλαÏ, ÏαιόÏ. οἱ δὲ μοÏÏόν ÏαÏι âhuman being, mortal; dark, grey; others: μοÏÏόνâ (Hsch.);
-
Call. fr. 467, taken from Ammoniusâ (5th c. CE) commentary to Aristotleâs De interpretatione (38.16): διὸ καὶ Ïὸ âá¼Î´Îµá½·Î¼Î±Î¼ÎµÎ½ á¼ÏÏεα μοÏÏοίâ ÏηÏιν á½ ÎÏ Ïηναá¿Î¿Ï âtherefore the Cyrenaean poet says: we mortals have built citiesâ.16
The evidence as regards the accentuation is conflicting, and it is possible that the gloss in Hesychius refers to more than one attestation. Still, the Callimachus fragment confirms the existence of a word meaning âmortal, manâ. Is it possible that both βÏοÏá½¹Ï and this μοÏÏá½¹Ï (if that was its accentuation) continue PGr. *mrÌ¥tó-? This depends on the evaluation of the names in âμοÏÏÎ¿Ï and ÎοÏÏοâ, the evidence for which was collected and discussed by Masson (1963: 219):
-
ÎγεμοÏÏÎ¿Ï (Lesbos, 4th c. BCE+);
-
ÎλεομοÏÏÎ¿Ï (Aeolis, 2nd c. and Cyclades, 3rd c. BCE);
-
ÎνᾱÏιμοÏÏÎ¿Ï (probably a Cretan mentioned on Abydos);
-
ΧαÏιμοÏÏÎ¿Ï (Lato, IC I, 16, 34); name of an Aetolian (3rd c. BCE);
-
ÎοÏÏονᾱÏÎ¿Ï (Thera, early 5th c. BCE).
Starting from the assumption that âμοÏÏÎ¿Ï can only be an Aeolic reflex of *âmrÌ¥tóâ, Masson reconstructs a pre-form *mórtoâ beside *mrÌ¥tóâ for Proto-Greek in order to account for the West Greek names.17 This would imply that Greek preserved more than one inherited word for âmortalâ from this root. Now, the PIE words for âmortalâ and âdeadâ are notoriously hard to reconstruct,18 but Massonâs identification of âμοÏÏÎ¿Ï and Ved. mártaâ is not evidently correct. It has been submitted that Ved. mártaâ derives not from *mórtoâ, but from *mértoâ,19 but in any case, there is no unambiguous evidence pointing to PGr. *mórtoâ (or *mortóâ) rather than *mrÌ¥tóâ among the forms just discussed. Masson claims that the names in âμοÏÏÎ¿Ï are general Aeolic and West Greek, but all secure examples of these names are attested in Lesbian, Theran, and Cretan.20
It is therefore attractive to suppose that âμοÏÏÎ¿Ï is the regular outcome of *mrÌ¥tóâ in Cretan and Theran. One might then think that the simplex μοÏÏá½¹Ï in Callimachus stems from the dialect of his native town Cyrene, a colony of Thera.21 As for the Lesbian names in âμοÏÏοÏ, an o-colored reflex of *rÌ¥ needs no further explanation, but the vowel slot is awkward (because *rÌ¥ > Aeolic âÏοâ, see section 3.3). One would then have to assume influence of the verbal root *merâ on the vocalization to âμοÏÏÎ¿Ï for a pre-stage of Lesbian.22 In this context, the gloss á¼Î¼Î¿ÏÏενΠá¼Ïέθανεν (Hsch. ε 2399) deserves attention, as it shows that a reflex of the verbal root *merâ was indeed preserved in some (probably poetic) form of Greek.23
In sum, since reconstructing an additional form *mórtoâ âmortalâ (beside *mrÌ¥tóâ) for Proto-Greek would be uneconomical, Cretan names in âμοÏÏÎ¿Ï could be an additional argument for a conditioned reflex âοÏâ < *rÌ¥ in Cretan.24 The conditioning factor âafter labialsâ for the reflex âοÏâ would make good sense from a phonetic point of view.25
3.2 Other West Greek Dialects
In this section, we will turn our attention to Laconian and its colonies (especially Theran and Cyrenaean, section 3.2.1), then consider the evidence from Literary Doric (section 3.2.2), and finally make some remarks on the dialects of Elis (section 3.2.3) and the Argolid (section 3.2.4). I have found no noteworthy details for the dialects of Megara (and colonies), for Cos, Rhodes, Karpathos and the other Doric-speaking islands in the Dodecanese, nor for Messenia. For other regions (Achaea, North West Greek), the details are not very interesting either, as appears from the respective dialectal grammars.26 I will not present a complete overview for all West Greek dialects, but merely try to illustrate the precarious nature of the evidence.
3.2.1 Laconian and Colonies
The dialect of Sparta itself is not very well documented in the (preâ)classical period, but its colonies have produced more inscriptions. In Magna Graecia, Heraclea and Tarentum have yielded important epigraphic material; in the Eastern Mediterranean, Thera was probably colonized by Laconians, and Theran settlers then founded Cyrene in Libya.
The evidence for Theran consists mainly of personal names. As far as names are trustworthy evidence, they may provide evidence for the vocalization to âαÏâ (and perhaps also âοÏâ after a labial consonant) that we just established for Cretan:27
-
ÎαÏÏ ÏÏÎ¿Î»ÎµÎ¼Î¿Ï (IG XII,3 787) and ÎαÏÏÏ [Î¼Î±Ï (IG XII,3 814), both from the archaic period.28
-
ÎαÏÏιâ is attested in ÎαÏÏÎ¹Î´Î±Î¼Î±Ï (passim) and in ÎαÏÏÎ¹Î½Î¹ÎºÎ¿Ï (IG XII,3 419, 3rd c.), see Bechtel (1917: 256).29
-
ΣÏαÏÏοâ in ΣÏαÏÏοÏÎ¿Ï (IG XII,3 330, 2nd c.).
-
ÎοÏÏοâ as a first member in ÎοÏÏοναÏÎ¿Ï (IG XII,3 Supp. 697, early 5th c.). Masson (1963: 220) takes this as the outcome of PGr. *mortoâ, but in view of reasons given above, it seems more likely that PGr. had only *mrÌ¥toâ.
ÎαÏÏÏ â may be the regular outcome of *thrÌ¥suâ or it may have an analogical full grade, so it is not entirely probative.30 The forms with ÎαÏÏιâ, ΣÏαÏÏοâ, and ÎοÏÏοâ, however, are absent from most other Greek dialects. The fact that attestations of these forms are concentrated in Cretan and Theran could suggest a common development of these dialects, but it could also be due to language contact or migrations of the bearers of these names.
The inscriptions from Cyrene, a colony founded by Theran settlers, have been edited by Dobias-Lalou (2000). She discusses the outcome of the syllabic liquids on pages 34â35. Not too much can be deduced from the evidence in appellatives:
-
The noun καÏÏÎ¿Ï âharvest, yieldâ (frequent from the 5th c. onwards, Dobias-Lalou 2000: 195) has the same form in all other dialects; therefore a Koine form or an early borrowing cannot be excluded.
-
In view of its special meaning âchaffâ in Cyrenaean, καÏÏÎ¿Ï could well be a genuine dialectal form (Dobias-Lalou 2000: 195â196). For the reconstruction of *rÌ¥ in the root καÏÏâ, see section 9.6.6.
-
The form γÏοÏÎµÏ Ï âsecretaryâ (SEG 9.13, 16) is otherwise peculiar to the Peloponnese and Crete, but it probably does not reflect a pre-form with *rÌ¥ (see section 9.2.2). The verbal root is γÏαÏâ in Cyrenaean, like in all other Greek dialects.
-
The title ÏÏÏαÏÎ±Î³Î¿Ï and the denominative verb ÏÏÏαÏÎ±Î³ÎµÏ have the same form as elsewhere in West Greek, with the exception of Theran and Cretan.
Many of the personal names attested in Cyrenaean may show the influence of Koine or Epic Greek.31 This does not apply, however, to the first compound member ÎαÏÏι- (Dobias-Lalou 2000: 34) in ÎαÏÏιÏθενηÏ,32 ÎαÏÏαγοÏαÏ,33 and ÎαÏÏιμαÏοÏ.34 With the exception of Theran, names with ÎαÏÏιâ are not found in other Greek dialects, not even in Cretan.35 They could therefore contain information about the regular Theran and Cyrenaean development of *rÌ¥, and they outweigh ÏÏÏαÏαγοÏ, because that form could easily be due to Koine influence. Since Cyrene was colonized from Thera, the vocalization *rÌ¥ > âαÏâ would have taken place before the settlement of Cyrene. Cyrenaean provides no further counterevidence to this assumption. It is true that evidence gained from personal names must be used with caution, but it is not unlikely that the names in ÎαÏÏιâ constitute an archaism, as opposed to Cretan ÎαÏÏαιâ and Ionic-Attic ÎÏαÏαιâ, both reflecting a remodeled form *krÌ¥taiâ.36 A regular Cyrenaean reflex âαÏâ is possibly confirmed by the form καÏÏÎ¿Ï âchaffâ.
3.2.2 Literary Doric
How to evaluate the outcome âαÏâ (and perhaps âοÏâ) in Cretan and Theran with regard to the vocalization in other West Greek dialects? The main question is whether there is any evidence at all for the outcome âÏαâ in these dialects. Unfortunately, it is difficult to reconstruct even scraps of the situation in most of the West Greek dialects.
For Laconian, the closest relative of Theran, the epigraphic material is sparse, but the literary evidence may perhaps offer some clues about the dialectal reflex. In Alcman (worked in Sparta, late 7th c.), Epicharmus (worked in Syracuse, a colony of Corinth, early 5th c.), Sophron (Syracuse, 2nd half 5th c.) and some other literary sources, we find the comparative κάÏÏÏν âbetterâ, from an earlier *krÌ¥ti̯Ån.37 In Cretan, this comparative has been restored as καÏÏονâ.38 Apparently, the zero grade of the positive καÏÏεÏÎ¿Ï has been introduced into the comparative both in Cretan and in the dialect(s) underlying κάÏÏÏν. But from which specific dialect(s) was κάÏÏÏν taken?
It is likely that κάÏÏÏν was not the regular outcome of *krÌ¥ti̯Ån in all Doric vernaculars. The Syracusan mimographer Sophron used a middle perfect ptc. á¼Î¼Î²ÏαμέναΠεἱμαÏμένα (fr. 114 K-A, acc. to EM 334.10, cf. á¼Î¼Î²ÏαÏαιΠεἵμαÏÏαι Hsch. ε 2313) as well as an aorist 2sg. á¼ÏÏÎ±Î´ÎµÏ âyou fartedâ (fr. 136 K-A, contrast á¼ÏαÏδον in Attic comedy). This could suggest that Syracusan has a regular reflex *rÌ¥ > âÏαâ, and that κάÏÏÏν was taken from another Doric dialect to become the form of the literary Koine. This dialect may have been Laconian, given that the oldest literary attestation of κάÏÏÏν is in Alcman, and given the prestige of his poetry. According to Hinge (2006: 38), a Laconian context is also suggested by two other sources for κάÏÏονεÏ.
If this is correct, Laconian would agree with the Spartan colony Thera (and with Cretan) in having the vocalization âαÏâ, and differ in this respect from at least Syracusan (colony of Corinth).39 The occurrence of κάÏÏÏν in the two Syracusan poets Epicharmus and Sophron is not decisive for the development in that dialect. In the glosses á¼Î¼Î²Ïαμένα and á¼ÏÏÎ±Î´ÎµÏ taken from Sophron, âÏαâ may well be the genuine Syracusan (and therefore Corinthian) vocalization.40 Once again, all this is quite uncertain in view of the limited evidence.
3.2.3 The Dialect of Elis
There is some evidence for *rÌ¥ > âÏαâ also in the dialect of Elis, but it is slight. Most of the evidence in the recent dialectal grammar by Minon (2007) cannot be used to determine the reflexes of *rÌ¥. For instance, it is impossible to determine whether θαÏÏεν (Minon No 20.1) derives from *thersâ or from *thrÌ¥sâ, because âαÏâ may reflect *âεÏâ in Elis. As in other dialects, the verb γÏá½±ÏÏ and its derivatives appear (cf. section 9.2.2 for further discussion). The value of most Elean glosses in Hesychius (discussion in Minon 2007: 549â560) is unclear.
There is, however, one good pair of candidates to show the regular reflex in Elis. The gloss βÏαÏάνανΠÏοÏύνην. Ἠλεá¿Î¿Î¹ (âstirring ladleâ, Hsch.) must be an instrument noun in âάνη derived from a root *u̯ratâ.41 In view of the meaning âstirring ladleâ, a derivation from the root *u̯ertâ âto turnâ suggests itself. The same root is attested in the gloss βÏαÏάνειΠῥαá¿Î¶ÎµÎ¹ á¼Ïὸ νόÏÎ¿Ï . Ἠλεá¿Î¿Î¹ (ârecovers from illnessâ, Hsch.), if we suppose that the meaning developed from âturns betterâ (Minon 2007: 554). This present formation in âá½±Î½Ï probably presupposes the existence of a thematic aorist *u̯rate/oâ (cf. βλαÏÏá½±Î½Ï : βλαÏÏεá¿Î½, á¼Î¼Î±ÏÏá½±Î½Ï : á¼Î¼Î±ÏÏεá¿Î½). It is conceivable that a causative s-aorist *u̯ert-sâ (presupposed by Hom. á¼ÏόεÏÏε âdrove off courseâ)42 coexisted with an intransitive thematic aorist *u̯rÌ¥t-e/oâ in Proto-Greek. If so, the latter form developed into *u̯rate/oâ in Elean, and the noun βÏαÏάνα was also built on the zero-grade root allomorph.
If these two glosses are considered reliable evidence, *rÌ¥ may have yielded âÏαâ in Elis even after a labial glide. The disagreement with the treatment in Cretan would be remarkable: the latter dialect has the outcome âαÏâ, and probably âοÏâ after labials. However, since the evidence comes from just two glosses, we must be careful. It must also be taken into account that the word for âdrachmâ is attested in Elis (Minon 2007: 355): several times as δαÏÏμα and once as δαÏÏναÏ, possibly reflecting a pre-form *drÌ¥khmnÄâ.43 However, the word for âdrachmâ could well be an inter-dialectal loan. Under these conditions, it would not be wise to base any firm conclusions on the evidence at our disposal.44
One epigraphic form from Elis is highly relevant for the development of *lÌ¥: αÏλανεοÌÏ âcompletely, all togetherâ (Minon No 4.4 and 8.3). The same form is attested in the gloss á¼Î»Î±Î½á½³ÏÏÎ á½Î»Î¿ÏÏεÏῶÏ. ΤαÏανÏá¿Î½Î¿Î¹ (âentirely, completelyâ, Hsch.), proving that this form is old in West Greek. As I will argue in section 10.6, these forms show that *lÌ¥ yielded âλαâ in Elean and Laconian even before nasals. Moreover, since the Cretan outcome may have been âλοâ after a labial consonant (cf. sections 10.3.1 and 10.6.1), it is possible that Proto-West-Greek preserved not only *rÌ¥, but also *lÌ¥.
3.2.4 The Dialects of the Argolid
The evidence for the development of the syllabic liquids from Western and Eastern Argolic is presented in full detail by Nieto Izquierdo (2008: 145â151 and 380â381). It comprises the following forms:
-
The verb γÏαÏÏ âto writeâ (e.g. impf. [ε]γÏαÏε, ptc. γεγÏαθμενοÏ) and the derivative γÏαθμα / γÏαÏÏμα âletterâ < *graphma;45
-
The nouns γÏοÏÎµÏ Ï âscribeâ, γÏοÏÎ¹Ï âstylusâ, and deverbal forms such as γÏοÏα, αγγÏοÏα, εγγÏοÏα, ÏÏ Î³Î³ÏοÏοÏ, etc. (see section 9.2.2 for further analysis, especially of the o-vocalism);
-
WArg. εξÏÏÏαÏεÏαι (Tiryns, SEG 30:380, no. 6.1, ca. 600â550â¯BCE(?)) corresponding to Att. á¼Îº-ÏÏÏá½³ÏÏ;
-
Epid. κÏαμαÏαι (IG IV2,1 122.3, ca. 320â¯BCE) corresponding to Att. κÏεμάÏαι âto hangâ;
-
The root ÏÏÏαÏâ âarmyâ in ÏÏÏαÏηα âarmy, expeditionâ (Del.3 84, Tylisos, 460â450â¯BCE), ÏÏÏαÏÎ±Î³Î¿Ï âgeneralâ (SEG 29:361, Argos, appr. 400â¯BCE), and often in personal names;
-
EArg. ÏÏαÏεζιÏÎ±Ï âmoney-changerâ (Epidaurus, 3rd c. BCE);
-
WArg. ÏαÏγμα âfenceâ (Del.3 89.8, cf. SEG 37:279, Argos, ca. 350â¯BCE);
-
EArg. ÏαÏξιν âfenceâ (Epidaurus, IG IV2,1 102 B, l. 75, 400â350â¯BCE);
-
EArg. ÏαÏÏμαÏα âid.â (Epidaurus, ibid. l. 253);
-
EArg. διαÏỊ̈α̣γ̣[μά]ÏÌ£Ïν (Epidaurus, IG IV2,1 115, l. 22, cf. SEG 25:393, ca. 330â300â¯BCE).
The forms with a root ÏαÏÏâ have older attestations than διαÏỊ̈α̣γ̣[μά]ÏÌ£Ïν; hence the latter must represent a Koine form. This is confirmed by the unassimilated root-final stop of ÏαÏÏμαÏα in Epidaurus. Likewise, the form ÏÏαÏεζιÏÎ±Ï may well be a Koine form, as it is attested relatively late and is a normal word in the Koine. Both γÏá½±ÏÏ (and derivatives) and forms with γÏοÏâ are genuine Argolic dialect forms, and the same probably holds for εξÏÏÏαÏεÏαι, κÏαμαÏαι and the forms with ÏÏÏαÏâ.
With these reductions in mind, it appears that Argolic has a reflex âαÏâ in ÏαÏξιν, ÏαÏγμα, ÏαÏÏμαÏα, as against âÏαâ in κÏαμαÏαι, γÏαÏâ, ÏÏÏαÏâ and ÏÏÏαÏâ. Clearly, κÏαμαÏαι and ÏÏÏαÏâ may have an analogical vowel slot (cf. Att. κÏεμάÏαι, ÏÏÏá½³ÏÏ), and the same may be true of γÏαÏâ (beside γÏοÏâ). The reflex ÏÏÏαÏâ cannot be analogical and therefore seems to prove a regular reflex âÏαâ in this dialect. However, it must be taken into account that all Greek dialects except for Cretan and Cyrenaean use the root ÏÏÏαÏâ, so some caution is in order. In the case of ÏαÏξιν, ÏαÏγμα, ÏαÏÏμαÏα, much depends on the reconstruction and etymology of the verb ÏÏá½±ÏÏÏ, which will be discussed in section 9.2.3.
Thus, no definite conclusion concerning the regular place of the anaptyctic vowel can be drawn for the dialects of the Argolid.
3.2.5 Conclusion on the West Greek Dialects
The only West Greek dialect for which we have clear-cut evidence is Cretan. In this dialect, *rÌ¥ yields âαÏâ, and perhaps âοÏâ after a labial consonant. There is slight evidence for a regular outcome âÏαâ in glosses from Elis and Syracuse, and for âαÏâ in Theran and Cyrenaean onomastic material. If the evidence for âÏαâ in the former two dialects is taken seriously, the divergence with Cretan would show that Proto-West Greek, and even Proto-Doric, kept *rÌ¥ intact. If so, the vocalization may well have taken place during the Dorian migrations in the early Dark Ages. The different reflexes can be ascribed to the different situations of linguistic contact between speakers of West Greek and the earlier populations in the regions where they settled.
3.3 The Aeolic Dialects
Our ability to reconstruct the prehistory of the Aeolic dialects is hampered in several respects. The most abundant sources of examples are the Lesbian poets Sappho and Alcaeus, but the status of this evidence is not always clear, because a number of forms may be hyper-Aeolic or of epic origin.46 Epigraphic material from Lesbian has already undergone considerable Koine influence at a time when inscriptions start to become numerous. Most Thessalian evidence is also late and much of it suffers from the same problem. Boeotian is a mixed dialect which contains many West Greek features. Finally, a large part of the evidence consists of personal names, where influence of other dialects or even the poetic language is a factor to be reckoned with.
In addition to these factual problems, there are practical ones. As yet, there is no comprehensive grammar of the Thessalian dialects, nor of Boeotian.47 The generative description of the Aeolic dialects by Blümel (1982) is of some use, but has no separate treatment of the reflexes of *r̥.
Notwithstanding all these problems, the combined evidence of our sources does allow us to draw a definite conclusion: the regular reflex was âÏοâ, without further conditioning, in all Aeolic dialects. I will now first review the epigraphic evidence, and after that turn to the extant fragments of Sappho and Alcaeus. Homeric words with âÏοâ will not be discussed here, but in chapter 7, as their Lesbian or Aeolic provenance is not certain.
3.3.1 The Numerals in the Aeolic Dialects
Let us start with the variation between Ïα/Î±Ï and Ïο/Î¿Ï in numeral forms in the Aeolic dialects. The attestations are conveniently listed in Blümel (1982: 271â275). Concerning the variations in form, he notes that âdie Einzelheiten der Abgrenzung zwischen phonologischen und morphologischen Ursachen sind noch nicht übereinstimmend geklärtâ (Blümel 1982: 52â53).
In section 2.7, it was proposed that the numerals in Aeolic dialects underwent analogical modifications similar to those taking place in other dialects. For instance, the Boeotian forms ÏεÏÏαÏÎ¿Ï and ÏεÏÏα- are usually explained as contaminations of original Aeolic *ÏεÏÏοÏοÏ, *ÏεÏÏοâ with Attic or West Greek forms (e.g. Waanders 1992: 379), but it is more likely that they underwent the same analogical influence of higher numerals (such as á¼ÏÏαâ, δεκαâ) that yielded Ionic-Attic ÏεÏÏαâ. Another parallel for this influence is Arcadian ÏεμÏοÏÎ¿Ï âfifthâ, which was clearly reshaped on the basis of δεκοÏÎ¿Ï âtenthâ and ενÏοÏÎ¿Ï âninthâ;48 Note that the forms δεκαÏÎ¿Ï and εναÏÎ¿Ï are actually attested in Boeotian inscriptions, contrasting with δεκοÏÎ¿Ï and/or ενοÏÎ¿Ï as attested in Lesbian and Thessalian inscriptions.
The same explanation can be applied to the Thessalian forms ÏεÏÏο-εÏηÏιδα and ÏεÏÏοÏοÏ. In the ordinals, Thessalian has δεκοÏÎ¿Ï with a medial o-vowel that may have spread from ânineâ or even âeightâ (cf. the shortened form οκÏο in Boeot. and Lesb.). In compounds, Thess. εξομεινον âperiod of six monthsâ (IG IX,2 506.4) is another clear instance of the spread of âοâ as a linking vowel, comparable to the spread of âαâ in Ionic-Attic ÏενÏαâ, á¼Î¾Î±â (after á¼ÏÏαâ, â¦, δεκαâ).49 Therefore, it is not completely certain that Thess. ÏεÏÏοâ is the regular outcome of *kwetrÌ¥â before consonants.
3.3.2 Epigraphic Evidence (Boeotian, Thessalian, Lesbian)
I start from the forms given in the dialect grammars.50 Most discussions of the Boeotian reflex of *rÌ¥ cite just two forms: the compound elements âÏÏÏοÏÎ¿Ï and ÎÏοÏ-, which are widespread in proper names from the region.51 Importantly, the word for âarmy; campaignâ does not only appear in names, but also in the verbal form εÏÏÏÏοÏÎµÏ Î±Î¸Î· (IG VII, 3174 and passim).52 Boeotian also has instances of a-vocalism such as ÏεÏÏαâ and ÏεÏÏαÏοÏ, but as we have just seen, these forms may be analogical. Thus, although Boeotian does not offer much information, ÏÏÏοÏÎ¿Ï definitely speaks in favor of a regular development *rÌ¥ > Ïο.
As for Lesbian, a first important piece of evidence is ÏÏÏοÏαγοÏ, denoting a magistrate. As Hodot (1990: 56) remarks, this title is in the process of being replaced by ÏÏÏαÏαγοÏ, a hybrid form consisting of dialectal (âαγοÏ) and Koine (ÏÏÏαÏâ) elements. The real Koine form ÏÏÏαÏηγόÏ, with Ionic âηâ, never occurs in Lesbian inscriptions. Other forms derived from this lexeme introduced a-vocalism much earlier, e.g. ÏÏÏαÏεια (Hodot, NAS 01, 4th c.). In the literary tradition, ÏÏÏá½¹ÏÎ¿Ï is attested in Sappho. Apparently, in the classical period the genuine dialectal form ÏÏÏá½¹ÏÎ¿Ï was preserved only in the title ÏÏÏοÏαγοÏ.
The second important epigraphically attested form is αμβÏοÏην âto transgressâ (IG XII,2 1.5), which again has a counterpart in literary Lesbian, á¼Î¼Î²ÏοÏε (on which see below). There is no further relevant evidence: the root γÏαÏâ (in forms like ανÏιγÏαÏÎµÏ Ï, γÏαÏην) is well-attested, as it is in all other dialects, but it need not have contained *rÌ¥. The attestation of ÏαÏξ (Hodot 1990, MAT 03.11 and 05.16, 21, end of 3rd c.) is relatively late, so it could be a loanword from Ionic. In conclusion, both ÏÏÏοÏÎ±Î³Î¿Ï and αμβÏοÏην are trustworthy evidence for the development *rÌ¥ > Ïο in Lesbian.
In Thessalian, the root *mrÌ¥khâ âshortâ is attested as a personal name ÎÏοÏÏ Ï (IG IX,2 460.13, Krannon, Pelasgiotis, 2nd c.) and in its older form in the female name ÎÏοÏÎ¿Ì (SEG 24.406, Perrhaebia, 500â450â¯BCE). The name ÎοÏÏÎ¹Î´Î±Ï (SEG 26, 672.32, Larisa, Pelasgiotis, early 2nd c.) may be due to a later metathesis, if it is indeed related. As we have seen, ÎÏοÏÏ â is also found as an onomastic element in Boeotian.
A regular Thessalian outcome âÏοâ is often thought to be supported by ÏεÏÏοεÏηÏιδα (RPh. 1911, 123.26, Larisa, 1st c.). Since this form has an unexpected spelling â¨Î·â© of the outcome of *Ä (which is otherwise spelled as â¨ÎµÎ¹â© in this dialect), and since the inscription has a number of Koine features, the evidential value of this form used to be questioned. Later, however, the form ÏεÏÏο- has been confirmed by ÏεÏÏοεÏειÏιδα (SEG 17.288 passim, Larisa, 1st c. BCE or later) and by the ordinal ÏεÏÏοÏÎ¿Ï (SEG 43.311, Skotoussa, Pelasgiotis, early 2nd c.).
Nevertheless, it is not completely certain that these forms contain a reflex of *rÌ¥: we must take into account that Myc. qe-to-ro- and Class. ÏεÏÏα- may be remodellings of the older form PIE *kwetru- based on the compositional form of âten-â (see sections 2.7 and 3.3.1). The same could be true of Thess. ÏεÏÏο-, as a linking vowel âoâ also occurs in Thess. εξομεινον âperiod of six monthsâ (IG IX,2 506.4). Moreover, an o-vowel is attested for the ordinals δεκοÏÎ¿Ï âtenthâ (at Larisa and Skotoussa, SEG 27.202, passim) and ενοÏÎ¿Ï âninthâ (SEG 43.311, Skotoussa). Therefore, given the model of the 1st compound members ÏεÏÏοâ, εξοâ, it cannot be completely excluded that an older form *ÏεÏοÏÏÎ¿Ï was changed into ÏεÏÏοÏοÏ.53
Another Thessalian piece of evidence has been adduced by GarcÃa Ramón (1999: 11â13): he argues that ÎÏοÏια, an epiclesis of Artemis at Atrax and Larisa in the Hellenistic period, is derived from *θÏá½¹ÏÎ¹Ï < *dhrÌ¥-tiâ âsupportâ.54 In his opinion, ÎÏοÏια refers to Artemis in her function as a supporter and protector of youngsters in a rite of initiation. An alternative derivation from the root *dhersâ of θÏαÏá½»Ï has been proposed (see e.g. LSJ), but GarcÃa Ramón objects that this adjective was continued in Thessalian as θεÏÏÏ Ï (with full grade root) on account of a different epiclesis, Îθανα ÎεÏÏÏ Ï.55 In either case, whether the underlying etymon is *dhrÌ¥-tiâ or *dhrÌ¥sâ, ÎÏοÏια would be an example for the vocalization of *rÌ¥. Being a name, however, the form must be handled with due caution.
For GarcÃa Ramón, ÎÏοÏια is an important form because it would show that the o-colored reflex in Thessalian occurs not only in contact with labial sounds.56 He also refers to the (as yet unpublished) Thessalian form οÏÏεν âmaleâ, which contains no initial digamma and derives from the zero grade also reflected in Hom. á¼ÏÏην. I would add that, contrary to what is often stated, the *rÌ¥ in ÏεÏÏοÏÎ¿Ï did not stand in a labial environment either: the simplification *tu̯rÌ¥ > *trÌ¥ took place early on (section 2.6).
We may conclude that there is secure evidence for o-coloring of the Thessalian reflex, but the regular vowel slot is less clear than in Boeotian or Lesbian. The only direct piece of evidence for âÏοâ is the epiclesis ÎÏοÏια. The names ÎÏοÏοÌ, ÎÏοÏÏ Ï may have a levelled vowel slot (PGr. *mrekhâ, cf. section 4.3.3), and for ÏεÏÏοÏÎ¿Ï we cannot exclude that it was influenced by the compounding form ÏεÏÏοâ, which itself may have taken its o-vocalism from higher numerals. In the unpublished form οÏÏεν, á½Ïâ may be an inner-paradigmatic restoration from ῥοâ after the full grade seen in á¼ÏÏην, or it could show a special development of *rÌ¥ in word-initial position (see section 9.1.7 for further discussion).
Thus, although the evidence does not completely exclude âοÏâ, the word-medial reflex in Thessalian was most probably âÏοâ, as in Lesbian and Boeotian.
3.3.3 The Relation between Lesbian Lyric and Ionian Epic
Linguistic evidence from the poems and fragments of Sappho and Alceaus is to be used with caution for more than one reason. As remarked above, this material may not only contain Ionic words with a-vocalism; it may also have undergone hyper-aeolicization at the hands of later editors or copyists. After the work of Lobel, it was thought for some time that Sappho composed her poems not in a literary dialect, but in the Lesbian vernacular (cf. the discussion in Bowie 1981: 60â¯ff.). In order to maintain this thesis, Lobel had to reject a number of Sapphic fragments as ungenuine, and to assume a large number of emendations in the other fragments. As Bowie remarks, however, some fragments that were declared non-Sapphic by Lobel had the same metre as others that he did consider genuine. Thus, Lobelâs criteria for emending forms or rejecting entire fragments lacked a solid basis.
Since the monographs by Hooker (1977) and Bowie (1981), two things have become much clearer. First of all, there has been substantial Ionic influence on the language of Sappho and Alcaeus. This influence can be ascribed, to a large extent, to the epic tradition, but it is equally possible that vernacular Lesbian vocabulary used by Sappho and Alcaeus had been influenced by that of the neighboring Eastern Ionic vernaculars (Bowie 1981). Secondly, Sappho and Alcaeus used a literary dialect for their genre which had a tradition of its own, as follows from the meters they use. Aeolic lyric may owe part of its vocabulary and phraseology to this tradition, but it is unclear what exactly the Lesbian contribution was.57
The influence of Ionic on the language of the Lesbian poets was clearly substantial. In practice, it is often difficult to decide whether a given Ionic form is due to epic influence or to borrowing from the Ionic vernacular, but that is irrelevant for present purposes.58 Both Lesbian poets make use of a fair number of epic lexemes and grammatical characteristics, especially (though not exclusively) in poems with epic subject matter. Convincing cases of Ionic or epic influence include the following:59
-
the long vowel s-aorist subjunctive (e.g. ÏαÏξώμεθα, Alc. 6.7), which is typical for Ionic-Attic;
-
á¼Î´ÎµÎ»Ïέα âsisterâ (Sapph., Alc.) < *ha-gwelphehÄâ, with a dental reflex of the labiovelar (Bowie 1981: 89â90);
-
the form Î á½³ÏÎ±Î¼Î¿Ï (Sapph. 44.16), a compromise between the Lesbian form Î á½³ÏÏÎ±Î¼Î¿Ï and the metrical structure of Homeric Î Ïá½·Î±Î¼Î¿Ï (Bowie 1981: 58).
Bowie (1981: 137) further mentions as epicisms the forms ÏεÏιÏέλλεÏαι, ÏίλναÏαι, γαá¿Î± (vernacular Lesbian γᾶ), ῤá¿Î± (for Hom. ῥεá¿Î±, contrast vernacular Lesbian βÏᾶ), á¼Î¼Ïá½· governing the dative, ÏοÏέονÏαι (with thematic contract verb inflection), and á¼ÏÏÏ Ïέλιξε (velar flexion of the s-aorist). This list could easily be extended.60 As candidates to have been borrowed from spoken Ionic into the Lesbian vernacular, i.e. forms for which it is unnecessary to assume epic influence, Bowie (1981: 136) mentions ἴεÏοÏ, ÏοιαύÏα, κάÏÏεÏοÏ, the 3pl. ind. aor. ending âÏαν, the pf. ptc. á¼Î¿á½·ÎºÎ¿ÏÎµÏ (in Aeolic, one would expect âονÏεÏ), and ἤÏÎµÏ (enclitic âÏÎµÏ is otherwise absent from Lesbian).
Like ÏαÏξώμεθα, a number of forms with Ïα/Î±Ï < *rÌ¥ that are found in Sappho and/or Alcaeus may stem from Ionic (for a full list, see section 3.3.5 below). For this reason, I disagree with OâNeil (1971) and Wyatt (1971) that Ïα/Î±Ï is the regular reflex in Lesbian under certain conditions.
3.3.4 Evidence for o-vocalism in Literary Lesbian
The only potential evidence for the reflex of *lÌ¥ is á¼á½¹Î»Î»ÎµÎµÏ (Alc. 348.3) This form may be either an epicism or an instance of *lÌ¥ > âολâ before nasals; see section 10.5.4 for further discussion. I will therefore focus on the reflex of *rÌ¥.
The following forms from Sappho and Alcaeus, in alphabetical order, must be considered as potential evidence for âÏοâ as a regular reflex (and âοÏâ as analogical):
-
βÏá½¹Ïε(α) (Sapph. 31.7)61
-
δÏá½¹[μÏμεν (Alc. 6.8)
-
δÏá½¹ÏÌ£[Ï]Ïιν (Alc. 119.15)
-
á¼Î¼Î¼á½¹Ïμενον (Alc. 39a.7)
-
ÏÏá½¹Ïην (Alc. 70.9), á½Î½á½³ÏÏοÏε (Alc. 72.8), ÏεδέÏÏοÏ[ε (Alc. 75.11)
-
ÏÏÏá½¹Ïον (Sapph. 16.1, Alc. 382.2).62
The following forms are well-attested in Epic Greek and can be analyzed as epicisms in Lesbian lyric.63 For this reason, they cannot be used as secure evidence in the present discussion (even if the form á¼Î¼Î²ÏοÏε is confirmed for Lesbian by epigraphic evidence: see above):
-
á¼Î¼Î²ÏοÏε aor. 3sg. (Sapph. 5.5), cf. Hom. aor. ἤμβÏοÏε;
-
á¼Î¼Î²ÏοÏá½·Î±Ï (Sapph. 141.1), cf. Hom. adj. á¼Î¼Î²Ïá½¹ÏιοÏ;
-
á¼ÏÏοδίÏα (Sapph. 1.1 and passim), cf. Hom.+ á¼ÏÏοδίÏη;
-
βÏόδÏν (Sapph. 55.2), βÏόδοιÏι (Sapph. 2.6), βÏοδοÏá½±ÏÎµÎµÏ (Sapph. 53; 58.19), βÏοδοδάκÏÏ Î»Î¿Ï (Sapph. 96.8): cf. the traditional epithets ῥοδοδάκÏÏ Î»Î¿Ï (Hom.+) and ῥοδόÏηÏÏ Ï (Hes.+).
I also leave aside the following forms:64
-
á½ÏÏεÏον âbeast, creatureâ (Sapph. 130.2) beside Ion.-Att. á¼ÏÏεÏόν: the original vocalism of the Lesbian form is unclear. In his monographic treatment of this form and the suffix âεÏá½¹â, Vine (1998: 74) concludes that á½ÏÏεÏον may have been contaminated with a form *á½ÏÏá½¹â âcreeperâ that is perhaps also presupposed by á½ÏÏηξ âyoung shoot, saplingâ;
-
Î¼á½¹Î»Î¸Î±ÎºÎ¿Ï âsoftâ (Sapph. 46.1, Alc. 338.8) has no convincing etymology, see section 10.1.7;
-
the forms γÏá½¹ÏÏαÏα and γÏá½¹ÏÏα (Balbilla) are probably hyper-Aeolisms in view of the universal occurrence of γÏαÏâ in Lesbian inscriptions;
-
The sequence ].ÏÏοÏÏε ÏίδαÏ[ (Alc. 179.12) may well contain the Aeolic form corresponding to epic á¼ÏÏÏá½±ÏÏÏ âto flash (of lightning)â, but the reconstruction of this etymon is uncertain;
-
ÏοικιλόθÏονâ (Sapph. 1.1) probably contains the word θÏόνα (to be kept distinct from θÏá½¹Î½Î¿Ï âthroneâ): cf. the discussion in section 2.5.2.
Returning to the potential evidence, the forms βÏá½¹Ïεâ, á¼Î¼Î¼á½¹Ïμενον, ÏÏÏá½¹Ïον, and the thematic aorists δÏομε/οâ, δÏοÏε/οâ, and ÏÏοÏε/οâ cannot be due to epic influence, as Homer attests these forms in a shape with âÏαâ or âαÏâ. These words with a reflex âÏοâ were at home in the Lesbian poetic tradition, and they probably entered this tradition as Lesbian vernacular forms. Indeed, the following forms with âÏοâ are backed up by epigraphic evidence from Lesbian or other Aeolic dialects:
-
á¼Î¼Î²ÏοÏε (αμβÏοÏην IG XII,2 1.5);
-
βÏá½¹Ïεα (Thess. ÎÏοÏοÌ, Boeot. ÎÏοÏÏ Î»Î»Î¿Ï, see above);
-
ÏÏÏá½¹ÏÎ¿Ï (Lesb. ÏÏÏοÏαγοÏ, Boeot. εÏÏÏÏοÏÎµÏ Î±Î¸Î·, names in âÏÏÏοÏοÏ).
As has already been noted, ÏÏÏá½¹ÏÎ¿Ï (beside Homeric and class. ÏÏÏαÏá½¹Ï, Cret. ÏÏαÏÏοÏ) shows that the o-colored reflex was regular also in a non-labial environment.
What does this evidence teach us about the regular place of the epenthetic vowel in Lesbian? In order to answer this question, let us now discuss some of the attested forms in more detail:
-
Like Ionic βÏαÏá½»Ï, Aeol. βÏá½¹Ïεα and Thess. ÎÏοÏÏ Ï (IG IX,2 460.13) may have leveled the old full grade slot (attested in Lat. brevis; cf. section 4.1.1 and 4.3.3).
-
There are three examples of thematic aorists with o-vocalism in Lesbian poetry. Of these, ÏÏá½¹Ïην is well-attested (prefixed forms á½Î½á½³ÏÏοÏε, ÏεδέÏÏοÏε); besides, δÏá½¹ÏÌ£[Ï]Ïιν and the restored form δÏá½¹[μÏμεν also clearly speak in favor of o-vocalism. Clearly, the vowel slot of ÏÏοÏâ could be analogical: cf. fut. á½Î½ÏÏá½³Ïει, pres. inf. á¼ÏιÏÏá½³Ïην. The same goes for the vowel slot seen in δÏόμÏμεν and δÏá½¹ÏÏÏιν.
-
As for á¼Î¼Î¼á½¹Ïμενον, the corresponding Attic form εἱμαÏÎ¼á½³Î½Î¿Ï guarantees the antiquity of the formation, a middle perfect *he-hmrÌ¥-toi with zero grade root. However, á¼Î¼Î¼á½¹Ïμενον does not constitute compelling evidence for a regular alternative treatment *rÌ¥ > âοÏâ in Lesbian, because in Homer we also find the older active perfect á¼Î¼Î¼Î¿Ïε (normally viewed as an Aeolism). Since the substitution of middle for older active perfect forms is widespread (cf. ÏεÏÏ Î³Î¼á½³Î½Î¿Ï beside older Myc. te-tu-ku-wo-a2),65 the vowel slot of Aeol. á¼Î¼Î¼á½¹Ïμενον may have been influenced by that of á¼Î¼Î¼Î¿Ïε.
The remaining forms clearly show that the regular Lesbian outcome of *CrÌ¥Tâ was CroTâ. The clearest instances are á¼Î¼Î²ÏοÏε (epigraphic αμβÏοÏην), ÏÏÏá½¹ÏÎ¿Ï (epigraphic ÏÏÏοÏαγοÏ), and the thematic aorist forms (á½Î½á½³ÏÏοÏε, ÏεδέÏÏοÏεν). The epenthetic vowel regularly appears after the liquid in the isolated forms á¼Î¼Î²ÏοÏε, αμβÏοÏην, and ÏÏÏá½¹ÏοÏ.66 This reflex âroâ is a clear characteristic of Lesbian and Aeolic generally. In this respect the Aeolic dialects differ from Mycenaean, and also from Arcadian, where the regular reflex was âorâ (as we shall see below).
3.3.5 Evidence for a-vocalism in Literary Lesbian
The following list contains all potential evidence for an a-colored reflex of *r̥ and *l̥ in literary Lesbian:
-
βÏá½±Î´Î¹Î½Î¿Ï âsuppleâ (Sapph. 44A(b).7, 102.2, and 115)
-
á¼Î±ÏÎ¿Ï âspringâ gen. (Alc. 296b.3), contracted ἦÏÎ¿Ï (Sapph. 136, Alc. 367)
-
καÏδία âheartâ (Sapph. 31.6, Alc. 207.9)
-
κάÏÏÎ¿Ï âharvestâ (Alc. 119.10)
-
κάÏÏεÏÎ¿Ï âstrongâ (Alc. 119.19; probably also Alc. 302 (col. 2).19)
-
á¼Î¼Î±ÏÏε âseizedâ (Sapph. 58.21), μαÏÏαι[ (Alc. 61.14)
-
νέκÏÎ±Ï ânectarâ (Sapph. 2.15 and 96.27)
-
á½Î½Î±Ï âdreamâ (Sapph. 134)
-
á½Î½Î·Î±Ï âbenefitâ (SLG, S286(2).10)
-
Ïá½±ÏÎ¸ÎµÎ½Î¿Ï âmaidenâ (Sapph. 56 passim, Alc. 42.8)
-
Ïá½±Ïβημι âto be scaredâ (Alc. 206 and 302.12)
-
ÏÏá½±Î³Î¿Ï âhe-goatâ (Alc. 167.5).
A number of these forms must be left out of consideration: á¼Î¼Î±ÏÏε and Ïá½±ÏÎ¸ÎµÎ½Î¿Ï are pan-Greek forms without a convincing etymology (cf. section 9.7.2). For ÏÏάγοÏ, a pre-form with *rÌ¥ is uncertain as we might be dealing with a secondary zero grade beside the present ÏÏώγÏ, from a root *treh3gâ (section 9.1.4). The epic verb ÏαÏÎ²á½³Ï was also utilized in Lesbian poetry and in the Attic tragedians; its a-vocalism may point to an Ionic-Attic origin, if the derivation from a zero grade of PIE *tergwâ is correct (cf. section 4.2.1). Furthermore, κάÏÏεÏÎ¿Ï is certainly a borrowing from Ionic, whether from the vernacular or from Epic Greek; see the arguments adduced by Bowie (1981: 99â100).67 It is also conceivable that Aeol. κάÏÏÎ¿Ï is of epic or Ionic origin, but this word is difficult to evaluate in any case, as it has the same form in all dialects where it is attested.
Examples for âÎ±Ï as the word-final treatment of *rÌ¥ are á½Î½Î±Ï, νέκÏαÏ, á½Î½Î·Î±Ï, and á¼Î±ÏοÏ. Given that it displays the change *Ä > η, á½Î½Î·Î±Ï must be a borrowing from Ionic.68 The three other forms also occur in Homer, and especially á½Î½Î±Ï and νέκÏÎ±Ï are liable to an analysis as epicisms. The gen. sg. á¼Î±ÏÎ¿Ï is commonly thought to have been built on the nom. á¼Î±Ï (Hom.+) < PIE *u̯es-rÌ¥. The two instances of contracted ἦÏÎ¿Ï may be of Ionic origin, and uncontracted á¼Î±ÏÎ¿Ï may be an epicism. Thus, there is no direct evidence for the Lesbian vernacular development of *rÌ¥ in word-final position.69
Two remaining forms require a more elaborate discussion. Both Lesbian poets use καÏδία as a word for âheartâ. On account of ÏÏÏá½¹ÏÎ¿Ï beside Ionic ÏÏÏαÏá½¹Ï, which is a secure example, it can be excluded that καÏδία contains the regular reflex of *rÌ¥ in Lesbian. Since καÏδίη was certainly the regular vernacular form in neighboring Ionic dialects, we must assume that this form was borrowed into Lesbian. There is another possible, but rather uncertain attestation κ]ο̣Ïδίαν (Alc. 130A.4, initial κâ suggested by Diehl).70 If this is indeed the correct reading, it could be a hyper-Aeolic form, as the vocalization âοÏâ would be at odds with the otherwise clear evidence for âÏοâ as the Lesbian reflex.
The adjective βÏá½±Î´Î¹Î½Î¿Ï occurs three times in Sappho, and its βÏâ certainly represents earlier *u̯râ (see the discussion in Bowie 1981: 80â84).71 The word has two main types of referent: (1) soft or supple body parts of women, e.g. feet, hands, cheeks, and also Aphrodite herself; (2) shoots, branches, and a whip that are âsupple, tapeableâ. An etymological connection with Ved. vrad âbecome softâ could therefore be envisaged, but this remains uncertain as the suffixation in âÎ¹Î½Î¿Ï remains without a good parallel.72 There are two options to explain âÏαâ in this clearly poetic word. First, it is conceivable that βÏá½±Î´Î¹Î½Î¿Ï stems from the Ionian epic tradition. Secondly, since βÏá½±Î´Î¹Î½Î¿Ï has no secure etymology, it can be argued that a pre-form with *rÌ¥ is not ascertained. In this case the word could be genuine Lesbian or belong to the Aeolic poetic tradition.
3.3.6 Evidence for Aeolic o-vocalism in Ancient Grammarians
As various previous authors have noted, much of the evidence from the ancient grammatical tradition cannot be relied upon. In many cases, there is no dialect indication: for instance, a gloss like μοÏνάμενοÏΠμαÏá½¹Î¼ÎµÎ½Î¿Ï (Hsch., cf. μάÏναμαι âto fightâ) need not be Aeolic, but could instead stem from Arcadian or Cyprian.
In other cases, the sources of the ancient grammarians cannot be determined. For instance, the adverb θÏοÏá½³ÏÏ and the noun ÏÏá½¹ÏÎ¼Î¿Ï (for ÏÏαÏÎ¼á½¹Ï âsneezeâ) are cited as âAeolicâ in the Compendium ÏεÏὶ διαλέκÏÏν attributed to Johannes Grammaticus, and they do not contradict the conclusions obtained so far: θÏοÏá½³ÏÏ has the expected Aeolic reflex of *rÌ¥, and if ÏÏá½¹ÏÎ¼Î¿Ï is a real form, various accounts are conceivable.73 The middle perfect forms Ïá½³ÏοÏθαι, μέμοÏθαι and á¼ÏθοÏθαι (wrongly referred to as âHomericâ by OâNeil 1971: 26) are cited as Aeolic in (pseudoâ)Herodianâs On Iliadic Prosody 67.74 If such forms indeed once existed, the analysis of á¼ÏθοÏθαι is clear enough, but Ïá½³ÏοÏθαι and μέμοÏθαι are difficult to interpret etymologically. Now, á¼ÏθοÏθαι may have an analogical vowel slot (ÏθεÏâ, ÏθοÏâ, ÏθαÏâ) and does not speak against a regular development to âÏοâ anymore than does á¼Î¼Î¼á½¹Ïμενον in Alcaeus (cf. above); the same may be true of Ïá½³ÏοÏθαι and μέμοÏθαι, whatever their etymology is.
Having said that, one pair of glosses attested in Hesychius clearly supports the Aeolic development of o-vocalism that we have just established on the basis of the literary and epigraphic evidence: ÏοÏνάμενΠÏÏλεá¿Î½ âto sellâ and ÏοÏνάμεναιΠκενÏούμεναι, ÏÏλούμεναι (Hsch. Ï 3042 and 3043).75 The forms reflect the nasal present-stem *prÌ¥-n-h2â that is also continued as Ionic Ïá½³Ïνημι âto sellâ, with the difference that in the latter form, the root vowel was secondarily adapted to that of the aorist ÏεÏá½±Ïαι.76 In ÏοÏνάμεν, the o-vocalism in combination with the infinitive ending âμεν clearly suggests an Aeolic origin. The vowel slot âοÏâ could be conditioned by the following nasal, or be analogical after the aorist ÏεÏá½±Ïαι.77
3.3.7 Conclusions on Aeolic
The forms ÏÏÏá½¹ÏÎ¿Ï âarmyâ and αμβÏοÏην âto errâ provide clear evidence for the development of an anaptyctic vowel âοâ after the liquid in Aeolic dialects. This development is shared by at least Lesbian and Boeotian. Furthermore, ÏÏÏá½¹ÏÎ¿Ï proves that the o-coloring was not conditioned by a neighboring labial sound. The Thessalian evidence is somewhat less straightforward, but all attested forms are compatible with the development established for Boeotian and Lesbian: ÎÏοÏια and ÏεÏÏοÏÎ¿Ï render this likely. We may therefore reconstruct *rÌ¥ > âÏοâ for Proto-Aeolic.
It is important to state this conclusion in clear terms, especially given the discussion of the reflexes of *rÌ¥ in the Aeolic dialects by Parker (2008: 446â447). Parkerâs general tenet is that the Aeolic dialects have no shared innovations: he describes most of the typical Aeolic features as choices made independently by Boeotian, Lesbian and/or Thessalian. However, GarcÃa Ramón (2009) has shown convincingly that Parkerâs arguments are misguided. Not only are there at least two shared innovations between Lesbian and Thessalian, but the three dialects in fact share a bundle of features (often non-trivial choices between alternatives) that cannot be due to language contact at a recent stage, as the dialect regions are geographically non-contiguous.
Now, the development *rÌ¥ > âÏοâ is perhaps the most salient of all common phonological innovations of the Aeolic dialects, especially now that it has been shown that the vowel /o/ regularly follows the liquid only here, not in Mycenaean (cf. chapter 2) or in Arcadian (section 3.4). When Parker states that â*rÌ¥ > Ïο/Ïα is a comparatively late change in various Greek dialectsâ (2008: 447), I agree that *rÌ¥ may well have been retained until the late (or subâ)Mycenaean period in many dialects, including Ionic-Attic (cf. chapters 6 till 9) and West Greek (sections 3.1 and 3.2). However, this does not imply that the change is ânot very important for grouping Greek dialectsâ, as Parker states with a misleading reference to Cowgill. Apart from Boeotian, Lesbian and probably Thessalian, there is no other dialect for which we know for certain that the reflex of *rÌ¥ was âÏοâ. Moreover, it is uncommon in other IE languages for the anaptyctic vowel to develop after the liquid: the only clear example of such a treatment is Celtic, where *CrÌ¥Tâ developed to *CriTâ (cf. OIr. cride âheartâ < *krÌ¥di̯oâ). Given that Thessalian, Boeotian and Lesbian are not geographically contiguous, the only logical conclusion is that the isogloss *rÌ¥ > âÏοâ is an innovation of their common ancestor, which we may call Proto-Aeolic.78
It is difficult, however, to determine the date of this Proto-Aeolic vocalization to âÏοâ more precisely. In my view, it is likely that the change took place before the end of the Mycenaean period: this allows us to understand why the Aeolic dialects did not develop an a-colored reflex, as Ionic-Attic and neighboring West Greek dialects did. Another important common Aeolic innovation, the generalization of âεÏÏι as a general 3rd declension dative plural ending (except in the s-stems), may also have taken place in the Mycenaean period, as I hope to show elsewhere.
3.4 Arcado-Cyprian
Fraenkel (1911: 250â251) was one of the first scholars to explicitly state that the o-colored outcome is regular in Arcado-Cyprian. He adduced the forms ÏαναγοÏÏιÏ, εÏθοÏκÏÏ from Arcadian, and âcypr. ÏλόÏει (â¦) das sich dem Sinne nach mit sonstigem ÏλάÏει decktâ. This thesis was quickly taken up by the handbooks, and it remained the standard view until Morpurgo Davies proposed that the instances of o-coloring were conditioned by a preceding u̯â. In her words, âboth in Arcadian and Cyprian the reliable instances of aR/Ra considerably outweigh those of oR/Ro. This amounts to saying that the data definitely favour the suggestion that aR/Ra and not oR/Ro is the regular treatment of RÌ¥ in these dialectsâ (1968: 808).
Since then, scholars have occasionally doubted that o-vocalism was the only regular outcome in Arcadian and/or Cyprian. For instance, Egetmeyer (2010: 144) remarks that the Cyprian outcome is uncertain because of a lack of clear examples. However, much of the alleged evidence for a-vocalism in both dialects was adduced for incorrect reasons, notably in Morpurgo Davies (1968). In my view, scholars like GarcÃa Ramón (1985) and Haug (2002: 49â67) are right to insist that only the o-colored outcome is regular in Arcadian and Cyprian. Nevertheless, whether the two dialects underwent a common development of *rÌ¥ remains, in my opinion, an open question.
The evidence below has been collected from Morpurgo Davies (1968), Haug (2002) and the dialectal grammar by Egetmeyer (2010: 144â147).
3.4.1 Cyprian: Evidence for o-vocalism
There are five more or less reliable forms with o-vocalism in Cyprian, three of which are attested in glosses ascribed by Hesychius to the dialect of Paphos (εá½ÏÏá½¹ÏÏεÏθαι, κοÏζία, and ÏÏÏοÏá½±), against two forms attested in the syllabary (ka-te-wo-ro-ko-ne, to-ro-su-ta-mo-se). Another form that was previously adduced (θόÏναξΠá½ÏοÏόδιον Hsch.) has no bearing on the discussion.
-
ka-te-wo-ro-ko-ne /kat-ÄÌworgon/ or /kat-ÄÌwrogon/79 âthey besiegedâ (ICS2 217.1, Egetmeyer 2010: 481) is a zero grade thematic 3rd pl. from the root of εἴÏÎ³Ï (Hom. á¼á½³ÏγÏ) âto shut in; keep awayâ. According to the traditional interpretation, this form is an aorist (cf. GEW s.v. εἴÏγÏ).80 Morpurgo Davies (1968) doubted whether the form reflected a zero grade root with the argument that there is no independent evidence for a thematic aorist from this verb, and she followed a suggestion by Schwyzer (1939: 777) to compare it with á¼Î½Ïγον, a thematic pluperfect found in the next line of this inscription (a-no-ko-ne âthey orderedâ ICS2 217.2). However, since á¼Î½Ïγον is clearly an exceptional case and is also attested in Homer, the interpretation of ka-te-wo-ro-ko-ne as a thematic aorist or imperfect must be preferred. Thus, the form serves as reliable evidence for an o-colored reflex.
-
A PN to-ro-su-ta-mo, either /Throsu-dÄmÅ/ or /Thorsu-dÄmÅ/, occurs in an inscription from Paphos that was dated to 750â600â¯BCE (cf. Neumann 2004: 138â139 for the reading). The antiquity of the inscription suggests that we are dealing with a genuine reflex of *rÌ¥ in Cyprian. Egetmeyer (2010: 146) argues that /Throsu-/ is the correct interpretation of the first member, but as Neumann (l.c.) remarks, there is no way to exclude /Thorsu-/.
-
The gloss εá½ÏÏá½¹ÏÏεÏθαιΠá¼ÏιÏÏÏá½³ÏεÏθαι. Î á½±Ïιοι (Hsch.), âto turn around or towardsâ, is mostly thought to derive from a yod-present *âtrÌ¥kw-i̯e/oâ.81 Although scholarly opinion is still divided concerning the assumed equivalence of Cypr. εá½â (as a preverb) and Ion.-Att. á¼Ïιâ,82 the most widely accepted interpretation of εá½ÏÏá½¹ÏÏεÏθαι recognizes in it the root *trÌ¥kwâ underlying ÏÏá½³ÏÏ.83 Morpurgo Davies (1968: 800) claimed that âin the absence of any other evidence a denominative formation on an âo-grade substantive cannot be excludedâ, but this seems highly unlikely: in all other Greek dialects, denominatives from o-grade thematic nouns are in âá½³Ï.84 The most plausible reconstruction remains *âtrÌ¥kw-i̯e/oâ, even if some doubts persist concerning the analysis of εá½â as a preverb. The vowel slot of εá½ÏÏá½¹ÏÏεÏθαι may be analogical, cf. ÏÏá½³ÏÏ.
-
The gloss κοÏζίαΠκαÏδία. Î á½±Ïιοι (Hsch.), a betting reading than κόÏζα,85 was disqualified by Morpurgo Davies (1968: 801, 812) with the remark âbut this is a gloss attested only in Hesychiusâ. If the other available evidence spoke against o-vocalism, this would perhaps be a legitimate way of arguing, but since there is no compelling evidence for a-vocalism in Cyprian (see below), it is best to take the gloss seriously, especially given its remarkable â¨Î¶â©.
-
ÏÏÏοÏάΠá¼ÏÏÏαÏá½µ. Î á½±Ïιοι (Hsch., Hdn.) âflash of lightningâ.86 The related gloss ÏÏοÏÏάνΠÏὴν á¼ÏÏÏαÏήν (Hdn., Hsch., without dialect indication) might well be Arcadian in view of ÎÎ¹Î¿Ï Î£ÏοÏÏαο in an Arcadian inscription (gen.sg., IG V,2 64, 5th c.). The aspirated stop in the gloss ÏÏÏοÏαίΠá¼ÏÏÏαÏαί (Hdn., Hsch.) might be folk-etymological after ÏÏÏá½³ÏÏ. As for the real etymology, Beekes (1987) has convincingly criticized the reconstruction *h2str-h3kwâ âstar-eyeâ, which is not evident semantically and, even worse, does not explain the forms without prothetic vowel or the lacking reflex of *h3.87 Still, since forms with o-vocalism appear precisely in Arcadian and Cyprian, a reconstruction with syllabic liquid might be considered. One would then have to reconstruct a Proto-Greek root *strÌ¥p-Ä, of unknown etymology, with a variant *astrÌ¥p-Ä continued in the Classical form á¼ÏÏÏαÏá½µ and in the epic denominative verb á¼ÏÏÏá½±ÏÏÏ. In this case, the reflex âÏαâ in the Classical Attic form would be difficult to rhyme with the claims made in this book. However, the word may well be of non-Indo-European etymology: the variation between forms with and without prothetic á¼â, and the difference in vocalism between Homeric ÏÏεÏοÏá½µ, á¼ÏÏεÏοÏá½µ and á¼ÏÏÏαÏá½µ (with âreducedâ vocalism) are both typical substrate phenomena (Beekes l.c.; Schrijver 2001: 419). Furthermore, it cannot be excluded that contaminations took place between two or more different pre-forms. In view of this, no conclusions can be based on these forms.88
-
The gloss θόÏναξΠá½ÏοÏόδιον âfootstoolâ (Hsch. θ 647 Latte) at one point received the addition Îá½»ÏÏιοι (thus e.g. GEW), but Latte no longer prints this because he thinks the ethnicon was wrongly taken over from the preceding gloss (cf. Egetmeyer 2010: 147). The word is clearly derived from Myc. to-no /thornos/, alph. θÏόνοÏ, but its pre-form may not have contained *rÌ¥ at all (see chapter 7).89
3.4.2 Cyprian: Evidence for a-vocalism
Morpurgo Davies (1968: 799â801) and Egetmeyer (2010: 145) list several pieces of evidence for a-vocalism, but none of them is compelling.
First of all, as explained in section 1.2, we must leave aside all forms where *r̥ may have been vocalized as part of a Common Greek development, e.g.:
-
Cypr. a-u-ta-ra /autar/ < PGr. *autrÌ¥ (Hom. αá½Ïá½±Ï);
-
Cypr. ka-i-re-te /khairete/ < PGr. *khari̯e/oâ < PIE *ǵhrÌ¥-i̯e/oâ.
Forms for which there is no clear reason to reconstruct a syllabic liquid can also be disregarded:
-
the gloss á¼Î±ÏΠαἷμα. Îá½»ÏÏιοι (Hsch. ε 31), which derives from PGr. *ehar < PIE *h1esh2r (cf. Hitt. Äshar âbloodâ);
-
the verb μάÏÏÏÏ (Morpurgo Davies 1968: 801) appears in the γλῶÏÏαι καÏá½° Ïá½¹Î»ÎµÎ¹Ï in the form ÎÏ ÏÏá½·Ïν. á¼Î¼Î±ÏÏενΠá¼Î»Î±Î²ÎµÎ½. However, the word also occurs in both epic and Lesbian lyric, and has no clear etymology (see section 9.7.2).
Some of the Cyprian forms with âarâ, âraâ look like imports from Ionic-Attic:
-
Personal names in âκÏá½±ÏÎ·Ï are also attested in Ionic-Attic and appear only late in Cyprian; they must therefore be borrowings (see Egetmeyer 2010: 327â330 and already Morpurgo Davies 1968: 800);
-
κάÏÏÏÏιÏÎ Î¸Ï Ïία á¼ÏÏοδίÏÎ·Ï á¼Î½ á¼Î¼Î±Î¸Î¿á¿¦Î½Ïι âfestival of Aphrodite in Amathousâ (Hsch.) is derived from the problematic word καÏÏá½¹Ï, which has this very shape in all Greek dialects where it is attested.
-
The form ÏαÏβεῠis ascribed to Cyprian by the γλῶÏÏαι καÏá½° Ïá½¹Î»ÎµÎ¹Ï (cf. Ruijgh 1957: 163). However, in spite of Bowra (1959), this list of dialectal words is in my view not fully trustworthy as a source for the spoken dialects of Ancient Greek.90 The verb ÏαÏÎ²á½³Ï is frequent in Homer, but also in the tragedians; it is therefore not excluded that the word is of Ionic-Attic origin. The reconstruction of the root as PIE *tergwâ (with ÏαÏÎ²á½³Ï reflecting *trÌ¥gw-eh1â) is possible, but not compelling.91
There are two Cyprian forms containing a sequence â¨Ca-raâ© for which a pre-form with syllabic nasal may be reconstructed:
-
As I will argue in section 9.1.4, the imperative ka-ra-si-ti /grasthi/ âeat!â can be the regular outcome of a pre-form *grnÌ¥s-dhi.
-
The form ta-ra-ka-ma-ta /dragmata/ âsheaves; first fruitsâ (ICS2 318 A III, 2) corresponding to alphabetic δÏάγμαÏα belongs to the root of δÏá½±ÏÏομαι, which did not have ablaut and whose âαâ may reflect a vocalized nasal (see section 9.2.1).
3.4.3 Arcadian: Evidence for o-vocalism
The epigraphic evidence unambiguously proves that Arcadian had an o-colored regular reflex, independent of the preceding consonant. This was already argued clearly by Haug (2002: 60); moreover, two forms that were recently discovered in an archaic Arcadian festival calendar (editio princeps: Carbon-Clackson 2016) must be added to the dossier.92 The forms are discussed in alphabetical order.
-
Arc. βÏοÏÏ [ (Dubois 1988: 43 with n. 212). Morpurgo Davies doubts the Arcadian origin of the form. Dubois could not find the stone in the museum of Dimitsana, but as he remarks (ibid., n. 212, cf. also Haug 2002: 60), âil est peu probable quââ¯il y ait eu dans ce musée beaucoup de pierres errantes éoliennes.â The place of the vowel in βÏοÏÏ can be analogical after the full grade in the forms of comparison, like that of Class. βÏαÏá½»Ï (cf. section 4.3.3).
-
According to Morpurgo Davies (1968), following Chantraine and Wackernagel, the Arc. form εÏθοÏκÏÏ (IG V,2 6.10â11 = Del.3 656) may have been influenced by the older root perfect (Att. δι-á½³ÏθοÏα). However, as Haug (2002: 60) remarks, the classical κ-perfect was normally derived from a middle perfect (cf. Attic á¼ÏθαÏκα from á¼ÏθαÏμαι). Moreover, in Ionic-Attic one never finds intrusion of the o-vowel from the active into the middle perfect. Therefore, I agree with Haug and with Dubois (1988: 44) that εÏθοÏκÏÏ probably implies the existence of an Arcadian middle perfect *εÏθοÏμαι, with an o-colored reflex of *rÌ¥.
-
ÎοÏÏÏ Î»Î¿ÏÎ¿Ï (Dubois 1988: II, 171) is attested on a 3rd c. proxeny decree from Orchomenos. Morpurgo Davies (1968: 794) remarks that the name refers to a person from Achaea and excludes the form as evidence. Haug (2002: 60) prefers to see in ÎοÏÏÏ â the regular development of a zero grade, and Dubois (ad loc.) follows Masson (1972) in seeing in this form an element of the pre-Doric substrate in Achaea. Note, in this context, the Cyprian manâs name to-ro-su-ta-mo (see above) and the Cretan PNâ¯s ÎοÏÏ ÏÏαÏÏÎ¿Ï and ÎοÏÏÏ Ï (Masson 1972, Leukart 1994: 191). It is hard to base any conclusions on this form, because it is a name.
-
Arc. ÏαναγοÏÏÎ¹Ï name of a festival, lit. âall-gatheringâ (IG V,2 3.26 = Del.3 654), also in the month name ÏαναγοÏÏιον (ibid. 3.3), ÏÏιÏαναγοÏÏÎ¹Î¿Ï (ibid. 3.7). The word is now also attested in its expected dialectal form ÏαναγοÏι (Carbon-Clackson 2016) with simple spelling of the geminate resulting from the dialectal development ÏÏ > ÏÏ. Finally, cf. also á¼Î³Î¿ÏÏιÏÎ á¼Î³Î¿Ïá½±, á¼Î¸ÏοιÏÎ¹Ï âgatheringâ (Hsch., without dialect identification). A zero grade reflex is also attested in the form αγαÏÏÎ¹Ï âmeetingâ (IG XIV, 659, lines 12 and 16) found in a Western Ionic colony.93 As a comparison between Arcadian and Western Ionic confirms, the original form was *agrÌ¥-tiâ.94 As Dubois remarks, Eastern Ionic á¼Î³ÎµÏÏÎ¹Ï (Hdt.+; epigraphically in Miletus) must contain the restored root of á¼Î³Îµá½·ÏÏ, while Arcadian ÏαναγοÏÏÎ¹Ï / ÏαναγοÏ(Ï)Î¹Ï and Western Ionic αγαÏÏÎ¹Ï show the etymologically expected zero grade root.
The form αγαÏÏÎ¹Ï was discarded as âdoubtful evidenceâ by Morpurgo Davies (1968: 794), for the reason that it occurs in a âlate inscription, in which the only other dialect formations are ÏÏηÏÏία and its derivativesâ. In her view, it is âquite possibleâ that αγαÏÏÎ¹Ï arose by vowel assimilation from á¼Î³ÎµÏÏιÏ, but this is clearly an ad hoc assumption.95 The fact that both αγαÏÏÎ¹Ï and ÏÏηÏÏια seem to denote institutions peculiar to this colony suggests that the form αγαÏÏÎ¹Ï preserves older morphology. Moreover, as Dubois (1995: 86) remarks, αγαÏÏÎ¹Ï shows the expected result of âÏÏâ in Western Ionic and cannot therefore be a Koine form. Finally, as Haug (2002: 60) remarks, an o-grade root is excluded in an abstract noun in âÏιÏ.
-
The Arcadian form of the word for âmaleâ (with ÏÏ < ÏÏ) has been known for a longer time in a form with crasis, ÏοÏÏενÏεÏον (Mantinea, 5th c., Dubois II, 94â¯ff. and 105). It remained unknown, however, whether this form had resulted from *Ïο αÏÏενÏεÏον or *Ïο οÏÏενÏεÏον. This question may now finally be resolved after the appearance of οÏεν, with single spelling of the geminate, on a newly published festival calendar (Carbon-Clackson 2016). The form αÏÏενα (Tegea, 4th c.) must be a Koine form: see below.
-
As for Arc. ΣÏοÏÏαο, epithet of Zeus, see the discussion of the Cyprian gloss ÏÏÏοÏá½± above. It remains uncertain whether the pre-form contained *rÌ¥.
-
Arc. ÏεÏοÏÏÎ¿Ï (Dubois 1988: 42â43). The form is attested twice as a gen. sg. ÏεÏοÏÏÎ±Ï and probably once in a broken attestation as a nom. ÏεÏ]οÏÏα. As a manâs name, ΤεÏαÏÏÎ¿Ï is attested only once. As with Attic Ïá½³ÏαÏÏοÏ, Arcadian ÏεÏοÏÏÎ¿Ï cannot be explained by analogy, because the cardinal form is ÏεÏÏεÏεÏ.96 I cannot accept the reasoning of Morpurgo Davies (1968: 795) that the single âÏâ (from *âtu̯â) in ÏεÏοÏÏÎ¿Ï can only be explained from an earlier form *Ïá½³ÏÏοÏÎ¿Ï or *Ïá½³ÏÏαÏοÏ. As argued in section 2.6, *âtu̯â was simplified before *rÌ¥ in this word prior to the vocalization of the syllabic liquid.
3.4.4 Arcadian: Evidence for a-vocalism
As Haug (2002: 59â61) makes clear, the counterevidence to a regular vocalization *rÌ¥ > âοÏâ in Arcadian merely consists of the forms δαÏÏμα, γÏαÏÏ and ÏÏÏαÏαγοÏ. There are two possible ways to explain these forms: either they are non-dialectal words, or they have âαÏâ or âÏαâ for some other reason.
-
As was already remarked e.g. by Ruijgh (apud Morpurgo Davies 1968: 813), ÏÏÏαÏÎ±Î³Î¿Ï could well be a borrowing from Doric. He compares the military term Att. λοÏαγόÏ, where the long âá¾±â excludes a native Ionic-Attic word, and which is generally accepted to be a Doric borrowing.
-
The root of γÏá½±ÏÏ has a-vocalism in all Greek dialects, except in the agent noun γÏοÏÎµá½»Ï âscribeâ attested in various dialects, mainly on the Peloponnese (see section 9.2.2 for a discussion of the details). Arcadian has γÏαÏε̣α̣ (IG V,2 343.31â32), ÏÏ Î³Î³ÏαÏο̣ν̣ (IG V,2 6.53), and γ]ÏαÏÎ·Ï (IG V,2 8.4), whereas γÏοÏÎµá½»Ï is only known from Koine texts.97 The a-vocalism of γÏá½±ÏÏ could be the reflex of a vocalized nasal (section 9.2.2).
-
It is hard to utilize δαÏÏμα as evidence: as a word designating a monetary unit, it may have easily been borrowed from another dialect. Indeed, the same form is found in the neighboring West Greek dialect of Elis, as well as on Crete. Moreover, the Boeotian dialect of Thespiae also offers instances of δαÏÏμα (Roesch, IThesp. 38 and 39), which cannot have the genuine reflex of *rÌ¥ in Aeolic, as we have seen above.98
-
The form αÏÏενα âmaleâ (Lex sacra from Tegea, 4th c., Dubois I, 80; II, 34â¯ff.) cannot be used as evidence, because the genuine Arcadian form with âÏÏâ < âÏÏâ is reflected in ÏοÏÏενÏεÏον and οÏεν (see above). Consequently, αÏÏενα must be a literary or Koine form.99
3.4.5 Conclusions on Arcado-Cyprian and Achaean
As Haug (2002) has convincingly shown, Morpurgo Davies was mistaken in positing a regular vocalization *rÌ¥ > âαÏâ, âÏαâ for Arcadian. The forms ÏαναγοÏÏÎ¹Ï and ÏεÏοÏÏοÏ, and now also ÏαναγοÏι and οÏεν in the festival calendar, clearly show that the o-colored reflex was regular in this dialect also in non-labial environments. The o-vowel of εÏθοÏκÏÏ further supports for this conclusion. The situation in Cyprian is a bit less clear, but here too, the gloss κοÏζία and the verbal form ka-te-wo-ro-ko-ne point to regular o-coloring. The gloss εá½ÏÏá½¹ÏÏεÏθαι and the PN to-ro-su-ta-mo support this conclusion.
As for the vowel slot, Arcadian ÏεÏοÏÏÎ¿Ï can only point to a regular and unconditioned vocalization to âοÏâ. The same outcome is found in ÏαναγοÏÏÎ¹Ï (ÏαναγοÏιÏ), οÏεν (ÏοÏÏενÏεÏον) and ΣÏοÏÏαο, although it must be noted that ΣÏοÏÏαο has no clear etymology, that οÏεν is a unique example for *rÌ¥â in word-initial position, and that the vowel slot in ÏαναγοÏ(Ï)Î¹Ï may have been influenced by that of the full grade form. In order to determine the regular vowel slot in Cyprian, we have to rely on glosses in view of the nature of the Cyprian syllabary. The evidence is meagre: κοÏζία points to âοÏâ, and εá½ÏÏá½¹ÏÏεÏθαι to âÏοâ, but the latter form might be analogical.
Although the evidence is less extensive, the situation in Arcado-Cyprian is similar to that in Mycenaean. There is no clear evidence for an a-colored outcome, and there is some reliable evidence for o-vocalism.100 In view of these similarities, one could be tempted to reconstruct a Proto-Achaean vocalization *rÌ¥ > âorâ (the Arcadian reflex was clearly âοÏâ), but we have to be careful. While a vocalization to âroâ can be excluded for Mycenaean, this dialect may have preserved *rÌ¥ (chapter 2), and the forms tu-ka-á¹áº¡-ṣỠand ạ-na-qo-ta perhaps support this (cf. section 2.4). Furthermore, a Cyprian outcome âroâ cannot be completely excluded on the basis of our evidence. It thus remains unclear whether Mycenaean and Cyprian had an outcome âorâ in the first place.
If Mycenaean did preserve *rÌ¥, the Arcadian reflex âοÏâ may have come into being in the sub-Mycenaean period, before speakers of West Greek dialects established themselves on the rest of the Peloponnese. The Cyprian reflex (if it was indeed âorâ) may then be an isogloss with Arcadian, but since a development to âorâ is phonetically more natural than a development to âroâ, an independent vocalization in both dialects is difficult to exclude.
3.5 Pamphylian
The view that Pamphylian, like Cretan, could undergo liquid metathesis has been codified in Brixheâs grammar (1976: 61â63). He adduces five items as evidence:
-
PN ÎÏοÏδιÏÎ¹Î¹Ï Ï, ΦοÏδιÏÎ¹Î¹Ï Ï ~ Hom. á¼ÏÏοδίÏη, Cret. ÎÏοÏδιÏα;
-
PN ΠοÏÏοÏα, corresponding to a hypothetical Ionic name *Î ÏοÏá½½ÏÎ·Ï âFaceâ;101
-
Î ÏειιαÏ, Î ÏειÏÏ Ï, Î ÏεεÏÏ ~ Ion. Î á½³Ïγη;
-
ÏεÏÏ- ~ Ion.-Att. ÏÏá½¹Ï, Hom. ÏÏοÏá½·, Cret. ÏοÏÏι;
-
ΣÏÎ»ÎµÎ³Î¹Î¹Ï Ï, ÎÏÏÎ»ÎµÎ³Î¹Î¹Ï Ï, supposed to derive from a pre-form *slegâ with t-epenthesis ~ non-Pamph. Σέλγη.102
In Brixheâs view, these five forms show that liquid metathesis may work in both directions, and that the phenomenon affected not only the outcome of *rÌ¥ or *lÌ¥, but also other sequences consisting of a vowel and liquid.
Upon closer consideration, however, these claims appear to be unfounded. The toponym Pamph. Î Ïειια < *Î Ïεγα alternates with Î á½³Ïγη in Ionic, and the ethnic designation ΣÏÎ»ÎµÎ³Î¹Î¹Ï Ï denotes an inhabitant of the town called Σέλγη in Ionic. Rather than proving metathesis, the alternations suggest that the Anatolian place names contained syllabic liquids in the donor language, and that these sounds were vocalized in two different ways in the Greek dialects in Asia Minor. This scenario is confirmed by the reflexes of the self-designation of the Lycians, not cited by Brixhe in this context. The Lycian form trmÌmili- probably represents /trÌ¥mili-/, at least originally.103 The Ionic counterpart is ΤεÏμίλαι (Hdt. 1.173, 7.92, a form also attested epigraphically in Pisidia), but Pamphylian also shows the form ΤÏÎµÎ¼Î¹Î»Î±Ï in a PN derived from the ethnonym. This exactly mirrors the distribution found in Pamph. Î Ïειια ~ Ion. Î á½³Ïγη and in Pamph. ΣÏÎ»ÎµÎ³Î¹Î¹Ï Ï ~ Ion. Σέλγη. Thus, certain sounds in names of Anatolian origin are reflected as âεÏâ, âελâ in Ionic, but as âÏεâ, âλεâ in Pamphylian. Moreover, since an Anatolian pre-form with *rÌ¥ is ascertained in Lycian trmÌmiliâ, we may hypothesize that the other toponyms were also borrowed from a language with syllabic lÌ¥ and rÌ¥. I propose the following scenario.104 When the borrowing into Pamphylian and Ionic took place, inherited PGr. *lÌ¥ and *rÌ¥ had already vocalized in these dialects. The sounds lÌ¥ and rÌ¥ from the donor language were initially rendered as [Él], [Ér] in Ionic, but as [lÉ], [rÉ] in Pamphylian. Subsequently, the shwa in these renderings was identified as the phoneme spelled â¨Îµâ© in both dialects.105
Leaving aside these ethnonyms and toponyms, the potential evidence for inherited Greek *rÌ¥ consists of the forms ÏεÏÏ-, ÎÏοÏδιÏÎ¹Î¹Ï Ï, and ΠοÏÏοÏα. As we have seen in our discussion of the Cretan material, it is conceivable that both ÏοÏÏι and ÎÏοÏδιÏα derive from a pre-form with syllabic liquid; and in chapters 6 and 7 we will encounter metrical evidence that supports this claim. This means that the outcome of *rÌ¥ in Pamphylian (at least after labials) was âοÏâ.106 However, in this way the form ÏεÏÏâ for ÏÏοÏá½· would remain unexplained. Assuming liquid metathesis from PGr. *preti, as per Brixhe, has the disadvantage that an e-grade form of this preposition is not directly attested anywhere else in Greek. Bechtel (1921â1924, II: 820) proposed that ÏεÏÏι reflects *porti < *prÌ¥ti in proclitic contexts, which deserves consideration. Wyatt (1978) suggested that ÏεÏÏι might be a cross between ÏεÏá½· and ÏοÏá½·.
In sum, there is not enough evidence to draw firm conclusions about the outcome of *rÌ¥ or *lÌ¥ in Pamphylian. If my scenario for the origin of Homeric muta cum liquida scansions (chapters 6 and 7) is correct, ÎÏοÏδιÏÎ¹Î¹Ï Ï and ΠοÏÏοÏα are suggestive of a development *rÌ¥ > Î¿Ï at least after labial consonants. However, the difference in vocalism between ÏεÏÏâ and ΠοÏÏοÏα would still remain problematic (there is no compelling reason to assume that the former reflects *preti with liquid metathesis); in general, the inherited material is too scanty to allow for a definite conclusion. It is clear, on the other hand, that syllabic liquids in words borrowed from Lycian and related Anatolian languages are reflected as âÏεâ and âλεâ in Pamphylian.
3.6 Conclusions
In chapter 2, it appeared that either âorâ or preserved ârÌ¥â is the regular reflex of *rÌ¥ in Mycenaean. A scrutiny of the epigraphic evidence for the first millennium dialects (with the exception of Ionic-Attic dialects) has yielded the following results:
-
Arcadian ÏεÏοÏÏÎ¿Ï âfourthâ shows that this dialect has an o-colored reflex even in non-labial environments, and that the anaptyctic vowel regularly develops before the liquid. As far as the vowel color is concerned, this conclusion is corroborated by the forms ÏαναγοÏ(Ï)Î¹Ï âassemblyâ and οÏεν / ÏοÏÏενÏεÏον âmaleâ.
-
In Cyprian, the verbal form ka-te-wo-ro-ko-ne âthey beleagueredâ and the personal name to-ro-su-ta-mo-se display a regular o-colored reflex, the latter in a non-labial environment. The vowel color is supported by the forms κοÏζία âheartâ and εá½ÏÏá½¹ÏÏεÏθαι attested for Paphos in Hesychius, and there is no secure evidence for an a-colored outcome. However, the regular vowel slot remains uncertain in view of the orthographical ambiguity of the syllabary.
-
The Aeolic dialects have regular o-coloring and develop the vowel after the liquid. This appears most clearly from Lesbian and Boeotian. The conclusions for Lesbian rely heavily on evidence from literary sources, but this is relatively clear-cut and is backed up by evidence from inscriptions. The Thessalian evidence is less conclusive: ÏεÏÏοÏÎ¿Ï âfourthâ may point in the same direction, but numeral forms are generally difficult to evaluate because analogical remodeling may have played a role.
-
Central Cretan did not undergo a liquid metathesis, as is widely believed, but developed the vowel before the liquid. The regular reflex in Cretan dialects is âαÏâ, and probably âοÏâ after a labial consonant.107 The situation on Thera (and in its colony Cyrene) seems to be similar, but the evidence is slight and consists mainly of personal names.
-
The situation in most other West Greek dialects is similar to that in Ionic-Attic (general a-coloring), but the precise details may differ per dialect, and the evidence is often too scanty to allow for solid conclusions. In Elis (βÏαÏάναν âladleâ, βÏαÏάνει âstirsâ in Hesychius) and Syracuse (middle pf. á¼Î¼Î²ÏαÏαι âis fatedâ, aor. á¼ÏÏÎ±Î´ÎµÏ âfartedâ) there is some evidence for âÏαâ as a regular reflex. The divergence between Central Cretan on the one hand, and the dialects of Elis and Syracuse on the other, shows that *rÌ¥ had not yet vocalized in Proto-West Greek. The situation in other West Greek dialects could benefit from a more detailed investigation.
-
There is no clear evidence for the outcome of PGr. *r̥ in Pamphylian, nor any compelling evidence for liquid metathesis in this dialect.
More generally, the results of this chapter can be summarized as follows. First of all, not all Greek dialects developed a vowel after the liquid, nor was there a fluctuation between both positions. Rather, the evidence suggests that in each individual dialect, there was only one regular position where the anaptyctic vowel developed. It developed after the liquid in Proto-Aeolic and some West Greek dialects, but before the liquid in Central Cretan and Arcadian. Secondly, Aeolic, Arcadian and Cypriot have unconditioned o-colored reflexes, but Central Cretan probably shows a conditioned outcome (Î¿Ï after labial sounds, Î±Ï elsewhere).
The inscriptional evidence for Ionic-Attic hardly adds anything to the picture obtained from literary sources, and will therefore not be treated separately in this chapter. In Western Ionic, the development of the syllabic liquids was identical to that in the rest of Ionic-Attic (cf. del Barrio 1991). The Euboean colonies in Italy yield the form αγαÏÏÎ¹Ï âassemblyâ (Naples), which probably reflects a zero grade root, whereas the literary Ionic-Attic form á¼Î³ÎµÏÏÎ¹Ï âmustering of an armyâ (Hdt.) was rederived from the verb with an e-grade root. See the discussion of Arc. ÏαναγοÏ(Ï)Î¹Ï in section 3.4.3. As for Attic, Threatte (1980) has no separate treatment of the syllabic liquids. An exceptional instance where Attic inscriptions add to the literary evidence is ÏαÏÏÏαι (inscr.) beside ÏÏάξαι âto fence in, fortifyâ (mss. of literary authors); it will be discussed in section 9.2.3.
Cret. καÏÏÎ¿Ï âyield, revenueâ could be the regular reflex of its pre-form, PGr. *krÌ¥póâ, but since this word shows âαÏâ in all dialects where it is attested, its evidential value is limited.
In the words of Bechtel (1921â1924, II: 710â711), âIn einigen Wörtern und Wortfamilien werden die Lautgruppen Ïα, Ïο zu αÏ, Î¿Ï umgestellt. Wie weit dieser Vorgang rein lautlicher Natur sei, wie weit analogische Wirkungen ihn begünstigt haben, kann nicht immer entschieden werdenâ Cf. also Thumb-Kieckers (1932: 160), who think that the same phenomenon is found in Argolic, Elis, Pamphylian, and Arcado-Cypriot.
For the compounded names in âμοÏÏοÏ, which are well-attested in Cretan and correspond to âβÏοÏÎ¿Ï in most other dialects, see section 3.1.2 (with further discussion). Another piece of evidence cited by Hirt is Pamphylian ÏεÏÏι, which allegedly reflects *preti and is related to ÏÏοÏá½·, ÏÏá½¹Ï. However, the dialectal affiliations of Pamphylian are unclear, and the same holds for the regular reflex of *rÌ¥ in this dialect (see section 3.5).
OâNeil (1971: 43â44) posits a liquid metathesis in Central Cretan only before dental or velar stops, but not before labial or (original) labiovelar stops. This is phonetically unmotivated and also contradicted by αÏοÏÏαÏεν âto run awayâ.
The Pan-Greek a-vocalism of γÏá½±ÏÏ speaks against a reconstruction PIE *grÌ¥bh-e/oâ. Moreover, there is evidence for an o-grade in nominal formations like γÏοÏÎµá½»Ï âscribeâ, and it cannot be excluded that γÏá½±ÏÏ obtained its vowel slot from a (no longer existing) verbal form with *grephâ or *grophâ. See section 9.2.2.
The form καÏÏÎ¿Ï could also be regular from *krÌ¥póâ, but it must be conceded that this word has the same form in all dialects where it is attested. As for δαÏÏμα, δαÏÏνα (if from PGr. *drÌ¥khmnÄ), there are various problems in the reconstruction of this word; cf. below on the dialect of Elis.
In Van Beek 2017b, I have argued that the δ in âάδâ can be due to voicing of an occlusive after an accented Proto-Greek syllabic nasal. Cf. also Olsen (1989).
Comparable epigraphic forms from other West Greek dialect areas are aor. αÏοÏÏÏαÏαι (Delphi, CID 2:34, col. II, 31; 4th c. BCE) and εξÏÏÏαÏεÏαι (SEG 30:380, no. 6, l. 1, Tiryns, ca. 600â550â¯BCE?). There is also evidence from literary sources: in Aristophanes (Ach. 788), ÏÏá½±ÏÏ for ÏÏá½³ÏÏ is reputed to be Megarean, and the form is also well-attested in Pindar and perhaps in Theocritus (á¼ÏÏαÏε Theoc. 3.16, but with v.l.). Moreover, Pindar uses both ÏÏá½±ÏÏ and ÏÏá½³ÏÏ, and á¼ÏÏαÏον occurs at Theoc. 2.147 (with v.l. á¼ÏÏεÏον).
See Letoublon & Lamberterie (1980: 324â325) for further discussion, also on the aspectual status in PIE of formations like *dhrǵh-e/oâ. Willi (2018: 351â355) now argues against the antiquity of zero-grade thematic root presents in Greek, and in Indo-European more generally.
The existence of ÏÏοÏι in Argolic is doubtful (cf. Wyatt 1978: 89 n. 1).
However, Mycenaean po-si could also represent /porsi/ or /prÌ¥si/, and Arcadian ÏÎ¿Ï could be the regular reflex of *ÏοÏÏ before a consonant. To my knowledge, this has not been noted before.
And possibly also orthotonic *préti, but that is irrelevant here.
Kloekhorst (EDHIL, q.v.) already reconstructs the Hittite form as PIE *prÌ¥ti, directly comparing Cretan ÏοÏÏι but without accounting for Ionic-Attic ÏÏá½¹Ï; the analysis proposed here and in chapter 7 may justify this idea.
In section 7.2.8, I argue that Aphrodite must be an inherited Greek epithet of the planet Venus (the evening and morning star), and tentatively propose to reconstruct PGr. (or common Greek) *aphrÌ¥-dÄ«tÄ âwho appears forthwithâ (at sunset). The reconstructed compound consists of the precursor of the adverb á¼ÏÎ±Ï âforthwithâ and a derivative in âtoâ of the PIE root *dih2â âappearâ, otherwise preserved in the Homeric aorist δέαÏο âappearedâ. On the Pamphylian forms ÎÏοÏδιÏÎ¹Î¹Ï Ï, ΦοÏδιÏÎ¹Î¹Ï Ï, see section 3.5.
The grammarian Orion (5th c. CE) cites the fragment as á¼Î´Îµá½·Î¼Î±Î¼ÎµÎ½ á¼ÏÏία μοÏÏοί. If the lectio difficilior á¼ÏÏία is the genuine Callimachean form, it would show the common dialectal change ε > ι before a vowel and a different accent (secondary, or directly from *á¼ÏÏέα?).
Masson concludes (1963: 221): â⦠on ne saurait plus affirmer comme jadis que μοÏÏá½¹Ï est une forme exclusivement éolienne, soit chez Callimaque, soit dans lââ¯onomastique. En effet, lââ¯existence des formes de noms propres en dorien et au nord-ouest assure que μοÏÏá½¹Ï nââ¯est pas un simple doublet de *μ(β)ÏοÏá½¹Ï, βÏοÏá½¹Ï, qui comporterait lui aussi un traitement éolien à partir dââ¯un modèle i.-e. *mrÌ¥tóâ, mais avec Î¿Ï au lieu de Ïο. La forme correspond plutôt à un i.-e. *mórtoâ, avec vocalisme o de la racine *merâ.â Massonâs judgment is followed by DELG (s.v. μοÏÏá½¹Ï) and was already anticipated in Boisacq 1916 and GEW (both s.v. βÏοÏá½¹Ï).
Indo-Iranian has three forms for âmortalâ: (1) Ved. márta- and OAv. (hapax) maÅ¡Ìaâ < PIIr. *mártaâ, (2) OAv. marÉta- < PIIr. *martáâ, and (3) Ved. mártyaâ, Av. maÅ¡Ìiiaâ, OP martiyaâ < PIIr. *mártiaâ (cf. EWAia s.vv. MAR and mártaâ). Furthermore, Ved. mrÌ¥tá- and Av. mÉrÉtaâ mean âdeadâ, not âmortalâ. However, since Indo-Iranian preserves the verbal root marâ âto dieâ, it cannot be excluded that at least some of these formations were later derivations.
Katz (1983) argues that Finno-Ugric borrowings from Indo-Iranian point to a pre-form (early) PIIr. *mértÉâ (where PIIr. *É notes the outcome of PIE *o in closed syllables), to be equated with (later) PIIr. *márta-. This would imply that PIIr. *mártaâ < PIE *mértoâ cannot be directly compared with the putative Proto-Greek *mórtoâ assumed by Masson. See further the discussion in Mayrhofer, EWAia s.v. mártaâ.
The Aetolian attestation cited by Masson (1963: 220) is found in an inscription from Egypt, and refers to an officer serving under Ptolemy Philopator (reigned 221â205â¯BCE). The same person is mentioned by Strabo and Polybius. Even if this relatively late piece of evidence is taken into account, one wonders whether enough is known about reflexes of the syllabic liquids in Aetolian to accept Massonâs conclusion that PGr. had a separate form *mórtoâ.
It is, of course, impossible to establish the dialectal provenance of μοÏÏá½¹Ï in Callimachus with certainty. It is also difficult to draw a conclusion from the gloss μοÏÏοβάÏινΠá¼Î½Î¸ÏÏÏοβάÏιν ναῦν (Hsch.), in view of the absence of a dialect identification.
For analogical âοÏâ in Lesbian, cf. Alc. á¼Î¼Î¼á½¹Ïμενον âhaving as a shareâ corresponding to Ion.-Att. εἵμαÏÏαι.
According to Klingenschmitt (apud LIV2 s.v. *merâ), á¼Î¼Î¿ÏÏεν reflects an older middle in *âto that was reinterpreted as an active form.
For the outcome of *lÌ¥ in Cretan, see section 10.6. The conditioning of the distribution between aâ and o-vocalism in Cretan could be challenged by the PNâ¯s ÎοÏÌ£ÏÏ Ï (IC II, 23.37 and 53, Polyrrhenia, dated between the 3rd and 1st c. BCE) and ÎοÏÌ£Ï ÏÏαÏÏÏ (IC II, 13.7, Elyros, 2nd c. BCE). But in Massonâs view (1972: 292, accepted by Leukart 1994: 191), these names with ÎοÏÏÏ â are an âélément ⦠du substrat pré-dorien ou âachéenâ en Crèteâ. That would presuppose, however, that Mycenaean (or its continuation in the sub-Mycenaean period) had an o-colored reflex also in a non-labial environment, which is possible (cf. the Arcadian reflex in ÏεÏοÏÏÎ±Ï ). On the possibility that Myc. PN to-si-ta reflects /ThrÌ¥sitÄs/ vel sim. < *dhrÌ¥siâ, see section 2.3.1.
A similar conditioned reflex has been proposed for Mycenaean and Arcado-Cyprian (e.g. Morpurgo Davies 1968, see section 3.4 below).
For North-West Greek, see Méndez Dosuna (1985); for the colonies in Magna Graecia, see the various dialect grammars by Arena and Dubois.
That âαÏâ was regular in Theran was already suggested by Bechtel (1921â1924, II: 534 and 556).
These forms show that Theran underwent the development âÏÏâ > âÏÏâ. The âÏÏâ found in ÎαÏÏικÏαÏÎ·Ï on another Theran inscription is probably a Koine form. Generally speaking, forms with ÎαÏÏιâ may replace older forms with *ÎεÏÏιâ, as in Hom. ÎεÏÏίλοÏοÏ.
As a second member, âκαÏÏâ is perhaps found in Îακ[α]ÏÏÏÏ (IG XII,3 1324).
See chapter 4 for further details.
ÎÏαÏÎ·Ï (2â¯Ã, 3rd c. BCE and later), âκÏαÏηÏ, (frequent in all periods), ÎÏαÏÏ â (frequent from the middle of the 4th c. BCE, Dobias-Lalou p. 35), as a simplex ÎÏαÏÏν and ÎαÏÏÏν (both 3rd c. BCE and later), ΣÏÏαÏοâ (SEG 20.735, Dobias-Lalou p. 14) and âÏÏÏαÏοÏ, á¼ÏÏαλέα (4thâ3rd c., CIG 5155 and 3rd c., SEG 9.92).
Frequent from the 4th c. BCEâ2nd c. CE. Bechtel (1917: 256) could only ascribe it to the Imperial period.
In SEG 9.45, 48 (5th c. BCE) and SECir. 244 (4th c. BCE).
Attested in two lists of temple servants, around the beginning of the CE. The sequence âαÏâ is also found in the festival name ÎαÏνεια, as attested in the PNâ¯s ÎαÏÎ½Î·Î¹Î±Î´Î±Ï (4thâ3rd c.), ÎαÏÎ½Î·Î±Î´Î±Ï (4th c.), and ÎαÏÎ½Î·Î´Î±Ï (highly frequent from the 4th c. onwards); for attestations see Dobias-Lalou (2000: 49). The festival belongs to the Laconian heritage of Cyrenaean, but it is unclear whether âαÏâ reflects a syllabic liquid in this word.
Names with ÎÏαÏιâ are attested sporadically in other dialects: ÎÏαÏιÏÏÎ¹Î´Î±Ï (IG V,1 1385.22, Thuria, 2nd c. BCE), ÎÏαÏι-Î´Î·Î¼Î¿Ï (Erythrae, No. 57, 5thâ4th c. and No. 60, early 3rd c. BCE, cited from McCabe, Erythrai inscriptions, text and list).
A first member ÎÏαÏαιâ is attested in inscriptions from various regions. The name ÎÏαÏÎ±Î¹Î¼á½³Î½Î·Ï occurs in Athens, Euboea, Ionia, and in an early example (SEG 22.345, 6th c.) that is perhaps from an Achaean colony in Magna Graecia; ÎÏαÏÎ±Î¹Î²Î¹Î¿Ï occurs on Delos. Cretan has ÎαÏÏÎ±Î¹Î´Î±Î¼Î±Ï (Bile 1988: 183 n. 133) with the expected reflex âαÏâ (contrast Theran ÎαÏÏιδαμαÏ; the Cretan form with âαιâ is due to a specifically Epic metrical lengthening). In section 5.2.11, I propose that ÎαÏÏιâ < *krÌ¥th1-iâ is the old compounding allomorph of καÏÏεÏá½¹Ï, and that the latter reflects *ḱrÌ¥th1-róâ.
For further attestations of κάÏÏÏν, see LSJ s.v. and Forssman (1980: 194 n. 77).
See section 5.2.1.
The reflex âαÏâ was perhaps also regular in Argolic, given forms like ÏαÏÎ¾Î¹Ï (on which see section 3.2.4 and 9.2.3).
I have found no relevant examples in the evidence for non-Attic vase inscriptions (Wachter 2001).
Also attested as ῥαÏάνανΠÏοÏύναν (Hsch.), without dialect identification, but clearly not from Ionic-Attic.
As argued by Forssman (1980), in Ionic-Attic this root may be reflected in Homeric á¼ÏÏÏ âto be lostâ < *u̯ert-i̯e/oâ. The verb is attested in many dialects (in Elean as ÏαÏÏÏ, with secondary lowering of er).
All the relevant inscriptions are dated to slightly before or after 500â¯BCE, so it is impossible to tell whether the form with âμâ or âνâ is older.
The noun καÏÏá½¹Ï âharvestâ is also attested twice in Elis, but it has the same form in all dialects where the word appears. Therefore, we cannot be certain that it reflects PGr. *krÌ¥póâ.
With Lejeune (1972: 76) and against Nieto Izquierdo (2008: 381â382), I prefer to view Arg. γÏαθμα and γεγÏÎ±Î¸Î¼ÎµÎ½Î¿Ï as due to dissimilation of *âphmâ at a morpheme boundary. As Nieto Izquierdo shows, attestations of the Koine form γÏάμμα are later than those of γÏαθμα, γεγÏÎ±Î¸Î¼ÎµÎ½Î¿Ï and γÏαÏÏμα.
The problems are clearly stated in Ruijgh (1961: 194). Cf. generally on these issues Hooker (1977) and Bowie (1981).
A grammar of Thessalian by GarcÃa Ramón and Helly is still in preparation. Vottéro (1998, 2001) announced the publication of a book on the phonetics and phonology of Boeotian, but to my knowledge, this has not yet appeared.
The Arcadian form ενÏοÏÎ¿Ï has recently come to light on an archaic festival calendar (Carbon-Clackson 2016).
Thess. ÏεÏÏÎ±Î³Î¿Ï Î½Î¿Ï (Larisa, late 3rd c.), corresponding to Class. ÏεÏÏάγÏÎ½Î¿Ï ârectangleâ, may be ascribed to Koine influence.
Bechtel (1921â1924, I: 242â243), Thumb-Scherer (1959); Blümel (1982); Hodot (1990).
E.g. ÎÏοÏÏ Î»Î»Î¿Ï (IG VII, 1908, Thespiae, 450â400â¯BCE). In the overview of Boeotian characteristics by Van der Velde (1929), the attestations of these forms are presented per locality; see also GarcÃa Ramón (1975: 62â63). A third form usually mentioned in this connection is εÏοÏÎ¹Ï (plus names in ÎÏοÏοâ, corresponding to Ionic-Attic á¼ÏαÏοâ), but this word does not derive from a pre-form with *rÌ¥, and the difference in vocalism must have another cause. The alleged PN ÎÏ]οÏÎ¹Î¿Ï ÏÏÏοÏÎ¿Ï is based on a false reading (see Masson 1972: 293).
This is the 3pl. pf. mid. of a verb ÏÏÏοÏÎµÏ Î¿Î¼Î±Î¹, with the athematic ending âαθη < *âαÏαι (with secondary âθâ and monophtongization of αι).
This was not a necessary development, however: Ionic-Attic preserves Ïá½³ÏαÏÏοÏ, ÏέμÏÏοÏ, á¼ÎºÏÎ¿Ï while also having ÏεÏÏαâ, ÏενÏαâ and á¼Î¾Î±â (with spread of the linking vowel âαâ) in compounds. On Homeric Ïá½³ÏÏαÏοÏ, see section 6.8.4.
Cf. also GarcÃa Ramón & Helly (2007: 305â306).
In my view, this objection is not cogent. As GarcÃa Ramón himself remarks, ÎεÏÏÏ Ï is a substantivized feminine âthe bold oneâ, âHer Boldnessâ of the archaic type ἰθῡÌÏ (f.) âcourseâ beside á¼°Î¸á½»Ï (adj.) âstraightâ (see Lamberterie 1990: 887â888). If the u-stem adjectives had root ablaut in Proto-Greek (see section 4.1.1), this substantivized form may have been derived from the full grade stem at an early date, before the adjective generalized the zero grade reflex.
âlässt sich der o-Vokalismus bei der Vertretung von *rÌ¥ als nicht durch die phonetische Umgebung bedingt erkennen.â (GarcÃa Ramón 2007c: 106).
In the words of Bowie (1981: 177), the lexicon of Sappho and Alcaeus âshares the characteristics and components of the poetic dictions of the other early Greek poets, both epic and lyricâ.
Bowie is reluctant to explain words that occur both in Lesbian poetry and in Homer as epicisms. The fact that a word is shared by Lesbian poetry and the epic language may mean two things. Either the word is inherited from an earlier, Common Greek poetic language, or one of the poetic languages borrowed the word from the other.
In what follows, fragment numbers refer to Voigtâs edition of Sappho and Alcaeus (Voigt 1971).
For instance, the productive epic adjective suffix âÎ±Î»á½³Î¿Ï (cf. section 4.2.2) is found in á½ÏÏαλέÏÏ (Sapph. 44.11), and Alcaeus is fond of á¼ÏÎ³Î±Î»á½³Î¿Ï âpainfulâ.
The form ]βÏαÏη[ in Alc. 300.9 (cited by OâNeil 1971: 24, but of unclear interpretation) need not belong here: it may be from a completely different lexeme, e.g. that of Hom. á¼Î²ÏαÏε âresoundedâ.
Perhaps also in ]ν ÏÏÏοÏ[ (Alc. 300.1).
Note also á¼á½¹Î»Î»ÎµÎµÏ (= Epic á¼Î¿Î»Î»á½³ÎµÏ).
On á½Î½Î¿Î¹ÏÎ¿Ï (Sapph.), see the next section.
On the relation between middle-passive and active perfect forms and the replacement of active perfects, see Van Beek and Migliori (2019).
The vowel slot of thematic aorists like á½Î½á½³ÏÏοÏε may, of course, be analogical.
Bowieâs analysis, however, is misguided to some extent by OâNeilâs (1971) poor linguistic treatment of the evidence.
On Aeol. η corresponding to Ionic ει in prevocalic position, see Slings (1979; p. 251 n. 36 on á½Î½Î·Î±Ï).
Ruijgh (1961) proposed that the regular Lesbian (and also Achaean) outcome of *rÌ¥ in word-final position was âοÏ, adducing ἦÏÎ¿Ï âheartâ and á½Î½Î¿Î¹ÏÎ¿Ï âdreamâ (Sapph. fr. 63.1) as examples. In fact, the attestation of ἦÏÎ¿Ï in Lesbian (Alc. fr. 6.20) is highly uncertain (cf. the edition by Voigt), so that only the Sapphic form á½Î½Î¿Î¹ÏÎ¿Ï would remain as an indirect piece of evidence for the Lesbian development. This form has been compared to Arm. anurÇ° âdreamâ and derived from *onÅr-i̯oâ, a reconstruction that is not without problems. However, the idea that á½Î½Î¿Î¹ÏÎ¿Ï is a contamination between á½Î½ÎµÎ¹ÏÎ¿Ï (the normal Greek form) and Lesb. *á½Î½Î¿Ï < *onrÌ¥ deserves full consideration. Cf. section 9.5 on word-final *ârÌ¥.
The Cyprian reflex of this word is attested as κοÏζία, in a gloss in Hsch. ascribed to the Paphians (see section 3.4.1).
In a number of cases (e.g. βÏοδοÏá½±ÏÎµÎµÏ Sapph. 53, and βÏόδÏν 55.2), βâ has been added by modern editors. In all three instances of βÏάδινοÏ, however, the mss. or papyri have initial βâ (reflecting digamma). Bowie criticizes Hookerâs view (1977: 28) that the βâ was a device to indicate that a short syllable was lengthened due to prevocalic initial á¿¥â. In fact, only in half of the cases in Sappho does the βÏâ close a final syllable that is short by nature (thus in á½ÏÏακι βÏαδίνῳ Sapph. 115; in Alcaeus, both cases of βÏâ generate a heavy syllable). Bowie thinks that words spelled with βÏâ are poetic archaisms of Lesbian: they preserve a reflex of *u̯â insofar as this was metrically useful, while in the vernacular, *u̯râ had already developed to râ by the time of Sappho.
See Chantraine (1933: 200â201) for the suffix âinóâ, and Mayrhofer (EWAia s.v. VRAD) for the suggestion to compare this with ῥαδινόÏ, Aeol. βÏάδινοÏ.
For instance, the o-grade root could be original, or an onset /ptro-/ (with the expected Aeolic reflex) may have been avoided.
These forms are adduced in the context of a discussion of the accentuation of the Homeric middle perfect infinitive á¼Î³ÏήγοÏθαι, and are intended to serve as parallels for an infinitive ending in âοÏθαι with proparoxytone accent (instead of the expected paroxytone). According to (pseudoâ)Herodian, such an accentuation is seen only the forms Ïá½³ÏοÏθαι, μέμοÏθαι, á¼ÏθοÏθαι in Aeolic (Ïá¿ Îἰολίδι).
In 3042, the codex has the accentuation Ïá½¹Ïναμεν; in 3043, the gloss κενÏούμεναι probably belongs elsewhere.
The âαâ in the gloss ÏοÏνάμεν may have been long or short: the acute accentuation need not be original.
For further discussion, see section 9.4.
According to GarcÃa Ramón (1975: 63), who starts from the position that *rÌ¥ was still intact in Mycenaean, âla conclusion sââ¯impose dââ¯elle-même: le proto-thessalien a développé *rÌ¥ > οÏ, Ïο à une époque où les Béotiens ne sââ¯Ã©taient pas encore séparés de la Thessalie, mais postérieure en tout cas à ca. 1200.â Note, however, that the regular vowel slot was not a relevant issue for GarcÃa Ramón. For a discussion of further phonological and morphological arguments in favor of positing a Proto-Aeolic stage, see GarcÃa Ramón (1975: 60â68) and (2009: 232â234).
On the basis of the syllabary, all four interpretations are possible, though a long vowel (an augmented form of the root *eu̯ergâ) is in my view more likely.
Tichy (1983: 287 n. 165) views ka-te-wo-ro-ko-ne as the imperfect of a root PGr. *u̯ergâ âto shut in, lock upâ, which she distinguishes (1983: 286â288) from PGr. *eu̯ergâ âto shut out, drive awayâ < *h2u̯ergâ (Skt. vrÌ¥nákti, ÄÌvrÌ¥nak). Together with Att. á¼ÏγÏ, á¼ÏÎ³Ï (forms with a short root vowel) and the Avestan opt. vÉrÉziiÄ n âto fence inâ, she derives the Cypriot form from an ablauting athematic root present. Whether this is correct or not is not directly relevant for present purposes, because ka-te-wo-ro-ko-ne would have a zero grade root also in Tichyâs interpretation.
See e.g. Egetmeyer (2010: 146).
For extensive discussion and further literature on Cypr. εá½â, uâ, cf. Egetmeyer (2010: 450â452). The best piece of evidence for *udâ as a relic alternative form of á¼Ïιâ is u-ke-ro-ne (ICS 217.A 5, 15), interpreted as a gen. pl. /u-khÄrÅn/ lit. âwhat is on the handâ, i.e. âsupplementary paymentâ. This interpretation is attractive in view of á¼Ïá½·ÏειÏαΠÏá½° á½Ïá½²Ï Ïὸν μιÏθὸν διδόμενα Ïοá¿Ï ÏειÏοÏá½³ÏÎ½Î±Î¹Ï âwhat is given to craftsmen on top of their wagesâ (Hsch. ε 5418). It remains unclear to me, however, how uâ < *udâ would relate to εá½â in the glossed form εá½ÏÏá½¹ÏÏεÏθαι.
â*trÌ¥kw-i̯é/óâ ⦠reste lââ¯hypothèse la plus solideâ (Egetmeyer 2010: 464). See there for other, less likely proposals.
Cf. Myc. to-ro-qe-jo-me-no /trokwe(i̯)omeno-/ âmaking tours of inspectionâ.
As it is usually cited, the form κόÏζα would presuppose a desyllabification of âiâ and the subsequent development of *âdi̯â to â¨Î¶â© in Cyprian. However, as Egetmeyer (2010: 125â126, with discussion of earlier literature) remarks in his discussion of the gloss, the codex of Hesychius has κοÏζία. He interprets this as an intermediary stage between κοÏδία and disyllabic /kordzÄ/. Whether this interpretation is correct or not, the reading κοÏζία must be maintained.
GarcÃa Ramón and Helly (2012: 61â63) read the form ΣÏÏοÏικα as an epithet of the Thessalian goddess Ennodia in a dedication from Larisa (SEG 54, 561; 3rd quarter of the 5th c.). If this is correct, the variant ÏÏÏοÏá½± is also secured for Thessaly.
The last point is not addressed by Peters (1980: 208 fn. 160), who thinks that the absent reflex of the initial *h2â can be due to laryngeal loss in a compound.
Thus also Haug (2002: 60). Note that the interpretation of Myc. to-pa-po-ro-i as /storpÄphoroihi/ âfor the torch-bearersâ (thus e.g. Waanders 2008: 37, Egetmeyer 2010: 146) is uncertain; for an alternative interpretation /torpÄ-phoro-/, connecting Att. Ïá½±ÏÏη denoting a type of basket, see chapter 2.
Compare the skepticism of Egetmeyer (2010: 147) on this gloss; I disagree with him, however, on the inclusion of θÏόνα âvaricolored embroideriesâ among the Cyprian evidence for *rÌ¥ (on this word, see section 2.5.2).
Leumann (1950: 273) thinks that the ascription ÎÏ ÏÏá½·Ïν of glosses (e.g. those in the καÏá½° Ïá½¹Î»ÎµÎ¹Ï list) may conceal the fact that these words occurred in the epic poem with the title Cypria. Although some of Leumannâs views are in my view far-fetched, a skeptical attitude towards the glosses marked as âCyprianâ seems in order.
For further discussion of ÏαÏÎ²á½³Ï and cognates, see section 4.2.1.
Unfortunately, the more precise provenance of this text within Arcadia is unknown.
In Van Beek 2013 I suggested that the Mycenaean month name a-ma-ko-to me-no /hamagortÅ mÄ(n)nos/ âin the month of the assemblyâ (cf. Taillardat 1984) reflects PGr. *smÌ¥â plus *agrÌ¥-toâ. However, the underlying form could also be an o-grade formation PGr. *ágor-toâ of the type νόÏÏοÏ.
Of course, the vowel slot of αγαÏÏÎ¹Ï could theoretically be analogical after the full grade of the verbal root. For the vocalization of *ârÌ¥sâ, see section 9.1.
See Van Beek (2011a) for a criticism of âvowel assimilationsâ in Greek, and cf. also the doubts expressed by Dubois (1988: 44 with n. 219).
Note, in this connection, that ÏεÏá½¹ÏÏÎ±Î¹Î¿Ï (Theoc. 30.2) is inadmissible as evidence for a Lesbian form Ïá½³ÏοÏÏοÏ*. The form in Theocritus (of unknown dialectal origin) may be analogical and based on the Doric cardinal Ïá½³ÏοÏεÏ.
According to Minon (2007: 301â302), the Elean alphabet was taken over from the Laconians. This would explain why γÏοÏÎµá½»Ï is found in that dialect. Is a similar explanation possible for the occurrence of γÏοÏÎµá½»Ï in Arcadian?
Haug (2002: 61) proposes to assume influence of the present stem of δÏá½±ÏÏομαι < *drnÌ¥ghâ on δαÏÏμα in Arcadian and Aeolic, but this does not explain the deviating vowel slot in comparison with Class. δÏαÏμή. One might therefore think that δÏαÏμή underwent the influence of the present stem, while δαÏÏμα, δαÏÏνα contain the regular outcome of PGr. *drÌ¥khmnÄ. See section 9.2.1.
Pace Morpurgo Davies (1968: 796), whose speculations on geographically different treatment of ârsâ are not supported by the evidence; cf. Dubois (1988: 80â83), who argues that âÏÏâ is found until the late 5th c., and that it later developed (from the 4th c. onwards) into a form with compensatory lengthening.
The regular outcome of *l̥ is unclear in all three dialects.
Brixheâs comment, âqui sans doute est une forme partiellement extra-dialectale pour *Î ÏοÏá½¹Ïαâ (1976: 61), is difficult to understand. Does he mean that the liquid metathesis points to extra-dialectal origin?
Brixhe (1976: 62) further mentions the forms κεκÏαμενοÏ, ΤÏÎµÎºÎ¿Ï Î´Î±Ï, ÎÏÎµÎºÎ¿Ï Î´Î±Ï, and ΣÏÏαÏοκλιÏÎ¿Ï Ï. An uncertain piece of evidence is the gloss κοÏÏá½±ÏοιÏÎ â¨ÎºÏοÏá½±ÏοιÏâ© á½Ïὸ ΠεÏγαίÏν (Hsch. κ 3659 L-C), which is an emendation by Latte of κοÏá½±ÏÎ¿Î¹Ï á½ÏοÏαÏγαίÏν in the codex.
Melchert (2004: 595) thinks that an anaptyctic vowel had developed before syllabic nasals and liquids in attested Lycian spellings like hrppi [hÉrp.pi], as this would explain the use of a geminate spelling âppâ.
This scenario was proposed also in Van Beek 2013. Skelton (2017: 113), apparently without having seen my dissertation, also concludes from the forms Î ÏειιαÏ, ΣÏÎ»ÎµÎ³Î¹Î¹Ï Ï, and ΤÏÎµÎ¼Î¹Î»Î±Ï that they were taken from a Lycian-like Anatolian language with syllabic liquids. However, her claim that Pamphylian speakers still pronounced these syllabic liquids, and that Ïε is an attempt to render this in Greek alphabet, is clearly untenable: in this way one cannot explain the emergence of an epenthetic stop in ΣÏÎ»ÎµÎ³Î¹Î¹Ï Ï.
This may also explain the reflex âÏεâ in the PNâ¯s ΤÏÎµÎºÎ¿Ï Î´Î±Ï, ÎÏÎµÎºÎ¿Ï Î´Î±Ï, which are the Pamphylian reflexes of a borrowed Lyc. trqqñtâ âStorm Godâ (cf. Hitt. tarḫuntâ).
Skelton (2017: 113) suggests that the spellings ÏεÏÏâ and ÎÏοÏδιÏÎ¹Î¹Ï Ï may actually ârepresent an attempt to write a syllabic liquidâ, but this seems risky in view of the scanty evidence. She also asserts that the forms âcould very well have come from Cretanâ, which would fit the foundation myth reported by Herodotus (1.173) âthat the Lycian Sarpedon led a group of Cretans to settle in Lycia.â (Skelton 2017: 110). However, as she equally admits, the difference between ÏεÏÏâ and Cretan ÏοÏÏι ârequires some explanationâ.
In chapters 6 and 7, I will argue that Epic Greek had a special reflex *rÌ¥ > âÏαâ, but âÏοâ after a labial consonant. The conditions for this change are the same as in Cretan, but the outcome is different. I therefore see no reason to assume a special relation between the Cretan development and that of Epic Greek.