ἤμελλον, ἠβουλόμην, ἠδυνάμην
(Moer. η 5, Philemo [Vindob.] 394.10)

D. General commentary

Two entries in the lexicon of Moeris (A.1, B.2) and one in the lexicon of Philemon (A.2) deal with the forms of the verbs βούλομαι ‘I want, wish, desire’, δύναμαι ‘I can, am able’, and μέλλω ‘I am about to’, showing the augment ἠ- instead of ἐ- in the imperfect and aorist indicative. The two lexicographers give opposite prescriptions: while Moeris recommends the long-augment forms as Attic, Philemon prescribes the short augment, at least for ἐβουλόμην. In fact, Moeris’ entries bring together – perhaps as a result of epitomisation – several forms whose augment in ἠ- has different historical explanations.

As has long been recognised (see Debrunner 1954, with a survey of previous scholarship), the possibility of a ‘long’ syllabic augment in ἠ- for this group of verbs is analogical on (ἐ)θέλω ‘I want, am willing’, where the initial ἐ- was originally part of the root (reflecting an initial laryngeal in Indo-European *h₁gʷʰel-, cf. Old Church Slavonic želěti ‘to wish, want’; but see Willi 2018, 384 n. 80 for the difficulties of this reconstruction and a different, tentative explanation of the ἐ-). This root regularly produced an augmented stem ἠθελ- with the so-called ‘temporal augment’. In the Homeric poems the present ἐθέλω is the norm, with the sole exception of θέλοιεν in Od. 15.317, so that ἐθελ- in the past tenses should be regarded as unaugmented. The allomorph θέλω is a post-Homeric development, probably produced by apheresis in phrases such as εἰ (μὴ) ’θέλεις, and is common in Aeolic lyric poetry and Ionic prose, while older Attic retains ἐθέλω (see Debrunner 1954, 104). In later Greek, θελ- spread in the unaugmented forms, while at the same time ἠ- was retained as the augment, producing a paradigm θέλω – ἤθελον – θελῆσαι – ἠθέλησα. Incidentally, the creation of the perfect τεθέληκα, condemned as an ‘Alexandrian’ feature by Phrynichus (B.1), in addition to ἠθέληκα, suggests that θελ- was perceived by some speakers of Post-classical Greek as the underlying verbal stem (this is further proved by Clarysse’s [2008] study of papyri and inscriptions from the Hellenistic, Roman, and Byzantine periods, which show that ἐθέλω is only used in more formal texts). The augment ἠ- then spread to the almost synonymous βούλομαι and the semantically close δύναμαι and μέλλω (the latter also phonetically close to θέλω); according to Debrunner (1954, 108), the coexistence of ἤθελον and ἔθελον in Homer first allowed the emergence of ἤμελλον alongside ἔμελλον in poetry; then the emergence of a paradigm θέλω : ἤθελον led to the reanalysis of ἠ- as a ‘long augment’, allowing for the creation of μέλλω : ἤμελλον, δύναμαι : ἠδυνάμην, and βούλομαι : ἠβουλόμην. This case therefore differs from the more complex one of several verbs that show a ‘long augment’ ἠ- from the beginning, especially before *w-, which may be inherited (see Ittzés 2005, 208–14, who is sceptical about a common origin with Vedic ā-, and now Chiattelli 2022), although both categories contributed to the synchronic perception that some verbs could have a syllabic augment ἠ-, rather than ἐ-, in the indicative of the past tenses (indeed, this allomorph of the augment spread further in later Greek: see E.).

The three analogical forms that are the focus of Moeris’ entry are attested at different times and with different frequencies (see Debrunner 1954, 101–4). ἠμελλ- is first attested in Hesiod (Th. 478 = C.1, 898) and Theognis (1.259, 1.906); Aristophanes uses it twice in anapaestic passages (C.4, C.5). ἠδυν- first occurs in the Prometheus Bound attributed to Aeschylus (C.2), never in Aristophanes (who has ἐδυν- 4x). ἠβουλ-, the youngest of these forms, is never metrically guaranteed in classical Attic: Eur. Hel. 752 (C.3) ἠβούλετο can also be read as ἐβούλετο, since the metre allows it, and [Eur.] fr. 1132.28 (ἠβούλετο) is a forgery from the 5th/6th century CE (see Karamanou 2006, 225–7), while the elided vowel in Ar. Ra. 1147 ’βουλόμην is probably ἐ-, since Aristophanes uses only ἐβουλ- elsewhere (6x), and the apheresis of ἠ- is quite uncommon. The predominance of ἐ- in older Attic is confirmed by inscriptions, where ἠ- first appears around 350 BCE and becomes the norm by 330 BCE (see Threatte 1996, 474); during the 4th century BCE the analogical forms with ἠ- also become more common in Attic prose. In later comedy the augment ἐ- is still the more usual one for these verbs, although there are a few attestations of ἠ- (see Arnott 2002, 196–7). ἠβουλ- is used by Alexis (C.7, see Arnott 1996, 733), Macho (C.10; see Gow 1964, 89) and in an adespoton fragment, possibly by Menander (C.9, see Arnott 2000, 563–5), although none of these occurrences is metrically guaranteed. ἠδυν- is guaranteed in Philippides (C.8), where the rare form ἠδύνω (only here and in Antisthenes, C.6), which attracted the interest of Moeris (B.2), is considered by Hartwig (2022, 33–4, 183) to be a colloquialColloquial language form like nom. plur. γυναί ‘women’ in Philippides’ fr. 2 (see entry γυναί). The usage in the Hellenistic and imperial periods shows several trend reversals. The ἠ- augment is the norm for βούλομαι, δύναμαι, and especially μέλλω in Ptolemaic papyriPapyri (see Mayser, Gramm. vol. 1,2, 93–4), but Roman and Byzantine papyri return to ἐ-, albeit to varying degrees: the long augment is particularly rare for βούλομαι, which was the last of the three verbs to receive it (see Gignac 1981, 228–31; a search on the website papyri.info supports his evaluation, yielding 9 occurrences for ἠβουλ- after 30 BCE, compared to 54 for ἠδυν- and 21 for ἠμελλ-). Koine authors also generally prefer ἐ-, although ἠδυν- is frequent, and ἠ- becomes common again in later Greek (see E.). In the New Testament δύναμαι and μέλλω oscillate between ἠ- and ἐ-, with many oscillations in the MSS, while ἠβουλ- is very rare (see Blass, Debrunner 1976, 53).

The rationale behind the Atticist entries devoted to this problem is discussed by Vessella (2018, 193–5). Moeris, in particular, is surprising in his preference for the long-augment forms, which are post-classical: Vessella (2018, 194–5) suggests that this is either ‘a consequence of the process of epitomisationEpitome’ or that ‘the entry generalises vowel lengthVowel length, attributing the long vowels to Attic’, perhaps to mark the difference with the koine, which preferred ἐ- (see AGP vol. 1, 315). Indeed, both of Moeris’ entries, as transmitted, include forms with an augment in ἠ- of a different origin. ηὐξάμην (A.1), with a regular temporal augment from αὐξ-, is not a distinctively Attic form, since Attic had a tendency to shorten augmented forms in ηὐ- to εὐ- (Lautensach 1899, 47–9, 146–9 argued that the rule in 5th-century BCE Attic was ηὐ- except for compounds with εὐ- ‘well’; this was disputed by Mastronarde 1989; see also the very useful discussion by Threatte 1980, 384–5 and Threatte 1996, 482–3). The peculiarity of the augment in ἠπίστω (B.2), a genuinely Attic form, is not its length but the fact that ἠ- appears on what was originally a prefix; at any rate, this entry deserves separate treatment, since it is also shows the ending -ω for non-Attic -ασο (see entry ἠδύνω, ἠπίστω). Philemon, on the other hand, may be reacting to the spread of ἠ- in Post-classical Greek; his proscription would be even more understandable if it were really limited to ἠβουλόμην, since this form, unlike ἠμελλ- and ἠδυν-, has no precedent in classical Attic literature before the 4th century BCE. This would be, as Vessella (2018, 193) observes, one of the cases where ‘Philemo is […] more accurate than Moeris’ in ‘reflect[ing] the state of affairs in Classical Attic’ (for a comparable example, see entry ἄθρους, ἀθρόους). Thomas Magister devotes two entries to this problem, giving apparently contradictory prescriptions. In B.6 he agrees with Philemon (on whom he often relies, as first noticed by Ritschl 1832, LXXV–VI) in advising against ἠβουλ-, observing the much higher frequency of short-augment forms in canonical authors, although he expressly extends the observation to ἠδυν-; in B.7, on the other hand, he prescribes ἠμελλ- over ἐμελλ-. While this apparent inconsistency may be due to Thomas’ reliance on different Atticist sources, it should be noted that the Attic credentials of ἠμελλ- are somewhat more substantial than those of the other two verbs, since a respected author like Aristophanes used it at least twice (C.4, C.5). This prescription, however, is at odds with the actual practice of Atticising authors. Indeed, the more rigorous followers of Atticism restrict the long augment to δύναμαι and βούλομαι: ἠμελλ- is completely avoided by Lucian and Philostratus, almost completely by Aristides and Aelian (perhaps in connection with its partial acceptance by the lexicographers?); on the other hand, ἠδυν- and ἠβουλ- are the norm in the above-mentioned authors (see Schmid, Atticismus vol. 4, 590 for the individual attestations). It is interesting in this context that Phrynichus twice uses ἔμελλον in entries (Ecl. 313Phryn. Ecl. 313, Ecl. 347Phryn. Ecl. 347) dealing with the syntactic construction ‘μέλλω plus infinitive’.

Later erudite sources were divided on the dialectalDialects affiliation of the long-augment forms: Choeroboscus (B.3) recognises them as examples of ‘Attic lengthening’ (possibly in connection with the widespread tendency to ascribe any long-vowel variant to Attic: see Vessella 2018, 93–4), while in the Ἐπιμερισμοὶ κατὰ στοιχεῖον γραφικά (B.4), probably dating from the 9th/10th century CE (see Genakou-Borovilou 2008), they are classified as Ionic. Finally, according to the treatise De barbarismo et soloecismo [5] (B.5), such forms are an example of barbarismBarbarism κατ’ ἐναλλαγήν, i.e., by changing one letter (see Sandri 2020, 131). This divergence of views suggests that the analogical long-augment forms were disputed in ancient scholarship: indeed, while they were not completely foreign to classical Attic, they had an undeniably marginal status in it, but at the same time they were not clearly linked to any other dialectal tradition.

E. Byzantine and Modern Greek commentary

The allomorph ἠ- of the syllabic augment survived into the Byzantine period, and even spread to verbs that did not have it in Classical Greek; the Atticist practice, confirmed by the usage of 2nd-century CE authors (see D.), may have been an additional factor in this development. Since η came to pronounced /i/, it is proper for this period to follow CGMEMG (vol. 3, 1394) in speaking of ‘/i/-augment’, however it may have been spelled. During the medieval and early modern period, the /i/-augment occurs routinely in certain verbs, while it is geographically restricted in some others (see CGMEMG vol. 3, 1398–406). Verbs for which the /i/-augment is common in all regions include θέλω, βούλομαι, and δύναμαι (but not μέλλω), showing that the old analogical forms in ἠ- eventually prevailed in the spoken language; the /i/-augment then spread by various analogical mechanisms to other verbs, for instance those with a stem originally in /e/ or /i/ like στήκω/στέκω (< perf. ἕστηκα) or (μ)πορῶ (< εὐπορέω), and those with a monosyllabic perfective stem, by analogy with εἶπα, εἶδα, ἦλθα, ηὗρα (which had a temporal augment of ancient origin), but also to verbs like φέρ(ν)ω, λέγω, or βλέπω for which there was no particular morphological reason. The stressed /i/-augment is used more extensively in some regions, such as eastern Crete, while in unstressed position [i] may be the phonological realisation of the /e/-augment in dialects with regular raising of unstressed mid-vowels. Modern Greek retains a syllabic augment η- in ήθελα (the imperfect of θέλω), ήξερα (imperfect of ξέρω ‘I know’), and ήπια (aorist of πίνω ‘I drink’); according to the general rule of augmentation in the modern language, the syllabic augment appears only when it bears the stress (see Holton, Mackridge, Philippaki-Warburton 2012, 186).

F. Commentary on individual texts and occurrences

N/A