Athematic infinitives in Lesbian, Homer and other Greek dialects
Innovations, archaisms or contact-induced borrowings?
p. 185-208
Résumés
This paper evaluates how the Lesbian athematic infinitive endings ‑μεναι and ‑ν may shed light on the prehistoric development of this dialect. There is a widely-accepted supposition that these endings are evidence of linguistic contact between Lesbian and Eastern Ionic. Instead, I argue that the distributions of ‑μεναι and ‑ν in Lesbian neatly reflect the situation to be reconstructed for Proto-Greek: *‑men was attached to ablauting stems, while *‑hen attached to non‑ablauting stems. Further evidence in support of this reconstruction is adduced from other dialects and from Homeric Greek.
Cette contribution examine les désinences ‑μεναι et ‑ν de l’infinitif athématique du lesbien (éolien d’Asie), et leur apport à la reconstruction de la préhistoire de ce dialecte. L’une et l’autre terminaisons sont attribuées communément aux contacts linguistiques étroits entre le lesbien et l’ionien oriental. Je propose, en revanche, que la répartition de ‑μεναι et ‑ν dans les thèmes verbaux où elles sont attestées relève de la situation reconstruite pour le proto-grec : *‑men était la désinence propre aux thèmes à apophonie, alors que *‑hen était utilisée dans les thèmes sans apophonie. Cette reconstruction s’appuie sur l’examen des désinences correspondantes dans d’autres dialectes grecs et dans l’épopée homérique.
Texte intégral
1. Introduction
1Lesbian has three different markers of the active infinitive:1 thematic ‑ην, athematic ‑ν, and athematic ‑μεναι, as exemplified by the following examples:2
| ἄγην ‘lead’ | (Att.‑Ion. ἄγειν) |
| ομνυν ‘swear’ | (Att.‑Ion. ὀμνύναι) |
| ἔμμεναι ‘be’ | (Att.‑Ion. εἶναι) |
2In this paper, I aim to elucidate the origin and distribution of the athematic endings ‑μεναι and ‑ν, and to evaluate their status as evidence for linguistic contact between Lesbian and Eastern Ionic, and between Lesbian and Epic Greek.
3It is often claimed that Lesbian ‑μεναι arose by contamination of original Aeolic ‑μεν and Eastern Ionic ‑ναι as the result of linguistic contact.3 For athematic ‑ν, too, an explanation involving contact influence of Eastern Ionic has been proposed,4 although it is mostly explained as an inner-Lesbian reanalysis of the thematic ending ‑ην.5 Thematic ‑ην itself is widely seen as an adaptation of Eastern Ionic ‑ειν to Lesbian phonology, replacing the allegedly older ending ‑έμεν as found in continental Aeolic dialects. Thus, the infinitive endings play an important role as alleged evidence for language contact between Greek dialects spoken in the Eastern Aegean during the late Dark Ages.
4As for thematic ‑ην, however, the argument for borrowing is not cogent: Lesbian ‑ην may continue Proto-Greek *‑ehen, while the generalization of ‑μεν from athematic to thematic verbs can be seen as a common innovation of Thessalian and Boeotian rather than a feature inherited from Proto-Aeolic.6 In this article, I argue that the Lesbian athematic endings ‑μεναι and ‑ν partly reflect an archaic distribution between the two oldest athematic endings, Proto-Greek *‑men and *‑hen. In support of this claim, I revisit the distribution of athematic infinitive endings in Classical Attic and Eastern Ionic, Western Ionic and Arcadian. This will also involve commenting on the origin and distribution of the endings ‑μεν, ‑μεναι and ‑ναι in Epic Greek, where the presence of ‑μεναι is commonly taken as evidence for the involvement of Lesbian singers in the tradition.
5The structure of this paper is as follows. § 2 sketches a preliminary outline of the athematic infinitive endings in Proto-Greek, as a background against which the dialectal developments can be seen. § 3 bolsters this with a bottom‑up reconstruction, based on the evidence for athematic infinitive endings in Attic-Ionic, Arcadian, Lesbian and Homer. § 4 shows how the variety of dialectal forms and distributions may have developed step by step from the Proto-Greek situation. Finally, § 5 reconsiders the evidence for linguistic contact between Lesbian and Eastern Ionic and evaluates how the infinitives may shed light on the prehistoric development of Lesbian.
2. The infinitives in Proto-Greek
6It is widely assumed among Indo-Europeanists that the morphosyntax of infinitives in the early prehistory of Greek closely resembled the situation of Vedic Sanskrit.7 Allegedly, Proto-Greek had a variety of nominal suffixes (including *‑men‑, *‑wen‑, *‑hen‑, *‑en‑, *‑eneh2‑) and inflected case-forms (dative, locative, allative) that formed quasi-infinitives to a verbal root. When the infinitive was grammaticalized, each individual dialect group selected a subset of these markers, and all tense-aspect stems of a verb received an infinitive.
7Such an approach, however, leaves the “systematic” morphological similarities and differences between the different Greek endings unaccounted for. Moreover, the infinitive is already a strongly grammaticalized form in the oldest attested varieties of Greek, Mycenaean and Homeric,8 and archaic modal uses of the infinitive in Homer9 cannot be tied to specific morphological markers.
8For these reasons, I propose to start from a relatively simple Proto-Greek situation, with an infinitive that was already fully grammaticalized and marked by a small set of dedicated endings. My aim is to show how the variety of dialectal endings may have developed from this reconstructed situation by natural and well-motivated linguistic changes. I will try to arrive at the Proto-Greek situation by means of dialect comparison as well as internal reconstruction.
9As a bare minimum, we must reconstruct for Proto-Greek the athematic endings *‑men and *‑hen, as well as an element ‑ai (of which the original function is difficult to recover) which yielded the extended forms *‑menai and *‑henai.10 Besides, inter-dialectal comparison proves that a thematic active ending *‑ehen, a sigmatic aorist active in *‑(t)sai (whose PIE pedigree is debated), and a middle ending ‑sthai must be reconstructed (tab. 1):11
Tab. 1 – The Proto-Greek infinitive endings.
| Ending | Extended form | |
| Active athematic | *‑men | *‑menai |
| *‑hen | *‑henai | |
| Active thematic | *‑ehen | |
| Active s‑aorist | *‑(t)sai | |
| Middle | *‑sthai | |
10Athematic ‑hen is directly attested in Myc. te‑re‑ja‑e /teleiāhen/ ‘fulfill’12 and reflected in such alphabetic forms as Arc. εξιεν ‘go out’ (= ἐξιέναι) and WIon. εξειν ‘be possible’ (= ἐξεῖναι). Unlike various previous scholars,13 I do not consider it necessary to reconstruct a vowel-initial infinitive ending *‑en(ai), because this would multiply the entities beyond necessity.14
11A separate ending *‑wenai is often reconstructed for Proto-Greek to account for Cypr. to‑we‑na‑i, attested twice on the Idalion Bronze. García Ramón 1994 and 2009 argues for the old comparison with Vedic dāváne, and reconstructs the PIE form as a quasi-infinitive. Peters 1989 reconstructs *‑wenai also for infinitives of other monosyllabic stems, such as εἶναι, θεῖναι, εἰδέναι. However, positing an ending *‑wenai raises the question of what the original distribution with the competing ending *‑men(ai) was, as the latter also occurs on monosyllabic stems. Since M. Peters has not provided an answer to this question, we must briefly consider whether the evidence really requires us to posit *‑wenai.
12The metrical evidence for θεῖναι, δοῦναι and their compounds in Homer does not favor the reconstruction *‑wenai (Cowgill 1964, p. 356). In 6 out of 9 cases, the contraction in the first syllable of δοῦναι is irresolvable, and in the other 3 instances δοῦναι can easily be an innovation for older δόμεναι.15 The numbers for θεῖναι are similar. From this evidence it follows that δοῦναι and θεῖναι had already been contracted from *δόεναι, *θέεναι (whether with *w or *h) in Ionic when the Iliad was composed; these trochaic forms were clearly useful, especially in compounds like ἀναθεῖναι.16 On these grounds, then, neither *dowenai nor *dohenai can be categorically excluded as the pre‑form. However, in possessive derivations in ‑όεντ‑ < *‑o‑went‑ in Homer there is a virtually complete absence of forms showing contraction over the morpheme boundary (the only exception is λωτοῦντα, Il., XII, 283), and the same holds for compounds in ‑οειδής < *‑o‑weidēs. Therefore, *dohenai is by far the most plausible reconstruction of Homeric δοῦναι.17 The w‑spelling of Cypr. to‑we‑na‑i may, in the end, be the orthographic representation of a secondary glide (see Cowgill 1964, pp. 358‑359 for possible scenarios), or it could be due to hypercorrection (Masson 1983).
13After these preliminary considerations, let me formulate the main questions to be answered in this paper.
- What was the reason for generalizing the extended forms *‑men(ai), *‑hen(ai) in various dialects?
- What was the earliest distribution between *‑men and *‑hen?
14According to a widespread view, the infinitives in ‑men and *‑hen reflect endingless locatives of PIE n‑stem verbal abstract nouns, or n‑locatives of stems in *‑m‑ and *‑s‑.18 Most scholars nowadays seem to view *‑menai and *‑henai as another case form of these same verbal nouns in ‑n‑: an old dative,19 locative,20 or allative.21 However, since no meaningful syntactic, semantic or pragmatic difference between infinitive forms with and without ‑ai is found, and since there was no case ending *‑ai or *h2ei in PIE, it would be arbitrary to view ‑ai as a case marker. Alternatively, Meillet 1931, pp. 192‑193, proposed that ‑ai was originally a particle that was added optionally to the infinitive endings ‑men and ‑hen.22
15Whatever the original function and morphological analysis of ‑ai, it is more fruitful to ask why extended forms in ‑ai gained the upper hand in some situations, but not in others. I submit that in different dialects, variants extended by ‑ai were preferred over their non‑extended counterparts in cases where disambiguation between homophones was useful. For instance, we may assume that, in pre‑Lesbian,23 δόμεναι and δόμεν were originally competing infinitive forms. At some point in the Late Bronze Age, when secondary indicative forms with and without the augment were still functionally opposed, the functional load of ‑αι had faded. Under these circumstances, using δόμεναι as an infinitive offered the advantage of distinguishing it from unaugmented 1st pl. indic. δόμεν ‘we gave’. As we will see in § 4.2, this scenario allows us to understand why Lesb. ‑μεναι occurs only with monosyllabic stems containing a short vowel.
16Within this framework, the Lesbian ending ‑μεναι can be connected directly with athematic ‑μεν attested elsewhere in Aeolic and West Greek dialects. It also throws a different light on the Homeric alternation between the endings ‑μεν and ‑μεναι: although this was clearly artificially extended for metrical reasons, it would be attractive if the optional addition of ‑ai (in part) reflected an archaic trait. This account of Lesbian ‑μεναι also encourages us to compare the Att.‑Ion. ending ‑έναι with the evidence for athematic *‑hen. As we will see, forms extended by ‑αι may have been restricted to (certain types of) monosyllabic stems also in early South Greek dialects.
17If ‑αι was originally an accessory feature of the athematic infinitive endings, it is natural to inquire into the distribution between *‑men and *‑hen in Proto-Greek. It is usually thought that the choice between these endings was a very early feature that defined the split between North Greek and South Greek dialects. Indeed, there seem to be no direct traces of ‑men in South Greek dialects, and hardly any traces of athematic *‑hen in North Greek dialects. There is, however, one exception: Lesbian has both ‑μεναι and ‑ν. Now, if athematic ‑ν is an old feature of Lesbian reflecting Proto-Greek *‑hen,24 this dialect could furnish direct evidence for the original distribution between *‑men and *‑hen. Further crucial evidence comes from the choice between ‑μεν and ‑ναι in Epic Greek: in § 3.5, I propose that their distribution points to a preservation of ‑μεν, in Eastern Ionic contemporaneous with Homer, after polysyllabic athematic present and perfect stems.
3. The dialectal evidence
18Starting from these hints contained in the Lesbian and Homeric evidence, I intend to sketch a scenario that not only explains how *‑hen(ai) could gain the upper hand in South Greek dialects, but also clarifies why ‑men gained complete dominance in North Greek dialects – with the exception of Lesbian.25 A possible scenario for the genesis of Lesbian ‑μεναι has just been proposed in § 2; an account of how and why ‑ν could spread in this dialect will be given in § 3.4.
3.1. Secondary creation of ‑ναι in Attic-Ionic
19The endings in use in Attic, Central and Eastern Ionic are ‑ναι and ‑έναι. In classical Attic and literary Eastern Ionic, the longer form ‑έναι overtly occurs in the perfect stem (e.g. λελυκέναι) and in ἰέναι; the shorter form ‑ναι is more widespread, and occurs both after long and short vowels.
20As was seen clearly by Schwyzer 1939, p. 808, ‑ναι may have arisen secondarily from ‑έναι by contraction, and then spread to other forms. The following points are relevant.
- The longer form ‑έναι is directly preserved in ἰέναι and is also presupposed by aor. θεῖναι, δοῦναι from *thehenai, *dohenai, as well as by εἶναι, which must reflect *ehenai (see below).
- The forms of the aorist passive in ‑(θ)ῆναι and the root aorist (e.g. στῆναι) may have arisen by contraction: *‑(th)ē‑henai, *stā‑henai. This idea is confirmed, as E. Schwyzer remarks, by the gloss ψαέναι· φθάσαι ‘come first, outrun’ (Hsch.): it reflects the same pre‑form as φθῆναι, the infinitive of ἔφθην, but in a dialect in which *ā(h)e had not yet been contracted and where φθ‑, πτ‑ became ψ‑.26
- The key form εἶναι ‘be’ must reflect *ehenai rather than *ehnai or *ehwenai. One reason is that an instance of Arcadian ηναι (IG V, 2, no. 343, l. 43) occurs in an inscription that has a geminate instead of compensatory lengthening in the form ε[κ]ρ̣ινναν (l. 48‑49).27 This excludes the reconstructions *ehnai and *ehwenai, forms which would probably have yielded *ἐνναι and *εὐεναι, respectively, in this variety.28 Another important point is that the rhythmical structure of Homeric ἔμεν and ἔμεναι is understood most easily if these have replaced *ehen and *ehenai.29 For such replacements in Homeric infinitives, see infra § 3.5.
21After contraction with preceding long vowels following h‑loss, the shorter form in ‑ναι may have spread from athematic aorists (including the productive ‑θῆναι) to other athematic stems. All present, perfect and aorist stem infinitives that might seem to speak for original ‑ναι can or must be explained by analogy.
- Root aorist infinitives like δῦναι, φῦναι (Hom.+) and perfect infinitives like ἑστάναι, βεβάναι (post‑Hom.) can be analogical after βῆναι, γνῶναι, ‑θῆναι, etc. As noted by García Ramón 1990, the concrete model may have been provided by the 2nd person imperatives, with which the infinitive shared one of its main functions:
| aor. | βῆθι | βῆτε | βῆναι | ||
| aor. | δῦθι | δῦτε | δῦναι | ||
| perf. | ἕσταθι | ἕστατε | ἑστάναι |
- Ablauting athematic present infinitives of the types δεικνύναι and ἱστάναι, διδόναι, τιθέναι in Attic can also be understood as analogical developments. The generalization of ‑ναι may have happened via a proportion with the 2nd plural imperative δείκνυτε, ἵστατε (García Ramón 1990, p. 163), but it could also be an elimination of the final remains of the unproductive ending ‑μεν.30
22An important additional argument for a secondary origin of ἱστάναι, ἑστάναι is the fact that such forms (in which ‑ναι stands after a short vowel) are unattested in Homer, even if they are, for practical purposes at least, metrically equivalent to attested Homeric forms in ‑μεν (e.g. ἑστάμεν ‘stand’).31 In § 3.5, I argue that forms like ἑστάναι did not yet exist in the variety of Ionic spoken by Homer. Moreover, as we shall see now, certain infinitives in ‑ν of Western Ionic dialects and Arcadian are unlikely to be secondary for ‑ναι.
3.2. The Western Ionic and Arcadian evidence
23The fragmentary evidence for athematic infinitives in Western Ionic (mainly from Eretria and Oropos) differs in important ways from that in other Ionic dialects and Attic, but is similar in other respects to the Arcadian and Lesbian evidence. Western Ionic and Arcadian share an old core of athematic infinitive forms in ‑ν that may well reflect ‑hen (as in Myc. te‑re‑ja‑e /teleiāhen/).
24The Western Ionic forms in ‑ν are often taken to be analogical innovations (García Ramón 1990), but in my view some of them might be old (for this view, but different in the details, see also Plath 1990). J.L. García Ramón clearly summarizes the evidence and argues that the oldest distributions are as follows.
- Monosyllabic stems originally had their infinitives extended with ‑αι: cf. aor. δουναι, °θειναι (as in Attic and Eastern Ionic). In later inscriptions, °δουν, °θειν are found for compound verbs.32 Parmenides (frg. B 8, 10) uses the simplex aorist infinitive φῦν at verse‑end.33
- Polysyllabic stems have forms without ‑αι, as in the reduplicated presents τιθεῖν, διδοῦν, ἱστᾶν (Eretria and Oropos, 4th c. BCE) and the inagentive aorist ἀνα‑σωθῆν (Eretria, 3rd c. BCE); cf. also μιγῆν in Parmenides (frg. B 12, 5). Slightly later inscriptions from these localities have τιθέναι, διδόναι instead of τιθεῖν, διδοῦν.
- As a simplex, εἶναι contrasts with compounded ἐξεῖν in the earliest evidence from Eretria and Oropos. In both localities, a simplex εἶν ‘be’ is attested later than εἶναι. However, εἶν is used already by Stesichorus (frg. S15, 7) in the 6th c. BCE,34 and it is the only form attested epigraphically in Olynthos (389‑383 BCE).
25J.L. García Ramón proposes that, at a first stage,35 the present infinitives τιθεῖν, διδοῦν and ἱστᾶν were created by analogy with thematic contract verbs like καλεῖν, based on a proportion with the 2nd plural imperative. While I agree that these infinitives of disyllabic stems were analogically created, I do not consider it plausible that all such infinitives were created ex nihilo, as J.L. García Ramón proposes; I rather assume that they replaced older forms in ‑μεν, as I will argue in more detail below (see infra § 4). Also, the proportion with the type καλεῖν assumed by J.L. García Ramón rests on the assumption that such verbs had an ablauting stem (with 2nd plural *κάλετε), which in my opinion is implausible. Nevertheless, that contract verbs may have played a role in the creation of τιθεῖν, διδοῦν is indeed suggested by the existence within the paradigm of forms like 3rd sg. τιθεῖ, διδοῖ in Eastern Ionic.36
26However this may be, the inagentive aorist [α]νασωθην (instead of °σωθῆναι) can hardly be accounted for as a thematization, as there is no evidence for thematic forms elsewhere in the paradigm of this aorist stem. The same holds for the pair εἶναι (simplex) beside εξειν (compound) in Oropos, and for other early occurrences of εἶν.37 It is unlikely that the infinitive of the substantive verb would be deprived secondarily of its characteristic ending ‑αι and reduced to a monosyllable. Rather, I would argue that the pair ἐξεῖν beside εἶναι reflects something old: when the extension ‑αι lost its original function, it was generalized in the simplex εἶναι, but not in compounds like ἐξεῖν. I will get back to possible motivations for this generalization in § 3.5. Furthermore, the pre‑forms *°ehen, *ehenai that would yield °εῖν, εἶναι can directly account for the odd Homeric infinitives ἔμεν and ἔμεναι (see supra § 3.1). Finally, if the variation between °ειν and εἶναι was retained longer in the substantive verb, it may well have induced a secondary interchange between simplex δοῦναι, θεῖναι and compounded °δουν, °θειν in 4th c. Western Ionic.
27In sum, the proliferation of athematic infinitives without ‑αι in Western Ionic presupposes an old Proto-Ionic core of such forms, including at least *‑ehen (compounds) versus *ehenai (simplex) in the substantive verb, and ‑(θ)ῆν < *‑(th)ē‑hen in inagentive aorists. The addition of ‑αι in class. ‑θῆναι is easily understood as a generalization of the ending used in the frequent monosyllabic aorists such as στῆναι, βῆναι and others.
28Turning to Arcadian, in IG V, 2, no. 4, l. 1 (4th c. BCE, Tegea) we find an infinitive εξιεν ‘to go out’ without ‑αι, as against ηναι ‘to be’ elsewhere in the same inscription.38 Other Arcadian infinitives that are traditionally cited are απειθηναι ‘to be disobedient’ and κατυφρονηναι ‘to despise’ in IG V, 2, no. 6 (l. 46 and 47 respectively, also from 4th c. BCE in Tegea), with the extension ‑αι.39 However, Peters 1986, p. 308, has drawn attention to two other Arcadian athematic infinitives without ‑αι, both in presents of verba vocalia: αντευεργετην in a decree from Megalopolis (Syll.3, no. 559, l. 20, 3rd c. BCE) and κοινανην at Telphousa (SEG XII, no. 371, l. 24, 3rd c. BCE).40 It is remarkable that the last‑mentioned inscription also uses the infinitives ηναι ‘be’ and δωναι ‘give’, both with a monosyllabic stem. For Arcadian, then, it is attractive to conclude that the extension ‑αι is original in forms that would otherwise be monosyllabic (ηναι, δωναι), while polysyllabic stems (including compounded verbs) originally did not have ‑αι (εξιεν, αντευεργετην, κοινανην). The reason behind this distribution will concern us in § 4.1.
3.3. Lesbian ‑μεναι
29Let us start with a caveat. It is customary to suppose without further ado that Homeric forms like ἴδμεναι ‘know’ and γνώμεναι ‘recognize’ are of Lesbian origin. The reason is that the ending ‑μεναι is attested epigraphically almost exclusively in Lesbian, not in other Greek dialects.41 This ascription of the epic ending to Lesbian need not be wrong, but it must be taken into account that Homeric infinitives in ‑μεναι can be productive inner‑epic extensions of forms in ‑μεν.
30Epigraphically, ‑μεναι is attested in Lesbian only for the three stems ἔμμεναι, θέμεναι and δόμεναι.42 Common to these forms is that their roots are monosyllabic and have a short vowel.43 It is also of interest that ἔμμεναι, θέμεναι and δόμεναι all occur both as simplices and in compounds. It is conceivable that the extended ending ‑μεναι was first generalized in simplices, and only later spread to compounds, but there is no direct evidence for this in Lesbian.44
31Of these infinitives, only ἔμμεναι is attested in literary Lesbian. In addition, an intriguing form εἰσανβάμεν’ ‘mount into’ (Sapph., frg. 58), with metrically guaranteed ‑βᾰ‑, has come to light in the Cologne Sappho papyrus.45 As this form is of considerable importance for my argument, I will now discuss it in some detail. At first sight, the short root vowel is unexpected: in Attic-Ionic and in Homer, root aorists normally do not show root ablaut, while the three kappa aorists (ἔθηκα, ἧκα, ἔδωκα) do have root ablaut. Contrast:
| 1st sg. indic. | ἔβην | vs. ἔδωκα |
| 1st pl. indic. | ἔβημεν | vs. ἔδομεν |
| imper. 2nd sg. | βῆθι | vs. δός (probably < PGr. *dóthi)46 |
| inf. | βῆναι | vs. δοῦναι (< *dóhenai) |
32How should we interpret the short root vowel of Sappho’s ‑βάμεν(αι), as against Att.‑Ion. βῆναι, Homeric βήμεναι? In fact, ‑βάμεν(αι) is at odds also with the behavior of the aorist stem στᾱ‑ ‘to stand’ in Lesbian. This appears epigraphically in the prefixed infinitive προσταν (IG XII Suppl., no. 137, early 2nd c. BCE). Moreover, Sappho (frg. 138, 1) has an imperative στᾶθι (which mirrors Att.‑Ion. στῆθι). This raises the question why the root aorists descending from Proto-Greek *stā‑ and *gwā‑ are inflected in two divergent ways in Lesbian.
33In reconstructions of the root aorist paradigm, an important role has been played by the Homeric 3rd dual indic. βάτην. It is plausible that βα‑ reflects an old zero‑grade *gwh2‑; in the productive root aorist paradigm the long-vowel stem was generalized, but only after Proto-Greek. Alternatively, Harðarson 1993, pp. 151‑152, thinks that βάτην reflects *gwm̥‑, from the originally ablauting paradigm of PIE *gwem‑ (cf. Ved. 3rd sg. ágan) while ἔβην would stem from a different, non‑ablauting root *gweh2‑. Again others have viewed βάτην as an artificial coinage of Epic Greek, based on the root of βάσκω.47
34However, the discovery of Sappho’s ‑βάμεν(αι) renders a secondary origin of Homeric βάτην highly unlikely. As Bettarini 2005 notes, in ‑βάμεν(αι) the relic ending ‑μεναι again occurs in combination with a light monosyllabic root. This must be contrasted with the behavior of στᾱ‑ in Lesbian, with inf. προσταν (epigraphic) and imper. στᾶθι (Sappho). Thus, ‑βάμεν(αι) strongly supports the idea that Homeric βάτην is an archaism, possibly an Aeolism, continuing an ablauting paradigm (whether one reconstructs the root as *gwm̥‑ or *gwh2‑).48
35This morphological difference between ‑βάμεν(αι) and προστᾱν, στᾶθι asks for an explanation. In Attic-Ionic, root aorists form a closed group (ἔβην, ἔστην, ἔφθην, ἔτλην, ἔδρᾱν, ἔγνων, ἐβίων, ἑάλων, ἔδυν, ἔφυν and a few others). The members of this group are characterized by their mostly intransitive (inagentive) semantics and the fact that their (mostly monosyllabic) stem ends in a long vowel. In fact, descriptively, their conjugation is virtually identical to that of aorists in ‑(θ)η‑, the main difference being that the latter necessarily have a polysyllabic stem. The features shared by these types of aorist (the stem-final long vowel, the athematic conjugation and the inagentive semantics) may have drawn a number of originally ablauting root aorists (such as βη‑ beside βα‑) into the (θ)η‑aorist conjugation. That the same change did not affect the three “kappa aorists” has various causes: their zero grade stem was protected by middle forms, they had an extension ‑κ‑ in their singular active indicative forms, and they were transitives, and thereby set apart from the ‑(θ)η‑aorist semantically.49
36In sum, the short root vowel of ‑βάμεν(αι) clearly represents an archaism, and the Lesbian evidence suggests that the difference between ‑μεν‑αι and ‑ν < *‑hen correlates with the type of conjugation. The (θ)η‑aorist paradigm, to which the long-vocalic root aorist paradigm was usually assimilated, never had ablaut and originally formed its infinitive with *‑hen (see supra § 3.2). In Lesbian, this is reflected in forms like μεθύσθην and in (secondarily transferred) στᾶθι, προστᾱν. On the other hand, monosyllabic aorist and present stems with paradigmatic ablaut used the morpheme *‑men (extended to ‑μεναι in Lesbian ἔμμεναι, δόμεναι, θέμεναι, ‑βάμεναι). I propose that with this distribution, Lesbian preserves an archaism from Proto-Greek: *‑men was used with ablauting stems, while *‑hen was originally limited to non‑ablauting stems.50
37Seen in this light, the scenario that ‑μεναι is a contamination due to linguistic contact with Eastern Ionic appears to be highly implausible. Most importantly, such a contact scenario does not explain why ‑αι occurs only in a few high-frequency verbs, and why it is not found in productive stem types such as the aorists in ‑(θ)η‑ (where Eastern Ionic actually had ‑(θ)ῆναι, while Lesbian retains ‑θην).
3.4. Lesbian ‑ν
38As far as can be checked, athematic infinitives in ‑ν are attested in Lesbian for all other types of stem (polysyllabic; monosyllabic ending in a long vowel).51 Representative examples are:
| Stem type | Lesbian | Attic |
| νυ‑present | ομνυν | ὀμνύναι |
| reduplicated present | διδων | διδόναι |
| pres. of verba vocalia | καλην | καλεῖν |
| root aorist | προσταν | προστῆναι |
| inagentive aorist | μεθύσθην (Alc.) | μεθυσθῆναι |
39Three main explanations of the origin of this ending have been put forward.52
- (1) ‑ν (with lengthening of preceding vowels, if short) spread by analogy from thematic ‑ην to athematic verbs. Wackernagel 1914, p. 103, followed by García Ramón 1990, p. 160, proposed the following proportional analogy with the thematic 2nd plural imperative in ‑ετε:
Als eine relativ jüngere Zeit das Bedürfnis empfand, auch von nicht wurzelhaften Präsensstämmen und den passiven Aoriststämmen Infinitive zu bilden, nahm das Äolische die Infinitive der Verba auf ‑ω zum Muster; nach λέγην: λέγετε bildete es δίδων zu δίδοτε, κέρναν zu κέρνατε, ὄμνυν zu ὄμνυτε, μεθύσθην zu μεθύσθητε.
- (2) ‑ν originated in the infinitive in ‑ην of verba vocalia in ‑ημι: see Schwyzer 1939, p. 807; Hamm 1957, pp. 171‑172; Blümel 1982, p. 209; Hodot, Dialecte éolien Asie, pp. 158‑159; Minamimoto 2015, p. 137. Since the 2nd and 3rd singular indicative forms of the Lesbian type κάλημμι look like thematic forms (κάλεις, κάλει like φέρεις, φέρει), the assumption generally made is that the infinitive κάλην was originally thematic, *καλέ‑ην.53 This form would have been contracted into κάλην, which was then reinterpreted as athematic κάλη‑ν. Subsequently, ‑ν would have spread to athematic aorists (προσταν, μεθύσθην) and present stems (διδων, ομνυν). For this analogical extension, Hodot, Dialecte éolien Asie, p. 159 proposes a proportion with the 3rd plural imperative, which had an athematic ending (καλέντον: κάλην = τιθέντον: X).
- (3) ‑ν directly reflects PGr. *‑hen (as attested in Myc. te‑re‑ja‑e) in the Lesbian types καλην (cf. Arc. κοινανην), μεθύσθην (cf. WIon. ˚σωθην) and προσταν (Taillardat 1960; Dubois, Dialecte arcadien, p. 177).54
40Note that all three scenarios assume that athematic ‑ν is analogical in polysyllabic present stems like ὀμνυ‑ and διδω‑.
41Scenario 1 is subject to considerable difficulties. After stems ending in a non‑ablauting long vowel, the ending ‑ν was not well‑marked (cf. κάλην, μεθύσθην, ‑στᾱν). In particular, the following question must be posed: if the original Aeolic athematic ending was ‑μεν, as is usually assumed, why were forms like *διδόμεν, *μεθυσθῆμεν, *στᾶμεν not extended to *διδόμεναι, *μεθυσθήμεναι, *στάμεναι, instead of shortening them? The ending ‑μεναι was much better marked than ‑ν, and it already occurred in various high‑frequency athematic verbs.
42Concerning present stem infinitives like δίδων and ὄμνυν, Wackernagel 1914, p. 103 noted: “Vielleicht gingen diesen Bildungen solche mit ‑μεν voraus. Aber ein διδόμεναι oder ζευγνύμεναι dem Altäolischen zuzutrauen haben wir gar keinen Grund”. With this remark, he seems to recognize the plausibility of διδόμεναι and ζευγνύμεναι as analogical reshapings of earlier forms in ‑μεν. However, since we do not in fact find διδόμεναι and ζευγνύμεναι, J. Wackernagel finds himself forced to suggest that Proto-Greek may have lacked infinitives for derived tense-aspect stems of the same verb.
43García Ramón 1990, p. 160, considers J. Wackernagel’s idea an “entscheidender Gewinn”, presumably because it circumvents the problem just described: how and why did Lesbian get rid of all infinitives in ‑μεν? However, the view that infinitives of derived stems were created ex nihilo in Lesbian is problematic: an infinitive in ‑σ(σ)αι must be reconstructed for the sigmatic aorist in Proto-Greek, and Myc. te‑re‑ja‑e is formed specifically to the present stem of the denominative verb *teleiā‑. And if denominative verbs already had separate present and aorist infinitives in Proto-Greek, the same was probably true in the case of primary verbs.
44The main disadvantage of J. Wackernagel’s scenario, then, is that it does not explain why the badly marked ending ‑ν was preferred over the well-marked alternative ‑μεναι. The spread of ‑ν would be more plausible if a precursor of this infinitive ending was already present from the start in certain frequent types of athematic stems. Various scholars have therefore taken the infinitive of athematic verba vocalia of the type κάλημμι (inf. κάλην) as pivotal: scenario 2.55 However, although I agree that the infinitive in ‑ν is old in this type, it is unclear whether it can reflect thematic ‑ην. The 2nd and 3rd singular indicative forms κάλεις, κάλει do not have to be thematic: they can be explained (with Willi 2012) from generalized sandhi variants of athematic forms in *‑ē‑si, *‑ē‑ti. If so, the Proto-Greek paradigm of such verba vocalia was simply based on non‑ablauting stems in *‑ē‑. Another problem is the mechanism by which ‑ν allegedly spread from verba vocalia to athematic aorists, especially if we assume that these originally had infinitives in *‑μεν. It is implausible that the 3rd plural imperative served as the pivotal form for such a drastic reshaping (as suggested by Hodot), as it occupies a marginal place in the verbal paradigm.
45These problems disappear in scenario 3, that is, if we assume that Proto-Greek had the ending *‑hen in all non‑ablauting stems ending in a long vowel. The basis for Lesbian ‑ν would consist of the present stems of verba vocalia plus all athematic aorists (apart from light monosyllabic roots). It would be understandable if this ‑ν (preceded by a long vowel) ousted ‑μεν, which was originally found in stems with ablaut, in forms like δίδων, ὄμνῡν for older *διδόμεν, *ὀμνύμεν. Another indication that Lesbian ‑ν reflects *‑hen can be seen in the distribution of the extended ending ‑μεναι. If the sole original athematic ending in Proto-Aeolic were ‑μεν, it is difficult to understand why *δόμεν was refashioned as δόμεναι, while putative *στᾶμεν was reshaped to give στᾶν (cf. attested προσταν) rather than *στάμεναι. The view that aorist infinitives like *δῶν were avoided because they were monosyllabic (Scarborough 2023, p. 237) is implausible for two reasons. First, it does not explain why προσταν actually occurs in Lesbian,56 and secondly, there was no general ban on monosyllabic words in Greek, but only on ultra-light lexical monosyllables of the structure C(C)V̆ such as *σχέ (cf. the generalization of the augmented indicative ἔσχε, and the analogical addition of ‑ς in the imperative σχές).57
46I conclude that Lesbian athematic infinitives in ‑ν reflect Proto-Greek *‑hen after stems ending in a long vowel: this included inagentive aorists (including non‑ablauting root aorists) and athematic verba vocalia.
3.5. The Homeric athematic infinitives
47It is normally admitted that the Homeric endings ‑μεν and ‑μεναι are Aeolisms, and that is surely what they are in forms like δόμεν and ἔμμεναι. Unexpectedly, however, certain athematic stems in Homer use the infinitive ending ‑μεν where one would expect a metrically equivalent Ionic form in ‑ναι (e.g. ζευγνύναι, ἑστάναι). In the present stem, leaving aside the frequent ἴμεν ‘go’ (also ἰέναι, ἴμεναι), three athematic verbs take the shorter ending ‑μεν:
- ζευγνύμεν ‘pair’ (Il., XV, 120) alongside ζευγνύμεναι (Il., III, 260), also metrically lengthened ζευγνῦμεν (Il., XVI, 145);
- ὀρνύμεν ‘raise’ (Il., IX, 353; Od., X, 22) alongside ὀρνύμεναι (Il., XVII, 546);
- μεθιέμεν ‘slacken’ (Il., IV, 351) alongside μεθιέμεναι (Il., XIII, 114), both with metrically lengthened ‑ι‑ (Att. μεθιέναι).58
48On the other hand, three other athematic present stems use ‑ναι:
- κιχῆναι ‘reach’ (Od., XVI, 357) alongside κιχήμεναι (Il., XV, 274);59
- ἀῆναι ‘blow’ (Od., III, 183, X, 25) alongside ἀήμεναι (Il., XXIII, 214; Od., III, 176);
- φορῆναι ‘carry’ (4x) alongside φορήμεναι (Il., XV, 310).
49The form φορῆναι (a non‑Ionic feature: Ionic has thematic φορέειν, φορεῖν) is an athematic contract verb infinitive, a type that displays no ablaut in Aeolic dialects and Arcadian. The verb ἄημι is non‑ablauting, too, as shown by 3rd dual pres. ἄητον (Il., IV, 5), 3rd plural pres. ἄεισι (Hes., Th., 875) instead of ἀεῖσι (cf. ἱεῖσι, τιθεῖσι), and the middle forms ἀήμενος, ἄητο.60 Thus, these Homeric infinitives suggest that ablauting present stems (‑νυμι, ἵημι) use the ending ‑μεν, while non‑ablauting athematic stems use ‑ναι.
50Apart from μεθιέμεν, there is little evidence in Homer for the infinitive of athematic reduplicated presents, and no examples of the classical forms διδόναι, τιθέναι, ἱστάναι, even if these would have been metrically useful.61 This rarity can be due to the telic actionality of such verbs, but it remains remarkable. The odd form διδοῦναι occurs once (Il., XXIV, 425) at verse end; however, it can be analyzed as an artificial hybrid of aorist inf. δοῦναι with a phrase like δῶρα διδοῦσιν (Od., XVIII, 279), and does not presuppose the existence of διδόναι.62 As for τίθημι, an infinitive τιθήμεναι occurs twice in Homer (Il., XXIII, 83 and 247);63 although Chantraine 1958, p. 104, considers it an artificial form, its ‑η‑ might reflect a linguistically real feature, stemming from an (Aeolic) dialect that transferred the reduplicated athematic presents into the class of athematic verba vocalia.64 This interpretation also seems plausible for κιχῆναι beside κιχήμεναι, as the paradigm of κίχημι largely coincides with that of τίθημι in Homeric Greek.
51Perfect infinitives are much more frequent than athematic present infinitives in Homer; without exception, they end in ‑μεν or ‑μεναι: cf. δειδίμεν, ἑστάμεν(αι), τεθνάμεν(αι), τετλάμεν, ἐκγεγάμεν, βεβάμεν, etc. Familiar classical Ionic and Attic infinitive forms such as εἰδέναι, ἑστάναι, τεθνάναι are completely absent, even if they were metrically equivalent to forms in ‑μεν.65 Instead of εἰδέναι, Homer uses both ἴδμεναι and ἴδμεν.
52This unexpected absence of ‑ναι in the perfect stem is usually neglected, but it was noted by Chantraine 1958, p. 488, who also remarked that perfect infinitives in ‑μεν, ‑μεναι are never attested in Aeolic inscriptions. Of course, non‑finite perfect forms in Aeolic dialects follow the thematic conjugation, and it is generally assumed that this feature was introduced early on: the thematic participle occurs in Boeotian, Thessalian and Lesbian, while the thematic infinitive is found in Thessalian (εζαλουκεµεν) and Lesbian (τε[θ]νακην, επιτεθεωρηκην); no perfect infinitives are attested in Boeotian.66 Even if one assumes that Aeolic preserved athematic perfect infinitives in ‑µεν before the Lesbian migration, there is no particular reason to view such forms in Homer as Aeolisms.
53Given that the Classical Attic-Ionic present and perfect infinitives in ‑ναι can be secondary creations (see supra § 3.1), the following conclusions suggest themselves:
- ‑μεν was the oldest infinitive ending in the perfect stem and in ablauting athematic present stems (ζευγνύμεν, ὀρνύμεν, μεθιέμεν);
- ‑μεν was still current in these stems in (Eastern) Ionic contemporary to Homer.67
54Before building on these conclusions, however, we must check whether ‑μεν cannot have been introduced artificially in the Homeric perfect and present stem infinitives.
55The point is that the choice between ‑μεν and ‑ναι in Homer seems governed by the length of the stem-final vowel: ‑ναι occurs almost exclusively after long vowels, while ‑μεν is found exclusively after short vowels.68 Now, a scenario involving secondary generalizations of ‑μεν and ‑ναι has been envisaged by Peters 1986, p. 308 and Peters 1989, pp. 97‑99. M. Peters proposes that Proto-Greek had *‑hen infinitives in polysyllabic stems, and that this ending (which was obsolete in the vernacular and threatened by contraction) was replaced in Homer by ‑μεν or ‑ναι depending on the length of the preceding vowel. He therefore reconstructs forms like *histáhen, *hestáhen for pre‑stages of Attic-Ionic and the epic tradition, and suggests that these perfect infinitives in *‑hen were consistently replaced within the tradition by ‑μεν.
56This scenario, however, does not explain why we find not a single trace in Homer of the alleged synchronic (Eastern or Western) Ionic perfect infinitive, whether this ended in long vowel plus ‑ν (*ἑστᾶν), in short vowel plus ‑εν (*δεδϝίεν) or already in ‑ναι (*ἑστᾶναι or ἑστάναι).69 It does not account either for the forms μεθιέμεν, ζευγνύμεν, ὀρνύμεν. That such present-stem infinitives are rare does not allow us to leave them out of account. If anything, one expects the vernacular form to be used especially in non‑formulaic contexts. I therefore conclude that infinitives in ‑μεν remained current in ablauting present stems and in perfects at least until the late 8th c. BCE in spoken Eastern Ionic, and that the replacement of ‑μεν by ‑ναι in these categories postdates Homer.
57Let me comment briefly on the substitutions made in Epic Greek. Most epic forms that were introduced from the Ionic vernacular have not yet ousted the corresponding traditional Aeolic forms, but are used alongside them. For instance, δοῦναι is used only thrice in metrical positions where δόμεναι (27x) could also be used; ἰέναι ‘go’ competes with metrically equivalent ἴμεναι, and εἶναι is clearly a recent introduction for ἔμμεν (6x). Interestingly, εἶναι always has the value of a trochee (like ἔμμεν) and does not compete with anapestic ἔμεναι, the more traditional epic form for which it could have been (but was not) substituted.
58On the other hand, there are substitutions in which productive rules of the language were applied. A morphophonological rule governing the distribution between ‑μεν and ‑ναι may explain why we find not only Ionic aorist infinitives in ‑ναι (στῆναι, μιγῆναι, κοιμηθῆναι), but also oddities like φορῆναι ‘to carry’. This present infinitive is important because it occurs four times in traditional formulaic contexts. However, it can neither be of Ionic origin (because the iterative type φορέω, inherited from PIE, never switched to the athematic inflection in Ionic), nor of “Achaean” origin (because Arcadian inscriptions preserve infinitives in ‑ην to verba vocalia).70 Therefore, φορῆναι must have secondarily acquired ‑ναι (the ending used after long vowels); before Homer, the form undoubtedly had a different shape, either φορῆμεν or *phorē̃hen. A similar explanation accounts for ἀῆναι ‘blow’ and κιχῆναι ‘reach’, two traditional epic lexemes; conversely, *e(h)en ‘be’ was recharacterized as ἔμεν rather than *enai because only ‑μεν was used after short vowels.
59How did this rule governing the Homeric distribution between ‑μεν and ‑ναι come into being? I propose that it directly reflects the situation as it existed in Homer’s Ionic vernacular. There, ‑ναι was almost always preceded by a long vowel, which reflected the contraction of short vowel plus *‑enai (cf. εἶναι, θεῖναι, δοῦναι, εἷναι). On the other hand, we find that ‑μεν always comes after short stem-final vowels in Homer. This follows, I propose, from the fact that ‑μεν was attached to the short-vowel stem of ablauting verbs, originally and still in the contemporaneous Ionic vernacular.
60Finally, some remarks about the problematic form εἰδέναι ‘know’. This Attic-Ionic form is first attested after Homer, but it cannot be a post‑Homeric analogical creation, as there would be no plausible model form ending in ‑έναι (which had always undergone contraction except in ἰέναι). We must therefore assume that εἰδέναι is an old form in Attic-Ionic. This allows us to explain the productivity of ‑έναι in perfect infinitives (e.g. λελυκέναι, βεβλαφέναι), based on a proportion with the participle εἰδώς (which has the same root vocalism as εἰδέναι). This means, however, that Homer used εἰδέναι in his vernacular, but did not introduce the form into Epic Greek. A possible reason for this is that εἰδέναι was completely isolated with respect to the perfect infinitives in ‑μεν(αι), which were productive in Homer (cf. also Hom. ἴδμεν, ἴδμεναι).
61There are two possible reconstructions of εἰδέναι. First, it could reflect a South Greek innovation *wíd‑hen(ai) that replaced earlier *wíd‑men(ai).71 Since οἶδα was the only non‑reduplicated perfect, it would make sense if only *wíd‑men was replaced by *wíd‑hen in Proto-Ionic (again motivated by homophony with the 1st plural indicative), while the ending ‑μεν was retained in polysyllabic perfect stems like ἑστάμεν. This retention was unproblematic because the accentuation of forms like ἑστάμεν differed from that of the corresponding 1st plural indicative ἕσταμεν. Secondly, εἰδέναι could reflect *wid‑wénai, an archaism retained only in Attic-Ionic that would be comparable to Old Avestan vīduuanōi (see García Ramón 2009; García Ramón 2020, p. 171). If the second alternative is followed, *wíd‑wenai would be an isolated retention in an irregular high-frequency verb, and thus would not affect the reconstruction of *‑men(ai) as the regular Proto-Greek ending of ablauting athematic stems.
4. The original distribution between *‑hen and *‑men
62Let us now recapitulate our findings concerning the original distribution between ‑μεν and *‑hen, and investigate whether the forms attested in South Greek dialects and Lesbian can be plausibly derived from this situation. My two main claims are:
- *‑men was at home in ablauting stems, while *‑hen was found in non‑ablauting stems;
- whether or not the extension ‑αι was generalized depended on considerations relating to Wortumfang.
63Thus, the oldest core of athematic stems with *‑hen infinitives consisted of presents of verba vocalia (Myc. te‑re‑ja‑e; Lesb. κάλην, φοίταν; Arc. αντευεργετην, κοινανην), inagentive aorists in ‑(θ)η‑, and non‑ablauting root aorists. The last‑mentioned type expanded when originally ablauting root aorists were transferred into the conjugation of inagentive aorists in ‑(θ)η‑. Presumably, this transfer affected all root aorists in Proto-Ionic, but Lesbian apparently underwent this development only in part (cf. ‑βάμεν(αι) in Sappho).
64Since I argue that ‑έναι reflects *‑hen plus ‑ai, the main remaining challenge is to show how and why *‑hen(ai) spread in Proto-South Greek to monosyllabic ablauting verbs, including at least the present stems *éhen(ai), *íhen(ai) and the kappa aorists *théhen(ai), *dóhen(ai), *i̯éhen(ai). That *‑hen(ai) is secondary in such stems in South Greek is suggested by the appearance of ‑μεναι in the corresponding Lesbian forms (ἔμμεναι, θέμεναι, δόμεναι), which implies that the Proto-Aeolic (and, I assume, Proto-Greek) forms had *‑men. So: what was the motive for substituting *‑hen for *‑men? And why was the form in ‑ai subsequently generalized in most varieties of South Greek?
4.1. From Proto-Greek to South Greek and Attic-Ionic
65The following table (tab. 2) summarizes the changes in the phonology and morphology of infinitives taking place between Proto-Greek and the classical language; innovations at each stage are given in boldface:
Tab. 2 – The athematic infinitives from Proto-Greek to Classical Attic and Ionic.
| *‑hen | Proto-Greek | South Greek 1 | South Greek 2 | 8th c. Attic | Class. Attic |
| root aorist | *stā́‑hen | *stā̃hen | *stāénai, *‑stāen | στῆναι | στῆναι |
| (θ)η‑aorist | *phanḗ‑hen | *phanē̃hen | φανῆναι | φανῆναι | |
| present of verba vocalia | *nīkā́‑hen | *nīkā̃hen | → thematic | ||
| *‑men | |||||
| root present | *éh‑men *í‑men | *éhen *íhen | *eénai, *‑éen *iénai, *‑íen | εἶναι ἰέναι | εἶναι ἰέναι |
| kappa aorist | *dó‑men | *dóhen | *doénai, *‑dóen | δοῦναι | δοῦναι |
| polysyllabic present | *didó‑men | *didómen | διδόμεν | διδόναι | |
| perfect | *hestá‑men | *hestámen | ἑστάμεν | ἑστάναι |
66In the first two columns of tab. 2, extended forms in ‑ai are not expressly indicated. I assume that such extensions could be made to any infinitive in Proto-Greek without morphological restrictions, and originally with some specific functional value.
67At stage South Greek 1, *‑hen was substituted for *‑men after monosyllabic stems containing a short root vowel. We may assume that *éhmen ‘be’ was changed into *ehen rather than *ehhen because /hh/ was illicit phonologically.72 Why did this change take place? In Proto-Ionic, inf. ‑μεν had become identical to the 1st plural indicative ending ‑μεν (probably from *‑mem, with 1st person marker ‑m). At this stage, the unaugmented indicative with secondary endings (often called “injunctive”) was still the unmarked simple past form. The augment, which preserved its functional value, could not be used to disambiguate between, for instance, *dómen ‘we gave’ and *dómen ‘to give’. Thus, I propose that homophony was the factor that drove these replacements in Proto-Ionic, as it did in pre‑Lesbian.
68Indeed, there were situations where ambiguity would arise between the 1st plural indicative and the infinitive.73 Moreover, it cannot be a coincidence that precisely in West Greek dialects, where the primary 1st plural marker ‑μες has been generalized at the expense of *‑me(n), ‑μεν becomes predominant as an athematic infinitive ending. Presumably, both developments are early innovations of West Greek.74 That these dialects, but also Thessalian and Boeotian, completely lost the athematic infinitive ending *‑hen (perhaps at a relatively late stage) could have various reasons; one possibility is that the loss of intervocalic ‑h‑ and subsequent contractions made this ending less transparent than ‑μεν.75 Obviously, there was no need to generalize the extended form in ‑ai.
69In Arcadian, both ‑μες and ‑μεν are actually attested as 1st pl. endings (Dubois, Dialecte arcadien, pp. 177‑178), and various scenarios could explain this situation. For instance, ‑μες may have been introduced under the influence of neighboring West Greek dialects, or the Proto-Greek distribution between the primary and secondary endings was preserved relatively long in Arcadian, before local varieties generalized one or the other. In both cases, the Arcadian evidence is compatible with the suggestion that the spread of *‑hen after monosyllabic stems in Proto-South Greek was driven by an avoidance of homophony between the infinitive and the secondary 1st plural indicative ending ‑μεν.
70When the function of ‑ai was lost, this element was incorporated into the infinitive ending *‑hen first with monosyllabic stems (South Greek stage 2). Again, the avoidance of ambiguity may have played a role in motivating this change. Due to the disappearance of intervocalic ‑h‑, monosyllabic infinitive forms like *éhen, *íhen, *théhen, *dóhen became identical to unaugmented 3rd plural past indicative forms, which had originally ended in PIE *‑ént but had lost their final obstruent and (being finite verb forms) generalized recessive accentuation.
71We have no direct evidence for unaugmented 3rd pl. indic. forms *íen, *théen, *dóen, *i̯éen, but there is reason to assume that such forms underlie the secondary Attic-Ionic forms in ‑σαν, attested from Homer onwards. While in Arcadian and Cypriot the kappa aorist has a disyllabic 3rd pl. form in ‑αν (e.g. Cypr. ka‑te‑ti‑ja‑ne, Arc. συνεθεαν), this was clearly not the only Proto-Greek ending. There is direct evidence for a 3rd pl. secondary ending ‑εν in the athematic optative (e.g. δοῖεν ‘may they give’), and in the 3rd sg. form ἦεν ‘was’ (reanalyzed from an original 3rd pl. form in Proto-Ionic). Moreover, as primary endings we find *‑enti (Myc. ki‑ti‑e‑si, e‑e‑si, Ion. διδοῦσι < *didó‑enti) beside ‑αντι > ‑ᾱσι (Att. τιθέᾱσι, διδόᾱσι). Following the argumentation of Kortlandt 1988, I assume that *‑en is old after unaugmented monosyllabic stems (*dó‑en), while *‑an was used in augmented forms and after polysyllabic stems (*é‑do‑an, *(e‑)dído‑an).76
72Judging by the evidence from Western Ionic, Arcadian and Lesbian discussed above, considerations relating to word size may also have promoted the generalization of forms in ‑αι: ἰέναι, εἶναι were much more clearly marked than *ἴεν, εἶν (although the latter is attested relatively early in Stesichorus). In compounds, forms like ἐξεῖν and ἐξίεν77 were retained more widely in Arcadian and Western Ionic.
73Still in the Dark Ages, in most varieties of Attic-Ionic, the extension ‑ai spread further to the inagentive aorists and to all other stems whose infinitive originally ended in *‑hen. This gives us the 8th c. Eastern Ionic situation contemporaneous to Homeric Greek. This development was limited to Attic and Eastern plus Central Ionic; it originally happened neither in Western Ionic (cf. ‑σωθῆν, ἐξεῖν), nor in Arcadian (cf. κοινανην, αντευεργετην), nor in Lesbian (cf. μεθύσθην).
74A final change took place after Homeric Greek: ‑ναι was reanalyzed as the ending and spread to polysyllabic athematic present and perfect stems, replacing the final remains of ‑μεν.
4.2. From Proto-Greek to Lesbian
75The phonological and morphological changes taking place in the prehistory of Lesbian are summarized in tab. 3. The three developments (with innovations again given in bold) are relatively straightforward: the extended form ‑μεναι was preferred over *‑men in monosyllabic root formations; ‑ν < *‑hen spread to all polysyllabic stems; and the perfect infinitive was thematicized:
Tab. 3 – The athematic infinitives from Proto-Greek to Lesbian.
| Proto-Greek | Pre-Lesbian 1 | Pre-Lesbian 2 | Attested Lesbian | |
| *‑hen | ||||
| root aorist | *stā́‑hen | *stā̃hen | *stā̃n | προ‑σταν |
| (θ)η‑aorist | *phanḗ‑hen | *phanē̃hen | *phánēn | μεθύσθην |
| present of verba vocalia | *nīkā́‑hen | *nīkā̃hen | *nī́kān | φοίταν |
| *‑men | ||||
| root present | *éh‑men | *éhmenai | *émmenai | ἔμμεναι |
| kappa aorist | *dó‑men | *dómenai | δόμεναι | |
| polysyllabic present | *didó‑men | *didómen | *dídomen | διδων |
| perfect | *hestá‑men | *hestámen | → thematic | τεθνάκην |
76As already mentioned in § 2, there is a plausible motive for an early generalization of the allomorph ‑μεναι after monosyllabic stems containing a short vowel: the avoidance of homophony with unaugmented 1st plural indicative forms in ‑μεν (such as δόμεν ‘we gave’). It is natural to make the following assumptions for an early pre‑stage of Lesbian: (i) recessive accentuation had been generalized in finite verbs, but not yet generally; (ii) augmented secondary indicative forms coexisted with their unaugmented counterparts; (iii) the infinitive in ‑μεν had penultimate accentuation (as it still has in Homer).78 In this situation, infinitives of polysyllabic stems (like διδόμεν, ὀμνύμεν) were still distinct from the 1st pl. pres. (and unaugmented impf.) indicatives δίδομεν, ὄμνυμεν. This explains why they at first retained their original form.
77After the early generalization of ‑μεναι exclusively after monosyllabic stems (at stage pre‑Lesbian 1), the extended allomorphs in ‑αι lost their productivity. Moreover, the dialect generalized phonologically recessive accentuation, probably after the migration to the Eastern Aegean. At this stage (pre‑Lesbian 2), another problem presented itself. Infinitives of polysyllabic athematic presents became homophonous with the 1st plural present indicative (inf. διδόμεν, ὀμνύμεν merged with 1st pl. indic. δίδομεν, ὄμνυμεν). This ambiguity was resolved by generalizing the only remaining productive infinitive morph, the contraction product of *‑hen with a preceding vowel, which always appeared as ‑ν following a long vowel. This led to the analogical creation of attested forms like διδων, ομνυν.
78This new scenario accounts for the attested distribution of ‑μεναι versus ‑ν in a natural way. Both ‑μεναι and ‑ν are archaic features. While ‑μεναι was retained only in relics, the productive use of ‑ν was analogically extended. The reason why forms like *διδόμεναι, *ὀμνύμεναι were never created (the problem noted by J. Wackernagel) may be found in chronology: the ending ‑μεναι had lost its productivity too early.
79Hitherto, infinitives in ‑men were thought to have been lost from all varieties of South Greek well before our earliest sources. Similarly, no variety of West Greek, Thessalian or Boeotian has reflexes of athematic *‑hen. This was the main reason for scholars to consider the Lesbian infinitive endings ‑ν and ‑μεναι secondary innovations. In reality, the restriction of ‑μεναι in Lesbian to monosyllabic roots containing a short vowel is best explained as a retained archaism. Moreover, the presence of ‑ν (reflecting *‑hen) in non‑ablauting stems can be another archaism inherited from Proto‑Greek.
80Finally, while Lesbian (unlike South Greek dialects) did not take part in the early spread of *‑hen to monosyllabic ablauting stems, it does share with all South Greek dialects the fact that the infinitives of such verbs have an extended form in ‑ai. Probably, the motivation was similar (homophony avoidance). This use of the ‑ai extension could be a convergent development of pre‑Lesbian and South Greek dialects, but is not necessarily a shared innovation.
5. The position of Lesbian and contact with Eastern Ionic
81Lesbian is normally thought to belong to an Aeolic subgroup together with Thessalian and Boeotian.79 In favor of this claim speak several non‑trivial innovations or choices shared by these three dialects, including:
- the unconditioned vocalization *r̥ > ρο (analogically ορ);80
- a labial reflex of labiovelars before e;
- *ri > re;
- the creation of a dative plural in ‑εσσι, and its spread throughout the 3rd declension;
- thematic non‑finite perfect forms;
- a sigmatic aorist stem of disyllabic roots with ‑σσ‑ (type ἐκάλεσσα);
- certain lexical choices, such as fem. ἴα ‘one’ instead of μία.
82Lesbian deviates from Thessalian and Boeotian in important respects, including:
- it has assibilation *ti > σι in roughly the same categories as in Attic-Ionic, including athematic and thematic verbal endings (3rd sg. ‑σι, 3rd pl. ‑οισι, ‑ισι), numerals, and productive derivations (e.g. adjectives in ‑σιος to stems in ‑τ‑);81
- psilosis;
- the infinitive endings;
- the prepositions/preverbs πρός and εἰς, ἐς (as in Ionic), as against ποτι, πο(τ)‑ and ἐν in Boeotian and Thessalian;
- βόλλομαι ‘want’, allegedly a contamination of pre‑Lesbian βέλλομαι and Ionic βούλομαι (Porzig 1954, p. 153).
83Within the Risch and Porzig classification of the Greek dialects,82 it is supposed that pre‑migration Lesbian started off as a “pure” Aeolic dialect, and that it was heavily influenced by Eastern Ionic in the centuries following the migration to Asia Minor, which is thought to have taken place around the turn of the first millennium BCE.83 However, while there are clear traces of Ionic influence on Lesbian in the inscriptional record (see generally Hodot, Dialecte éolien Asie, p. 25), early inscriptions from Lesbos and the Aeolid are scarce, starting only in the late classical period. The poetic language of Sappho and Alcaeus contains many Ionic forms (e.g. καρτερός ‘strong’ with ‑αρ‑ < *r̥), but these can be ascribed to the influence of Ionic poetry (especially epic) and do not necessarily reflect an influence of the Eastern Ionic vernacular on everyday Lesbian.
84If anything, the assibilation *ti > σι in Lesbian must be interpreted as resulting from regular sound change. This change can hardly be a late adaptation of Lesbian phonology to that of neighboring Eastern Ionic resulting from language contact, because the last‑mentioned dialect (like Attic and already Mycenaean) tolerated /ti/ again after it had undergone assibilation of *t(h)i. On the other hand, it is not plausible either that linguistic contact led to a reshaping of core verbal endings (those of the 3rd person). Therefore, assibilation was probably an isogloss shared by the mid‑second millennium precursor of Lesbian with South Greek dialects.84 Certain other isoglosses shared by Lesbian and Arcado-Cyprian have traditionally been taken to prove a close relation between these dialects; they could also be ascribed to this stage.
85After these early changes, pre‑Lesbian shared all the exclusive Aeolic isoglosses with Thessalian and Boeotian mentioned above, without taking part in later characteristic common innovations of South Greek, such as temporal adverbs in ‑οτε or infinitives like *dóhen. In other words, Lesbian may have started off as a South Greek dialect before the characteristic developments that it shared with the precursors of Thessalian and Boeotian led to the formation of an Aeolic group.
86The Lesbian infinitive endings can also be reinterpreted in this light. While ‑μεναι and thematic ‑ην are often ascribed to contact with Eastern Ionic, I have argued that the distribution between athematic ‑μεναι and ‑ν to a considerable extent reflects that between *‑men and *‑hen in Proto-Greek. If we compare the developments proposed for Lesbian with those taking place in South Greek dialects, interesting differences and similarities appear. The oldest development at stage South Greek 1 (*éhmen, *dómen being replaced by *éhen, *dóhen) was not shared by Lesbian, but the oldest specifically Lesbian development (‑μεναι favored over ‑μεν after monosyllabic stems, pre‑Lesbian 1) closely resembles the generalization of forms in ‑αι in all South Greek dialects (stage South Greek 2). Not only the appearance of ‑αι (probably a retained archaism), but also its distribution (first in monosyllabic stems) and the plausible motivation for its generalization (avoidance of homophony) can be interpreted as induced by linguistic contact between pre‑Lesbian and South Greek dialects.
87This shared generalization of an ending extended with ‑ai after monosyllabic stems must go back to the Mycenaean period at least, as all South Greek dialects (including Cypr. to‑we‑na‑i, Arc. ηναι) have it. However, its exact age relative to the assibilation of *t(h)i (another common South Greek development that is shared by Lesbian) is difficult to determine.
6. Conclusions
88Assuming that the infinitive was a grammaticalized category already in Proto-Greek, with separate forms for each tense-aspect stem and voice, I have proposed a maximally economical reconstruction of the infinitive endings in Proto-Greek that allows us to account for the most important dialectal developments. The key ingredients of my reconstruction are:
- a simple distribution between two athematic infinitive endings: *‑hen was used with non‑ablauting stems, *‑men with ablauting stems;
- the idea that the element ‑ai could be added to any infinitive ending in Proto‑Greek.
89In the different dialect groups, this situation was changed for two structural reasons: (i) homophony between the infinitive and indicative endings (1st pl. ‑men, 3rd pl. ‑en) was avoided, and (ii) ‑ai became functionally opaque. Homophony avoidance caused an early substitution of *‑hen for ‑men after monosyllabic stems in South Greek dialects. The functional bleaching of ‑ai made possible its incorporation into the extended endings *‑henai (South Greek) and ‑menai (Lesbian), again allowing avoidance of homophony. I see no compelling reason to reconstruct separate endings *‑wenai or *‑nai.
90The following developments can be distinguished chronologically:
- (1) in South Greek, all monosyllabic stems (root presents, kappa aorists, perhaps also *woida ‘know’) acquired *‑hen(ai) instead of *‑men(ai);
- (2) when ‑ai lost its function, *‑(h)enai and *‑men(ai) were generalized at the expense of *‑(h)en, *‑men in monosyllabic stems (cf. εἶναι, δοῦναι, Lesb. ἔμμεναι, δόμεναι).
91This second development was still common to all South Greek dialects and may have taken place in close contact with pre‑migration Lesbian, as its generalization of ‑μεναι after monosyllabic stems is structurally similar:
- (3) in Attic and Eastern Ionic ‑(έ)ναι spread also to compound verbs and polysyllabic stems;
- (4) ‑ναι replaced the remaining instances of ‑μεν.
92Development (4) was post-Homeric. Western Ionic and Arcadian escaped (3) at first: they preserve reflexes of *‑hen in compounds and after polysyllabic stems ending in a long vowel. A relatively late replacement of ‑μεν by the reflex of *‑hen in polysyllabic present stems also occurred independently in pre‑Lesbian (cf. ομνυν, διδων). Lesbian inherited the ending ‑ν < *‑hen in non‑ablauting stems, just like various South Greek dialects (cf. Myc. te‑re‑ja‑e).
93If several dialectal developments were indeed caused by homophony avoidance, the present scenario may strengthen claims about the reconstruction of the Greek verbal endings, e.g. concerning reflexes of a secondary 3rd pl. ending *‑en in South Greek. Another important consequence is that we may make sense of the Homeric distributions between athematic ‑ναι, ‑μεν and ‑μεναι. The epic language preserves various archaic features of infinitive morphology, while also showing traces of productive replacement:
- the avoidance of ‑μεν after long vowels in Homer can be ascribed to the fact that only *‑hen (replaced by ‑ναι) was originally used in such contexts (non‑ablauting stems);
- ‑μεν in polysyllabic ablauting present and perfect stems (e.g. ὀρνύμεν, ἑστάμεν) is an archaism that was also preserved in Eastern Ionic contemporaneous with Homer.
94Thus, part of Homer’s infinitives in ‑μεν are Aeolisms, another part may stem from Ionic. Finally, the Homeric option to use disyllabic ‑μεναι instead of ‑μεν, ‑ναι may simply reflect an older situation, where ‑ai could be added to any infinitive. Originally, this addition was no doubt accompanied by a functional distinction, but this distinction had been lost well before Homer.
95All in all, this suggests that the pre‑migration precursor of Lesbian developed in close proximity to South Greek dialects, sharing some of its early characteristic innovations, including the assibilation of *t(h)i, the extended infinitive in ‑ai after monosyllabic stems, and perhaps the generalization of the 1st pl. ending ‑μεν. Obviously, it also took part in characteristic Aeolic isoglosses. Both aspects can be accounted for if the precursor of Lesbian was spoken somewhere in Boeotia or on the south-eastern fringes of Thessaly. In this case, it would have been in contact with South Greek dialects to its south-east, but also with regions to the west (and north) where other Aeolic dialects were spoken.85 The infinitive endings can no longer be considered evidence for linguistic contact between Lesbian and Eastern Ionic.
Bibliographie
Abbreviations of editions and of works of reference for alphabetic Greek epigraphy are those of the list of the AIEGL published online: GrEpiAbbr, version January 2022, https://www.aiegl.org/grepiabbr.html (accessed 06/07/2023) and abbreviations of ancient authors of the Diccionario griego-español, http://dge.cchs.csic.es/lst/lst4.htm (accessed 06/07/2023).
Bartoněk 1968: A. Bartoněk (dir.), Studia Mycenaea. Proceedings of the Mycenaean symposium, Brno, Universita, 1968.
van Beek 2022: L. van Beek, The reflexes of syllabic liquids in Ancient Greek: linguistic prehistory of the Greek dialects and Homeric Kunstsprache, Leiden, Brill, 2022.
Benveniste 1936: É. Benveniste, Origines de la formation des noms en indo-européen, Paris, Klincksieck, 1936.
Bettarini 2005: L. Bettarini, “Note linguistiche alla nuova Saffo”, ZPE 154, 2005, pp. 33‑39.
Bettarini 2007: L. Bettarini, “Note esegetiche alla nuova Saffo: I versi di Titono (fr. 58, 19‑22 V.)”, ZPE 159, 2007, pp. 1‑10.
Blümel 1982: W. Blümel, Die aiolischen Dialekte: Phonologie und Morphologie der inschriftlichen Texte aus generativer Sicht, Göttingen, Vandenhoeck & Ruprecht, 1982.
Cassio 1996: A.C. Cassio, “Da Elea a Hipponion e Leontinoi: Lingua di Parmenide e testi epigrafici”, ZPE 113, 1996, pp. 14‑20.
Chantraine 1958: P. Chantraine, Grammaire homérique, vol. I, Phonétique et morphologie, Paris, Klincksieck, 1958 (2nd ed.).
Chantraine 1961: P. Chantraine, Morphologie historique du grec, Paris, Klincksieck, 1961.
Cowgill 1964: W. Cowgill, “The supposed Cypriote optatives duwánoi and dṓkoi. With notes on the Greek infinitive formations”, Language 40, 1964, pp. 344‑365.
Ellsworth 2011: M. Ellsworth, “The first palatalization of Greek”, in S. Jamison, H.C. Melchert, B. Vine (dir.), Proceedings of the 22nd annual UCLA Indo-European conference, Bremen, Hempen, 2011, pp. 13‑31.
Finkelberg 2017: M. Finkelberg, “Lesbian and mainland Greece”, in G. Giannakis, E. Crespo, P. Filos (dir.), Studies in Ancient Greek Dialects. From Central Greece to the Black Sea, Berlin/Boston, De Gruyter, 2017, pp. 447‑456.
García Ramón 1975: J.L. García Ramón, Les origines postmycéniennes du groupe dialectal éolien, Salamanca, Universidád de Salamanca, 1975.
García Ramón 1977: J.L. García Ramón, “Le prétendu infinitif ‘occidental’ du type ἔχεν vis‑à‑vis du mycénien e‑ke‑e”, Minos 16, 1977, pp. 179‑206.
García Ramón 1990: J.L. García Ramón, “Proportionale Analogie und griechische Morphologie: Athematische Infinitive im Attischen und im Westionischen”, in H. Eichner, H. Rix (dir.), Sprachwissenschaft und Philologie: Jacob Wackernagel und die Indogermanistik heute, Wiesbaden, Reichert, 1990, pp. 150‑169.
García Ramón 1994: J.L. García Ramón, “Correspondencias y ecuables en la reconstrucción indoeuropea: chip. to‑we‑na‑i /dowenai/, véd. dā‑váne, ie. *də3‑u̯én‑”, in J. de la Villa Polo, V. Picón García, L.M. Macía Aparicio, E. Torrego (dir.), Quid ultra faciam? Trabajos de griego, latín e indoeuropeo en conmemoración de los 25 años de la Universidad Autónoma de Madrid, Madrid, Ediciones de la Universidad Autónoma de Madrid, 1994, pp. 45‑55.
García Ramón 1997: J.L. García Ramón, “Infinitive im Indogermanischen? Zur Typologie der Infinitivbildungen und zu ihrer Entwicklung in den älteren indogermanischen Sprachen”, Incontri linguistici 20, 1997, pp. 83‑92.
García Ramón 2009: J.L. García Ramón, “Formal correspondences, different functions. On the reconstruction of inflectional categories of Indo-European”, in V. Bubeník, J. Hewson, S. Rose (dir.), Grammatical change in Indo-European languages: papers presented at the workshop on Indo-European linguistics at the XVIIIth international conference on Historical Linguistics, Montreal, 2007, Amsterdam/Philadelphia, Benjamins, 2009, pp. 237‑250.
García Ramón 2010: J.L. García Ramón, “On the genetic classification of the Ancient Greek dialects: comparative reconstruction versus hypercriticism and atomism at work”, in Studies in Greek linguistics. Proceedings of the 30th annual meeting of the Department of Linguistics, School of Philology, Aristotle University of Thessaloniki, Thessaloniki, Institouto Neoellinikon Spoudon, 2010, pp. 219‑236.
García Ramón 2020: J.L. García Ramón, “Anatolian and Greek infinitives: continuity of Indo-European morphosyntax vs. areal development”, in L. Repanšek, H. Bichlmeier, V. Sadovski (dir.), vácām̆̇si miśrā́ kr̥ṇavāmahai. Proceedings of the international conference of the Society for Indo-European Studies and IWoBA XII, Ljubljana 4‑7 June 2019, celebrating one hundred years of Indo-European comparative linguistics at the University of Ljubljana, Hamburg, Baar, 2020, pp. 163‑188.
Golston 2014: C. Golston, “Minima”, in G. Giannakis, V. Bubeník, E. Crespo, C. Golston (dir.), Encyclopedia of Ancient Greek Language and Linguistics, vol. II, Leiden, Brill, 2014, pp. 442‑445.
Hackstein 2002: O. Hackstein, Die Sprachform der homerischen Epen, Wiesbaden, Reichert, 2002.
Hamm 1957: E.‑M. Hamm, Grammatik zu Sappho und Alkaios, Berlin, Akademie-Verlag, 1957.
Harðarson 1993: J.A. Harðarson, Studien zum urindogermanischen Wurzelaorist, Innsbruck, Institut für Sprachwissenschaft der Universität Innsbruck, 1993.
Helly, Tziafalias 2013: B. Helly, A. Tziafalias, “Décrets inédits de Larisa organisant la vente de terres publiques attribuées aux cavaliers”, Topoi: Orient-Occident 18, 2013, pp. 135‑249.
Hettrich, Stüber 2018: H. Hettrich, K. Stüber, Infinitivische Konstruktionen im R̥gveda und bei Homer, Mainz/Stuttgart, Akademie der Wissenschaften und der Literatur/Franz Steiner Verlag, 2018.
Keydana 2017: G. Keydana, “‘Finite’ infinitives in Ancient Greek”, Indo-European Linguistics 5, 2017, pp. 49‑76.
Kortlandt 1988: F.H.H. Kortlandt, “The Greek 3rd pl. endings”, Münchener Studien zur Sprachwissenschaft 49, 1988, pp. 63‑69.
von der Lahr 1992: S. von der Lahr, Dichter und Tyrannen im archaischen Griechenland: das Corpus Theognideum als zeitgenössische Quelle politischer Wertvorstellungen archaisch-griechischer Aristokraten, Munich, Tuduv, 1992.
Masson 1983: O. Masson, “Remarques sur quelques passages de la tablette chypriote d’Idalion (ICS 217)”, BSL 77, 1983, pp. 261‑281.
Meillet 1931: A. Meillet, “Les cas employés à l’infinitif en indo-européen”, BSL 32, 1931, pp. 188‑193.
Méndez Dosuna 2017: J.V. Méndez Dosuna, “The language of the Dodona oracular tablets. The non‑Doric consultations”, in G. Giannakis, E. Crespo, P. Filos (dir.), Studies in Ancient Greek dialects. From Central Greece to the Black Sea, Berlin/Boston, De Gruyter, 2017, pp. 265‑296.
Miller 2014: D.G. Miller, Ancient Greek dialects and early authors, Berlin, De Gruyter, 2014.
Minamimoto 2015: T. Minamimoto, “The infinitive διδουμεν ‘to give’ and the ‘Aeolic’ inflection in Thessalian Greek”, in S. Jamison, H.C. Melchert, B. Vine (dir.), Proceedings of the 26th annual UCLA Indo-European Conference, Bremen, Hempen, 2015, pp. 135‑140.
Parker 2008: H. Parker, “The linguistic case for the Aiolian migration reconsidered”, Hesperia 77, 2008, pp. 431‑464.
Peters 1986: M. Peters, “Zur Frage einer achaiischen Phase des frühgriechischen Epos”, in A. Etter (dir.), o‑o‑pe‑ro‑si. Festschrift für Ernst Risch, Berlin, De Gruyter, 1986, pp. 303‑319.
Peters 1989: M. Peters, Sprachliche Studien zum Frühgriechischen, Habilitationsschrift, University of Vienna, 1989 (unpublished).
Plath 1990: R. Plath, “Mykenisch e‑re‑e”, Münchener Studien zur Sprachwissenschaft 51, 1990, pp. 169‑182.
Porzig 1954: W. Porzig, “Sprachgeographische Untersuchungen zu den altgriechischen Dialekten”, IF 61, 1954, pp. 147‑169.
Richardson 1993: N. Richardson, The Iliad: a commentary, vol. VI, books 21‑24, general ed. G.S. Kirk, Cambridge, Cambridge University Press, 1993.
Rix 1992: H. Rix, Historische Grammatik des Griechischen, Darmstadt, Wissenschaftliche Buchgesellschaft, 1992 (2nd ed.).
Ruijgh 1992: C.J. Ruijgh, “L’emploi mycénien de ‑h‑ intervocalique comme consonne de liaison entre deux morphèmes”, Mnemosyne 45, 1992, pp. 433‑472.
Scarborough 2023: M. Scarborough, The Aeolic dialects of Ancient Greek: a study in historical dialectology and linguistic classification, Leiden, Brill, 2023.
Schwyzer 1939: E. Schwyzer, Griechische Grammatik, vol. I, Allgemeiner Teil, Lautlehre, Wortbildung, Flexion, Munich, Beck, 1939.
Stüber 2000: K. Stüber, “Zur Herkunft der altindischen Infinitive auf ‑sáni”, Münchener Studien zur Sprachwissenschaft 60, 2000, pp. 135‑167.
Taillardat 1960: J. Taillardat, “Notules mycéniennes”, REG 73, 1960, pp. 1‑14.
Vanséveren 2000: S. Vanséveren, “Prodige à voir”: recherches comparatives sur l’origine casuelle de l’infinitif en grec ancien, Louvain‑la‑Neuve, Peeters, 2000.
Wackernagel 1914: J. Wackernagel, “Akzentstudien III”, Göttingische Gelehrte Nachrichten, 1914, pp. 97‑130 (reprinted in Kleine Schriften, vol. I‑II, Göttingen, Vandenhoeck & Ruprecht, 1969, pp. 1154‑1187).
Wathelet 1970: P. Wathelet, Les traits éoliens dans la langue de l’épopée grecque, Rome, Edizioni dell’Ateneo, 1970.
Willi 2012: A. Willi, “Kiparsky’s rule, thematic nasal presents, and athematic verba vocalia in Greek”, in P. Probert, A. Willi (dir.), Laws and rules in Indo-European, Oxford, Oxford University Press, 2012, pp. 260‑276.
Notes de bas de page
1 Many thanks to the editors Alcorac Alonso Déniz and Julián V. Méndez Dosuna for their extensive and frank comments on the first draft. I would like to express my gratitude also towards Prof. Albio C. Cassio, who discussed several details with me via email. Any errors of fact or judgment that remain are my own responsibility. The following abbreviations are used in this paper: aor. = aorist, Arc. = Arcadian, Att. = Attic, class. = classical, Cypr. = Cypriot, imper. = imperative, indic. = indicative, inf. = infinitive, Ion. = Ionic, Lesb. = Lesbian, mid. = middle, Myc. = Mycenaean, pass. = passive, PGr. = Proto-Greek, PIE = Proto-Indo-European, pl. = plural, ptc. = participle, sg. = singular, Ved. = Vedic, WIon. = Western Ionic.
2 For the Lesbian attestations, I refer to the standard handbooks, in particular Hamm 1957; Blümel 1982; and Hodot, Dialecte éolien Asie. Forms that are only attested epigraphically will not receive diacritic marks.
3 This view goes back at least to Porzig 1954, p. 153, and is widely accepted, cf. Wathelet 1970, pp. 315‑324; Rix 1992, p. 238; Hettrich, Stüber 2018, p. 53.
4 Ruijgh 1992, p. 450.
5 Wackernagel 1914, p. 103; García Ramón 1990, p. 160; Schwyzer 1939, p. 807; Hamm 1957, pp. 171‑172; Chantraine 1961, p. 276; Hodot, Dialecte éolien Asie, pp. 158‑159.
6 See recently Scarborough 2023, pp. 177‑179.
7 Cf. Rix 1992, pp. 237‑238; García Ramón 1997; García Ramón 2009; García Ramón 2020; Hettrich, Stüber 2018.
8 Keydana 2017, p. 55: “The development of the infinitive had already reached a very advanced state in early Greek, since modal semantics are no longer obligatory even in the oldest attestations”, with examples from Mycenaean.
9 García Ramón 2020; Hettrich, Stüber 2018.
10 Meillet 1931, pp. 192‑193 called ‑αι “un élément accessoire, ajouté à la forme, mais non essentiel”; see also Taillardat 1960.
11 It is difficult to decide whether the extension ‑ai in *‑menai and *‑henai has any relation with the final diphthong of *‑(t)sai and *‑sthai.
12 Myc. /teleiāhen/ is the infinitive of an athematic denominative present stem /teleiā‑/. Cf. possibly also e‑re‑e /erehen/ ‘to row’, provided that this verb was athematic in Mycenaean (Plath 1990), which in my view is questionable.
13 E.g. Schwyzer 1939, p. 807; Dubois, Dialecte arcadien, p. 111; García Ramón 1990, p. 162; Rix 1992, p. 238; Ruijgh 1992.
14 I follow García Ramón 1977 in explaining the West Greek short-vowel thematic infinitive in ‑εν as secondary for ‑ειν, ‑ην by generalization of the pre‑consonantal sandhi form.
15 Two of these three instances are Τρωσὶν δὴ βόλεται δοῦναι κράτος (Il., XIII, 319) and δοῦναι δὲ μέγα κράτος Ἀργείοισιν (Il., XV, 216); contrast αἴ κ’ ἐθέλῃσι θεὸς δόμεναι κράτος (Il., XIII, 743). Apparently the older (Aeolic) form with ‑μεν(αι) was not replaced everywhere (cf. ἔμμεν beside εἶναι in Homer).
16 Peters 1989, p. 103, claims that verse-final θεῖναι and δοῦναι in Homer may reflect metrically lengthened θέμεν and δόμεν, but this is clearly an emergency measure.
17 Two further Att.‑Ion. forms that are alleged to support the reconstruction *‑wenai are εἶναι (which would reflect *eh‑wenai according to Peters 1989) and the perfect infinitive εἰδέναι ‘know’ (Peters 1989, pp. 106‑107). Both arguments will be discussed below (see infra §§ 3.1, 3.2, 3.5).
18 Cf. García Ramón 1997, pp. 64‑67; Stüber 2000 on PIE *‑sén(‑i).
19 E.g. Wackernagel 1914, p. 102, connecting Lesb. ‑μεναι with dative Ved. ‑mane. This view can now be maintained only if one assumes that original *‑ei was secondarily altered to ‑αι in Greek, for instance under influence of the sigmatic aorist infinitive ending ‑σαι.
20 Rix 1992, p. 238; Hettrich, Stüber 2018, p. 53.
21 E.g. García Ramón 2020: ‑μεναι is either a conflation of dative *‑men‑ei and allative *‑men‑eh2, or an allative *‑men‑eh2 recharacterized with ‑i.
22 See also Benveniste 1936, pp. 87‑99 and 129‑134, who viewed PIE *‑men and *‑sen as reflexes of an old cas indéfini. This was referenced favorably by J.L. García Ramón in an early publication (see García Ramón 1977, p. 180); cf. also Dubois, Dialecte arcadien, p. 177 with endnotes. Vanséveren 2000, pp. 71‑73 discusses the problems with a comparison between ‑αι and Vedic infinitives of the type gántavái, but also with the analysis of ‑αι as a case ending.
23 By pre‑Lesbian I mean an unattested pre‑stage of Lesbian that is still on the same node of the Greek dialect tree (i.e. not so early that one would have to speak about, e.g., Proto-Aeolic or Proto-Greek).
24 Taillardat 1960.
25 The Cyrenaean infinitives τιθΕν, διδων (attested in the 1st c. BCE) are analogical creations replacing older forms in ‑μεν in this dialect: cf. Dobias, Dialecte Cyrène, pp. 131‑132.
26 In ψαέναι· φθάσαι the quantity of the root vowel is not indicated; given that we are dealing with a root aorist corresponding to ἔφθην, it was probably long (but cf. the vocalism of mid. ptc. φθάμενος). The development to φθ‑ > ψ‑ is attested for Cretan, e.g. ψίνω for φθίνω, but the infinitive ending ‑έναι is not attested there. According to Peters 1989, p. 102, ψαέναι was taken from pre‑Homeric Boeotian epic.
27 On geminate reflexes of *VhR, *Vhw in Arcadian, see Peters 1986, p. 306, n. 8.
28 Cowgill 1964, p. 356, notes that the verb form δαμιοϝοργε (with preserved intervocalic digamma) occurs in an early Arcadian inscription (SEG XI, no. 1112) which also has εναι for εἶναι. However, this is not a valid argument against reconstructing εἶναι as *ehwenai: *w may have been restored intervocalically in the compound δαμιο‑ϝοργο‑, as the sound was retained longer in word-initial position (cf. ϝεργο in the same inscription). Against the reconstruction *ehwenai, Cowgill 1964, p. 356, also argues that Homer does not have any trace of *εἰέναι, but Peters 1989 rightly remarks that ἔμμεναι has the same metrical shape.
29 To assume that ἔμεν and ἔμεναι were formed analogically to Aeolic ἴμεν, ἴμεναι ‘to go’, whether on the basis of the participle forms ἐών, ἰών (Ruijgh 1992, p. 450, n. 52) or of 3rd pl. ἔᾱσι, ἴᾱσι (Peters 1986, p. 316, n. 46), seems more plausible on paper than as a real linguistic development: one wonders why ‘go’ would have influenced the much more frequent verb ‘be’.
30 Considering only the phonological form of ἱστάναι, διδόναι, τιθέναι, it is theoretically possible to view them as reflexes of inherited PIE forms *si‑sth2‑én‑, etc. (Rix 1992, p. 238, followed by Hettrich, Stüber 2018, p. 53); ‑έναι would be the locative of an abstract noun in *‑éneh2. However, such a reconstruction would be purely mechanical, and there is no guarantee that abstracts in *‑éneh2 are an old type: the Vedic action noun vanánā ‘desire’ (usually cited in this connection) is isolated, and the same holds for the Old Irish verbal noun orcun ‘destruction’ < *h3erg‑eneh2 to orcaid ‘destroys’ (cited by Hettrich, Stüber 2018).
31 Normally the diphthong ‑αι would be shortened (epic correption) when followed by a vowel, hence ἑστάναι and ἑστάμεν could be used in the same metrical environments.
32 In Theognis, we find μεταδοῦν (104, prevocalic) and συνιεῖν (565) to συνίημι ‘understand’. However, it remains unclear what relation these forms could have to Western Ionic dialects. Plato claimed that Theognis was not from Megara Nisaea in mainland Greece, but from Megara Hyblaea in Magna Graecia, in which case Theognis could have acquired forms like μεταδοῦν from neighboring Western Ionic colonies. However, the current consensus is that Theognis was from Megara Nisaea (see von der Lahr 1992, p. 9, n. 1).
33 Parmenides is from the Eastern Ionic colony of Elea; Cassio 1996, p. 18, argues that Western Ionic linguistic features may have spread in Magna Graecia from Western Ionic to Eastern Ionic colonies.
34 It is assumed that Stesichorus, who was from Magna Graecia, could use forms like εἶν because these were current in neighboring Western Ionic colonies in Southern Italy. See Cassio 1996, p. 17‑18, who speaks of εἶν as “una forma euboica che il grande lirico occidentale ha imparato a conoscere a Imera o in altre colonie euboiche di Magna Grecia e Sicilia”.
35 That τιθεῖν (and ἱεῖν) are relatively old is perhaps suggested by the “thematic” infinitives πιέναι, πελέναι, κλεπτέναι as attested for usual πιεῖν, πέλειν, κλέπτειν in inscriptions and literary sources from Magna Graecia. Cassio 1996 argues that these forms in ‑έναι are explained most easily in a linguistic context where τιθεῖν coexisted with τιθέναι, i.e. in colonies with a Western Ionic dialect (or that were in contact with such dialects).
36 Hackstein 2002, pp. 95‑120, discusses the spread of thematic forms in originally athematic presents, including reduplicated stems, and argues that WIon. τιθεῖν, διδοῦν and ἱστᾶν are thematicizations. However, Willi 2012 analyzes the forms τιθεῖ and διδοῖ as the regular reflexes of athematic forms in *‑ē‑ti, *‑ō‑ti, and notes that this reconstruction immediately accounts for the semi-thematic paradigm of verba vocalia in Aeolic. It is perhaps more prudent to say that the conjugation of (ablauting) reduplicated athematic presents was influenced by that of (non‑ablauting) verba vocalia. Plath 1990 views WIon. τιθεῖν, διδοῦν and ἱστᾶν as old athematic forms in *‑hen, comparing the Vedic infinitive ending ‑sáni that is used after roots originally ending in a laryngeal. However, this would make it more difficult to account for the preservation of ‑μεν after polysyllabic ablauting stems in Homer (e.g. ζευγνύμεν, ὀρνύμεν).
37 With previous scholars, Hackstein 2002, p. 105‑106, analyzes WIon. εἶν, ἐξεῖν as thematic forms. This is implausible, as there is no secure evidence for thematicization in εἰμί ‘be’ in Attic-Ionic dialects (the rare epic forms opt. ἔοι and impf. ἔον can be seen as Aeolisms or artificial creations).
38 Dubois, Dialecte arcadien, pp. 173‑174, accepts the antiquity of variation between *‑ehen and ‑en in thematic verbs and thinks that Arc. εξιεν is a “thematic” infinitive because ‑εν occurs as a thematic infinitive ending also at Tegea. I agree with García Ramón 1977 that this interpretation of εξιεν is highly implausible.
39 Cf. also aor. γραφηναι (Dial. graec. ex., no. 657, l. 2, 4th c. BCE).
40 Cf. also Dubois, Dialecte arcadien, p. 176.
41 The evidence for ‑μεναι in mainland dialects is very weak. The best evidence by comparison is the form εἴμεναι, put into the mouth of a Megarian by Aristophanes (Acharnians, 775), and judged authentic by Wackernagel 1914, p. 102.
42 For the epigraphic evidence, see Blümel 1982, p. 209; Hodot, Dialecte éolien Asie, pp. 157‑160.
43 Blümel 1982, p. 209.
44 Wackernagel 1914, p. 103.
45 The interpretation of the preverbs εἰσ‑αν‑ and the restoration of the preceding word are debated, but ‑βάμεν’ as an infinitive of ἔβᾱν is secure. For a critical discussion from a philological and paleographical perspective, see Bettarini 2007 with references.
46 See Ellsworth 2011; Willi 2012 on sandhi developments involving word-final PGr. ‑t(h)i.
47 E.g. Schwyzer 1939, p. 742; Rix 1992, p. 214.
48 Ablaut in the root aorist paradigm is also suggested by relic middle forms such as the isolated φθάμενος (8xHom.) beside non‑ablauting active φθη‑, and by κτᾰ‑ < zero-grade *ktn̥‑ (e.g. in act. inf. κατα‑κτάμεν, κτάμεναι, mid. ptc. κτάμενος). Furthermore, note that ablauting roots of the structure *CReH‑ automatically merged with the non‑ablauting (θ)η‑aorist paradigm, e.g. ἔγνων (*ǵneh3‑m) alongside ἔγνωμεν (*ǵn̥h3‑me), and that ablauting roots of the structure *CeRH‑ could be easily transferred into the non‑ablauting paradigm by generalizing the zero grade allomorph, e.g. ἔτλην (*tl̥h2‑) and ἔβλην (*gwl̥h1‑). In fact, these last‑mentioned forms presuppose an original paradigm with root ablaut.
49 Note that the perfect preserves productive root ablaut (e.g. τέθνηκα, 1st pl. τέθναμεν, inf. τεθνάμεν after βέβηκα, βέβαμεν, inf. βεβάμεν) because there was no competing non‑ablauting perfect stem. Certain perfect participles resisted this productive ablaut scheme longer (cf. Hom. τεθνηώς, τετληώς).
50 For further discussion (including the alternative reconstruction by Peters 1986), see infra § 3.5.
51 See Blümel 1982, pp. 210‑215 for the evidence. Τhe perfect infinitive has become thematic in Lesbian (‑ην) and in Thessalian (‑εμεν), so it will not be taken into account here.
52 A fourth explanation was proposed by Ruijgh 1992, p. 450, who assumed that (post-migration) Eastern Ionic preserved reflexes of the older South Greek ending *‑hen, in forms like inf. aor. pass. *‑thē‑hen > *‑θῆν (attested as ‑θῆναι from *‑thē‑henai). He suggested that *‑hen was borrowed into Lesbian, yielding ‑ν after contraction. C.J. Ruijgh also claimed that Lesbian ‑μεναι arose from Proto-Aeolic ‑μεν under the influence of (alleged) Eastern Ionic pairs like ἰέν beside ἰέναι. These assumptions are difficult to substantiate. Peters 1989, p. 97‑98 claimed that pre‑migration Ionic already generalized ‑ναι in view of the Argive forms δαηναι and δασθεναι, which in his view are due to an Ionic substrate. This claim, too, rests on shaky foundations.
53 Hodot, Dialecte éolien Asie, pp. 157‑161, even assumes that the original Lesbian contract presents were completely thematic, and accounts for athematic forms in the paradigm of κάλημμι by contractions and subsequent analogies. I find this scenario implausible because it does not account for the 1st sg. ending ‑μι.
54 This reconstruction of Lesbian ‑ν was also mentioned as a possibility by Schwyzer 1939, p. 807. Peters 1989, p. 98, also seems to accept it, but is implicit on this point.
55 Hamm 1957, p. 171, n. 69: “Die Endung ‑ην der Verben auf ‑ημι wurde, als ‑η‑ν aufgefasst, auf die θην‑Aoriste […] übertragen”.
56 Pace Wackernagel 1914, p. 103, it is irrelevant that προσταν is a compound, as all infinitives of monosyllabic short-vowel roots occur epigraphically as compounds, too (e.g. δομεναι beside ἀποδομεναι in Kyme).
57 See generally Golston 2014. In lexical words, monosyllables were acceptable as long as they had a bimoraic vowel (γνῶ) or a complex coda (θρίξ). Cf. also πούς ‘foot’ (phonologically lengthened from *πός, contrast τρίπος ‘tripod’), neuter πᾶν ‘every’ (contrast ἅπᾰν, σύμπᾰν) or the letter names εἶ, οὖ.
58 Inf. ἱέμεναι ‘make one’s way’ (Il., XXII, 206) may be etymologically distinct from ἵημι.
59 Cf. indic. κίχεις (Od., XXIV, 284) and mid. ptc. κιχήμενος.
60 The variation in the Odyssey between ἄη and ἄει as the 3rd sg. imperfect form (Hackstein 2002, p. 119) suggests that ἄημι was also attracted by the athematic contract verb inflection, cf. (ἐ)τίθει to τίθημι.
61 To φημί ‘say; think’, only middle φάσθαι is used in Homer; neither φάναι nor φάμεν(αι) is attested.
62 Cf. Schwyzer 1939, p. 808, n. 4; Chantraine 1958, p. 104; García Ramón 1990, p. 159; cf. also the artificial future infinitive δῶρα διδώσειν (Od., XXIV, 314). The suggestion of Richardson 1993, ad loc. that διδοῦναι is a metrically lengthened form of διδόναι, as in the unique ζευγνῦμεν (Il., XVI, 145), is implausible.
63 Besides, τιθέμεν is used by Hes. (Op., 744), while Western Ionic τιθεῖν is found at Thgn., 286. In Homer, the long vowel occurs also in the mid. ptc. τιθήμενος (Il., X, 34).
64 For this transfer, see Minamimoto 2015 on the Thessalian infinitive ‑διδουμεν, and cf. also the form καττιθειμεν that has recently come to light in Larisa (Helly, Tziafalias 2013). I thank Prof. Albio C. Cassio for kindly drawing my attention to these papers.
65 In early Epic Greek, the form τεθνάναι is attested first in a fragment of Hesiod (frg. 266a, 11, cf. 266c, 4 M‑W) and in fragments of Mimnermus (frg. 2, 10) and Semonides (frg. 3, 1).
66 See Scarborough 2023, p. 119.
67 That Att.‑Ion. ‑ναι after short vowels was secondary for ‑μεν was suggested in passing already by Schwyzer 1939, p. 808 (“die alle aus Formen mit ‑μεν umgebildet sein können”), but it has, to my knowledge, never been taken seriously.
68 See Chantraine 1958, p. 485. The use of ‑μεναι, the only disyllabic ending, is not subject to any restrictions.
69 Note that forms like *ἑστᾶν, *ἑστᾶναι would have been a metrically useful addition to the inventory of perfect infinitives in Epic Greek.
70 See Peters 1986, pp. 307‑308.
71 I assume here that *‑henai (which originally occurred exclusively after vowel-final stems, as *h can only reflect lenited *s in forms that were inherited as such from Proto-Greek) could also be attached to consonant-final stems when *h was no longer an allophone of *s. This situation had been reached already in the Mycenaean period (when /s/ may already appear between vowels).
72 As Alcorac Alonso Déniz suggests to me, in view of 2nd sg. PIE *h1esi (pre‑PIE *h1és‑si) as reflected in Att. εἶ ‘you are’, one could assume that *esen (< *h1es‑sen) existed in Proto-Greek already before the lenition *s > h. However, given that geminate ‑ss‑ was clearly tolerated at morpheme boundaries in such forms as Proto-Greek fut. *essomai (> Att. ἔσομαι), this would presuppose a very high antiquity for the athematic infinitive in *‑sen, an assumption for which I see no sufficient reason.
73 For instance, κάτθεμεν ἐν λεχέεσσι (Od., XXIV, 44) means ‘we put you to bed’, while using the infinitive κατθέμεν would yield the meaning ‘you must put to bed’. In this compounded verb, the accentuation helps to distinguish both forms.
74 Ruijgh 1992, p. 450, n. 51, surmised that Aeolic (Thessalian, Boeotian) originally also generalized the ending ‑μες, just like West Greek, while Lesbian would owe ‑μεν to contact with Eastern Ionic. If so, this would explain why continental Aeolic had no problems with generalizing the infinitive ending ‑μεν. However, we must note that there is evidence for ‑μεν as a primary 1st pl. ending in Thessalian: subj. επινοεισουμεν (IG IX, 2, no. 517, l. 13, 214 BCE), cf. Scarborough 2023, pp. 121‑122 with discussion. Méndez Dosuna 2017, pp. 282‑283, also draws attention to subj. ὀ‹γ›κομίξο̄μεν in an oracular tablet from Dodona written in Thessalian (I.Dodone Evangelidi, no. 336, A, l. 1‑2, 350‑300 BCE?). The form ἐκρίνναμεν (SEG XLIII, no. 311, B, l. 57, Skotoussa, 197‑185 BCE) furnishes evidence for the retention of ‑μεν as a secondary ending.
75 The same motivation may underlie the Attic-Ionic generalization of extended form ‑ναι, ‑έναι at the expense of ‑ν in polysyllabic stems.
76 I hope to provide more elaborate argumentation for this claim in a future publication.
77 The accentuation of ἰέναι might be secondary for older *ἴεναι. I assume that the pre‑desinential accentuation of ‑μεν infinitives (e.g. διδόμεν) is a relatively archaic trait that was also originally a feature of infinitives in *‑hen. When the analogical spread of ‑ναι occurred, pre‑desinential accentuation was retained (e.g. διδόναι for older διδόμεν), aided by aorist infinitives in ‑ῆναι that had undergone contraction, as well as by the s‑aorist infinitive in ‑σαι.
78 Note that the sigmatic aorist active infinitive in ‑σαι also had penultimate accentuation.
79 Pace Parker 2008, cf. García Ramón 2010; Scarborough 2023.
80 For *r̥ > ρο as a specifically Aeolic development, see van Beek 2022.
81 In Thessalian, Boeotian and West Greek, the assibilation is found mainly in action nouns in ‑σις and in some forms in ‑σιος, ‑σια, but not in verbal endings and numerals.
82 Porzig 1954; cf. García Ramón 1975.
83 Linguistically, the migration must postdate the geminate outcome of the first compensatory lenthening (probably shared by Thessalian and pre‑Lesbian) and predate the completion of the diphthongization of Vns in Lesbian.
84 See Cowgill apud Bartoněk 1968, p. 182; Miller 2014, pp. 321‑322; Finkelberg 2017.
85 In this respect, I find myself in partial agreement with Finkelberg 2017.
Auteur
-
Lucien van Beek
Universiteit Leiden
IdRef : 139501010
Le texte seul est utilisable sous licence Creative Commons - Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International - CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
Dire la ville en grec aux époques antique et byzantine
Actes du colloque de Créteil, 10-11 juin 2016
Liliane Lopez-Rabatel, Virginie Mathé et Jean-Charles Moretti (dir.)
2020
Dérivation nominale et innovations dans les langues indo‑européennes anciennes
Alain Blanc et Isabelle Boehm (dir.)
2021
Stylistique et poétique de l'épigramme latine
Nouvelles études
Daniel Vallat et Florence Garambois-Vasquez (dir.)
2022
Contacts linguistiques en Grèce ancienne
Diachronie et synchronie
Alcorac Alonso Déniz, Julián V. Méndez Dosuna, Enrique Nieto Izquierdo et al. (dir.)
2024
Les représentations du désir féminin dans la poésie latine, de Catulle à Juvénal
Contraintes et libertés de l’espace poétique
Giacomo Dimaggio
2024
Définir l'épopée en Grèce ancienne
Contribution à une histoire de l'idée d'épopée d'Hérodote à Tzetzès
Flore Kimmel-Clauzet
2024
Lector in poematis
Le lecteur dans la poésie gréco-latine
Bruno Bureau, Bénédicte Delignon, Jean-Christophe Jolivet et al. (dir.)
2025
