19. The following Homeric forms are usually regarded as instances of irregular contraction of verbs in -αω, -εω, -οω.
- φορή-μεναι, φορῆ-ναι
to get dry (§ 42)
-οω: σάω 3rd singular imperfect and also 2nd singular imperative of σαόω (I keep safe).
These forms cannot be explained by the ordinary contraction with the thematic ε or ο, e .g. φοιτή-την cannot come from *φοι-ταέτην, φορῆ-ναι from *φορεέ-ναι, ἀλιτή-μενος from *ἀλιτεό-μενος, σάω from σάοε, etc. On the other hand, as Curtius has shown (Stud. iii. 377-401, Verb. i. 352 ff.), they agree exactly with those athematic forms in which the vowel before the ending is long except before -ντ and -ι, such as the present κιχή-μεναι, ἀή-μεναι (§ 12), the aorist στή-μεναι, τλῆ-ναι, γνώ-μεναι, etc., and (as we may add by anticipation) the passive aorists in -ην and -θην.
Moreover, the same type of inflection appears in the peculiar verbs in -μι of the Aeolic dialect, as φίλη-μι, 1st plural φίλη-μεν, 3rd plural φίλεισι (for φίλε-ντι), participle φιλή-μενος; and also in the Latin verbs in -āre and -ēre, except in the 1st singular; e.g. amā-mini is parallel to ἀρή-μεναι, docemini to φορή-μεναι, docemus, doce-nt to φίλη-μεν, φίλεισι.
Further traces of this formation may be seen in those Attic verbs in -αω and -οω which take η and ω instead of ᾱ and ου respectively (as ζάω, ζῇς, ζῇ, etc., ῥιγόω, inf. ῥιγῶν), and in the optative in -ῳην, -οιην (for which however in the case of verbs in -εω we expect -ειην, as in κιχείην and Aeolic φιλείη).
These facts seem to show that the formation now in question is of high antiquity, and Curtius even maintained that it was older than the ordinary conjugation of the verbs in -ᾰω, -εω, -οω. In these verbs, as he pointed out, there is evidence to show that the vowel before the thematic ending was originally long (e.g. in Homeric διψᾱ́ων, πεινᾱ́ων, ὑπνώοντες, Aeolic ποθήω, ἀδικήει, etc.). The forms in -ᾱω, -ηω, -ωω, again, may represent an older (and Aeolic) -ᾱμι, -ημι, -ωμι, just as δεικνύω is for older δείκνῡμι; and these again may be explained by contraction from -ᾱι̯ημι, -ηι̯ημι, -ωι̯ημι, the Greek representatives of the Sanskrit -ayâmi. The Latin amo, doceo, pl. amāmus, docēmus, would fall into this scheme, if we suppose that they belong to the stage at which the thematic endings had not extended beyond the 1st singular.
Against this theory it is urged by Brugmann (M. U. i. 86) that the thematic conjugation of these verbs is found also in Sanskrit, Zend, Slavo-Lithuanian and Germanic—all which members of the Indo-European family, if Curtius is right, must have recast their derivative verbs on the same thematic model. It is more probable therefore that these verbs were originally thematic, and according to the final vowel of the base appeared as verbs in -αω (as νικά-ω), -εω (as ποθέ-ω), or -οω (as δηϊό-ω). On this assumption, again, the Homeric forms now in question may be variously explained. Where we find η for εε or αε, as in φιλήμεναι, γοήμεναι (instead of the ει, ᾱ required by the ordinary rules), we may suppose, with Wackernagel (K. Z. xxvii. 84), that the contraction belongs to an earlier (pre-Hellenic) period. The existence of such a period is proved (e. g.) by the temporal augment, as in ἦ(σ)α for an original ἐ-εσα. Then the participles ἀλιτήμενος, φιλήμενος and the like may be explained by supposing a form in -εμενος, cp. Latin leg-imini, docēmini, so that φιλήμενος would be a primitive contraction from φιλε-έμενος (φιλε-ι̯ε-μενος). The solution however is confessedly incomplete. It does not (directly at least) explain Aeolic φίλημεν, φίλεισι, Latin amāmus, docēmus, amant, docent. It only explains the long vowel of φιλή-σω, ἐφίλη-σα, φιλητός, etc., if we also suppose that the -ι̯ε of the present was carried through all the tenses. And it does not give any satisfactory account of the common contracted forms, νικᾶτε, φιλεῖτε, δηλοῦτε, etc., since these must have come from νικάετε, φιλέετε, δηλόετε, etc., at a period in which the ordinary Greek rules of contraction were in force.
A wholly different explanation is proposed by Brugmann himself (l. c.). He shows, as we have seen (§ 14), that there is a large class of athematic forms with stems ending in a long vowel–ᾱ, η, ω–which is of the nature of a suffix. Such are ἔ-βλ-η-ν (βᾰλ-, βλ-η), ἔ-πτη-ν (πετ-, πτ-η), ἔ-γνω-ν (γεν-, γν-ω-), and many others, which have their representatives in all languages of the Indo-European family. By an extension of this type has been formed the specifically Greek class of the passive aorists in -ην, as ἔ-φανη-ν, ἐ-τύπη-ν and one or two in -ων, as ἑ-άλω-ν.
Similarly, again, the analogy of the verbs in -μι, and especially of those tenses which do not vary the quantity of the stem (as κίχημι, ἄημι, πλῆ-το, ἔγνων) has affected the derivative verbs, and has thus produced the athematic forms in question—φιλήμεναι like ἀήμεναι, ἀλιτήμενος like κιχήμενος, and so on. The forms τιθή-μεναι (Il. 23.83 & 247), τιθή-μενον (Il. 10.34) are probably due to the influence of the same group of verbs. A similar process explains the Aeolic conjugation of verbs in -μι (γέλαιμι, φίλημι, δοκίμωμι), the difference being that in Aeolic it was carried much further. In Homer we have nothing answering to the 1st singular φίλημι, the 1st plural φίλημεν, the 3rd plural φίλεισι, or the corresponding imperfect forms.
We cannot be sure, however, that all the examples of this type which appeared in the original text of Homer have been preserved. Wackernagel has observed that nearly all the words now in question are forms which would be unfamiliar in the Greece of classical times. The list is made up chiefly of duals (προσαυδήτην, φοιτήτην, etc.) and infinitives in -μεναι. It is not improbable (e g.) that the familiar form προσηύδα has supplanted an original athematic προσηύδη. On the other hand in Il. 11.639 ἐπὶ δʼ αἴγειον κνῆ τυρόν the meter points rather to the uncontracted κνάε.