87. The general rule is that the accent is thrown back as far as possible; and the chief departures from this rule are found in the infinitives and participles, which are in reality nouns. In the forms of the verb properly so called the following exceptions have to be noted.
- εἰμί and φημί. The 2nd singular imperfect φα-θί is oxytone.
The disyllabic forms of the present indicative, εἰμί, ἐσσί, φημί, φησί, etc., are enclitic, and, when they do not lose the accent altogether, are oxytone; but ἔστι is accented in the ordinary way when it occurs at the beginning of a sentence, or after certain words (οὐκ, καί, ὡς).
Such was the commonly accepted account; but the ancient grammarians were not agreed as to the enclitic character of the dual and plural forms (on ἐστὸν see Charax 1151; on φαμέν, φατέ, φασί, ibid.; on ἐσμέν, ἐστέ, εἰσί, Eust. 1457, 48). Again, one grammarian denies that φημί was ever enclitic (Charax 1152); another holds that it should be written φῆμι, at least in such instances as φῆμι γὰρ οὖν κατανεῦσαι, κτλ., (Tyrannio ap. Eust. 1613, 18). In all likelihood the original forms were, singular ἔστι, φῆμι, plural ἐσμέν, φαμέν, and we may suppose that φημί and ἐστί are not properly oxytone, but are unaccented forms made oxytone as enclitics (ὠξύνθη διὰ τὴν ἐποῦσαν αὐτοῖς ἔγκλισιν Apoll. Synt.). The Sanskrit verbs of the same kind follow the rule of accenting the stem in the singular, the ending in the dual and plural; and this must be connected with the difference of quantity between strong and weak stems (§ 6). See Benfey, Vedica und Linguistica, pp. 90 ff.
The 2nd singular εἰς is enclitic, though the corresponding Attic form εἶ is not; but see § 5. As to φῄς there is a contradiction; it is not enclitic according to Arc. 142, 8, but enclitic according to Schol. A. Il. 17.147—both notices being supposed to rest on the authority of Herodian (ed. Lentz, i. 553, 4 and ii. 105, 5).
- The 3rd plural ἱστᾶσι, τιθεῖσι, διδοῦσι, δεικνῦσι, are properispomena (Herodian, i. 459, ed. Lentz).
This can hardly have been the original accentuation, since they are not contracted forms, but represent ἵστα-ντι, etc. Probably it comes from the Attic ἱστᾶσι (contracted from ἱστά-ασι, cp. τιθέ-ασι, etc.). The Doric forms are written τιθέντι, etc., by Eustath. Od. 1557, 45; but we do not know that this represents the usage of any living dialect.
- Subjunctives such as φανῇ, δαῶμεν are circumflexed, as being contracted forms (for φανήῃ, δαήομεν). On εἰδέω, εἰδῇς, εἰδῇ, εἰδέωσι see § 80.
Optatives in which -ιη- becomes -ι- before heavy endings are accented on the ι throughout, as διακρινθεῖτε, δαμεῖεν.
But middle forms to which there is no corresponding active follow the general rule: δύνωμαι, δύνηαι (so Herodian, but Tyrannio wrote δυνῶμαι, δυνῆαι, Schol. Il. 6.229), κέρωνται (Il. 4.260), ἐπίστηται (§ 280); ἐπίσταιτο, ὄναιο, ὄνοιτο.
- The imperatives εἰπέ, ἐλθέ, are oxytone (and so in Attic εὑρέ, ἰδέ, λαβέ). Similarly Tyrannio wrote πιθέσθε, λαβέσθε (Schol. V. Il. 18. 266); cp. the Attic βαλοῦ, etc.
The rule in Sanskrit is that the verb loses the accent, except in subordinate clauses, or when it begins the sentence. Hence the verbs εἰμί and φημί in fact retain the original accentuation, which was doubtless that of the Indo-European language. The imperatives εἰπέ, ἐλθέ, etc., are evidently words that would often be used to begin a sentence.
The ordinary accent of a Greek verb, the so-called "recessive" accent, represents the original enclitic condition. The optative φαιήν, for example, is originally oxytone. On the Sanskrit rules it loses its accent, and we should have (e. g.) ἐγώ-φαιην. But owing to the Greek rhythmical law this is impossible. Accordingly the accent goes back as far as the Greek rules will allow, and we have ἐγω-φαίην.
- The final -αι of the endings -μαι, -σαι, -ται, -νται, and of the infinitive is treated as short. These are all cases in which -αι represents the original final sound of the word. But the -αι of the optative, which is for original -αιτ, counts as long.