General Abbreviations
7 Chronology of Changes in Attic and Ionic
7.9 Chronology of changes
The forms discussed in §7.5 contain a lot of chronological information. To block Att. *κρα�νη by reversion of /ǣ/ to /ā/ after /r/, Att.-Ion. */krnǣ/ had to undergo an early dissimilation to */krɛ�nǣ/. The same dissimilation is found in *par-
h₂ews-éh₂ ‘(item) next to the ear’ > *parawh > Att.-Ion. */parǣw/ > */parɛ̄w/
> */parɛ̄/ (loss of /w/) > Att. /pare/ (shortening and reversion) παρεα� ‘cheek’. The derivation of Ἑρμῆς ‘Hermes’ from Attic-Ionic *Hermhǣs from *Her-
mhās (cf. Myc. dat. sg. e-ma-a₂ /Hermhāi/ [Py.] ‘to Hermes’) exhibits the same
Chronology of changes 69
The difference between Att. πρᾱότης ‘mildness’ (< Att.-Ion. *prǣhewotǣt-) and κόρη ‘girl’ (< Att.-Ion. *kórwǣ; cf. Myc. kórwā) shows that reversion after /r/ ceased to apply after (at least) postconsonantal w-deletion.
Consider Att. ὄρη ‘mountains’ (older *óreha) vs. χρέᾱ ‘debts’ (earlier *khrɛ�weha) and ὑγιᾶ (acc. sg. m., nom./acc. pl. nt. of ὑγιής ‘healthy’). Miller (1976a) accounts for variation between ὑγιᾶ and ὑγιῆ [Xenophon, Plato, etc.] vs. earlier ὑγιᾶ [Thucydides] and eventual victory of ὑγιῆ by loss of reversion. The absence of *ὄρᾱ for ὄρη ‘moun- tains’ is accounted for by earlier loss of reversion in the most opaque environment (after /r/). Peters (1980: 299) rightly counters that the complete absence of any forms of the *ὄρᾱ type strongly suggests that reversion 1 (after /r/, called rhotic lowering by Samuels 2006) was earlier than reversion 2 (after /e/ and /i/, Samuels’ dissimilation). The first application of QM (§7.8) applied after h-deletion (and possibly also after dissimilation). Hermɛ�ǣs may have undergone it, although the standard doc- trine is that front vowel sequences were not affected (Peters 1980: 251; Méndez Dosuna 1993: 99). Consequently, in the derivation below only shortening (no glide formation) is indicated in this word.
The following derivations incorporate the two applications of Shortening/QM to account for the difference between ὀρχηστῶν ‘of dancers’ and βασιλέων ‘of kings’ amid Reversion 1 and 2, dissimilation, contraction, etc.²7
Hermhǣs orkhɛ̄stǣhɔ̄n parǣw prǣhewótǣs óreha basilɛ�wɔ̄n khrɛ�weha 1) h-deletion Hermǣs orkhɛ̄stǣɔ̄n — prǣewótǣs órea — khrɛ�wea 2) Revers1 — — — prāewótǣs — — — 3) Dissim. Hermɛ�ǣs — parɛ̄w — — — — 4) Shrt/QM1 Hermɛ́ǣs orkhɛ̄stæ̯ɔ̄n — — — — — 5) Contract. Hermæ῀s orkhɛ̄stɔ̃n — prāwótǣs órǣ — khrɛ�wǣ 6) w-delet. — — parɛ̄ prāótǣs — basilɛ�ɔ̄n khrɛ�ǣ 7) Shrt/QM2 — — parɛ — — basiléɔ̄n khrέǣ 8) Revers2 — — parɛ — — — khrέā 9) /ǣ/ > /ɛ̄/ Hermɛ̃s — — prāótɛ̄s órɛ̄ — —
27 Apart from the double application of Shortening/QM recognized by only a few, the stages in the derivation differ little from those of Bartoněk (1966: 103), Lejeune (1972a: 234 ff., 262 f., 369), Peters (1980: 303; 1995), or Samuels (2006). The main differences involve rules that are not criti- cally ordered with respect to one another. Technically h-deletion probably preceded the change of /ā/ to /ǣ/ since the former began in Mycenaean. The key point of agreement is that Attic reversion applied at two different times, first rhotic lowering, then later front vowel dissimilation. Also, on all accounts, rule (2) belongs to proto-Attic-Ionic, and the rest to Attic, although most (including a species of reversion) appear to be independently shared with at least some varieties of Ionic.
70 Chronology of Changes in Attic and Ionic
Some of the intermediate stages of the derivation remain in older forms pre- served by the poetic tradition, e.g. Homeric Ἑρμείᾱς (see below) beside Hom., Ion., Att. Ἑρμῆς, Hom. παρειαί ‘cheeks’ (acc. pl. παρεια�ς ii.153), sg. χρεῖος ‘debt’ (11.686, 688). In each case, the spelling reflects preservation of the form in a met- rical slot requiring a long vowel, but with metrical lengthening of the contempo- rary short vowel. That is, metrically the form is /parɛ̄aí/ (stage 6) but graphically it is not the expected *παρηαί but rather a lengthening of contemporary παρεαί, hence the spelling ει for /ē/ (§8.9).
Sihler’s chronology (1995: 51), with postconsonantal w-deletion (*kórwǣ > κόρη ‘girl’) after the merger of /ǣ/ with /ɛ̄/ [c5ᵉ], cannot possibly be correct since the oldest Attic inscriptions have no trace of /w/. The difference between κόρη and νέᾱ ‘new’ (nom. sg. f.) from *néwǣ do not show that /w/ was lost earlier between vowels (Sihler, p. 52) but rather that reversion took place in two stages and that the second (after w-deletion) operated only after front vowels. In that case, loss of /w/ in the *kórwǣ type could have been around the same time as the intervocalic loss (stage 6). Stage (8) then is the cutoff point for prehistoric Attic. Only (9) occurred within the historical period.
To account for the hero name Ἀμφιάρᾱος [xv.244, 253] (< *-rǣwos), Peters (1995) posits a West Ionic reversion between /r/ and /n/ or /w/ that preceded intervocalic /w/ deletion. Attic also had several more complex environments for reversion, as in σικύᾱ ‘bottle-gourd’, δικρόᾱ ‘forked, cloven’ (fem.), καρύᾱ ‘hazel- nut tree’, ἀθρόᾱ ‘crowded together’ (f.), ἀρα� ‘prayer, curse’ < *arw (§22.20), etc. (Miller 1976a: 151 f.).
Since Peters (1993: 93; 1995: 196) also includes a West Ionic environment for reversion after /e/ following loss of /w/, Homer’s Ἑρμείᾱς raises one final point of potential significance. It seems suspicious that a number of assured nomina- tive singulars in -ᾱ(ς) have the same structure: Αἰνείᾱς, Αὐγείᾱς, Ἑρμείᾱς, Ῥείᾱ (GH i. 199; cf. Peters 1980: 263‒273). One cannot help but wonder whether some variety of West Ionic shared with Attic one portion of the dissimilatory reversion, viz. *Herm(h)ǣs > *Hermɛ�ǣs > *Hermɛ�ās (> Ἑρμείᾱς) in place of contraction. Names like Εὐαγρείᾱς (IG 12/9: 723: Eretria), Αἰνέᾱς (IG 12/9: 249.139, 216: Eretria [c3]), are attested in Euboean, but (i) there is no guarantee that these people are West Ionians, and (ii) also attested are names like Πυρρίε̄ς (IG 12/9: 56,342 ff.: Styra [c5]). Since names with Πρηξι- tend to be native Eretrians (§13.10), Πρηξίᾱς Πρηξι�νου (IG 12/9: 249.203: Eretria [c3]) may be genuine. If any of the names in -(ε)ίᾱς are verifiably West Ionic, because of the same dissimilation, it then becomes less of a leap to accept Cumaean θεα� ‘goddess’ as West Ionic (§9.3), but see §25.4.
Conclusion 71
7.10 Conclusion
Quantitative metathesis (QM) applied with considerable regularity in Attic, pro- ducing entirely novel paradigms. In all varieties of Ionic, QM applied in morpho- logically unmotivated forms and in derived compounds, but in some varieties of especially East Ionic, QM did not override standard paradigmatic forms at least to the output of w-deletion. Disyllabic forms were especially stable and did not undergo the glide formation that would yield QM, hence forms like νηός / νεός ‘of a ship’. The outputs of QM following *y/h-deletion are most predictably monosyl- labic, esp. gen. sg. -εω.
In Attic, the Ionic-Attic */ǣ/ reverted to /ā/ in several well-known environ- ments (after /r, e, i/) and in several complex environments, exemplified by σικύᾱ ‘bottle-gourd’ and ἀθρόᾱ ‘crowded together’. Although reversion is not generally admitted for Ionic, at least West Ionic may have had several environments, exem- plified by Ἀμφιάρᾱος and Ἑρμείᾱς, if these dialectal names were not simply pre- served by the tradition.
None of the arguments against 5th century Attic /ǣ/ are probative. Evidence from phonological and morphological restructurings supports a late c5 merger of /ǣ/ with /ɛ̄/. The fact that many adjustments occur around the same time and are most reasonably interpreted as reactions to the merger of /ǣ/ with /ɛ̄/ would be an incredible accident if the merger had occurred several centuries earlier. Since /ǣ/ and /ɛ̄/ remained distinct well into the 5th century in some areas of the Cyclades, there is no a priori reason to reject a late merger of the two in Attic. After all, the island of Ceos, where /ǣ/ was merging with /ɛ̄/ late in the 5th century, was just fifteen miles southeast of Attica. It is reasonable to interpret the merger as an areal feature.
The rationale for the Attic merger of /ǣ/ with /ɛ̄/ toward the end of the 5th century is evident from Teodorsson’s study (1974: 264 ff.) and Threatte’s copious documentation (1980). Inscriptions suggest that numerous processes were in progress that served to raise long (especially front) vowels. The recently tensed, peripheral character of inherited /ɛ̄/ caused it to merge with /˂ɛ̄/ (< /ǣ/), and ‘both’ were rising to [ē] while /ē/ was itself moving toward [ī]. Some speakers, as is typical in Labov’s studies, were more advanced in the shift than others, as indicated by early spellings of both (original) /ɛ̄/ and /˂ɛ̄/ (< /ǣ/) with Ι (§6.7 ff.).
MF Ancient Greek Dialects