50. The Ι-Class. The suffix was probably -ιε(ο) in a pre-historic period of Greek: it appears in stems of the following forms.
- In -ιω, -αιω, -ειω, -υιω or -ῡω (for -ι-ιω, -α-ιω, etc.), the ι blending with the final vowel of the stem.
- With epenthesis of ι, in -αινω, -αιρω (for -αν-ιω, -αρ-ιω).
- With assimilation, in -λλω (for -λ-ιω), -σσω (for -κ-ιω, -τ-ιω), and -ζω (for -δ-ιω, -γ-ιω).
- By compensatory lengthening in -εινω, -ειρω, -ῑνω, -ῡνω, -ῡρω (for -εν-ιω, -ερ-ιω, -ῐν-ιω, -ῠν-ιω, -ῠρ-ιω). That the ει of -εινω, -ειρω is not a true diphthong (and therefore not due to epenthesis) is shown by the corresponding Doric -ηνω, -ηρω.
- In -αω, -εω, -οω, -αυω, -ευω, -ουω (for -α-ιω, etc.).
51. Verbs in -ιω, etc. The verbs in which the original ι becomes ι, thus forming -ιω, -αιω, -ειω, -υιω, are almost confined to the Homeric dialect. The chief examples are as follows.
ἴδιον I sweated
μήνιε be angry
ἀνα-κήκιε gushed forth
κονίο-ντες raisiny dust.
In these verbs (except perhaps the first two) the verb stem ends in ι, so that (e. g.) κονίο-ντες is for κονι-ιο-ντες; so probably τίω I honor, φθίω I waste away, for τι-ιω, φθι-ιω. The ι therefore is naturally long, but may be shortened before a vowel; hence it is usually doubtful in quantity.
- -αω: usually with loss of σ or ϝ.
ναίουσι dwell (aorist νάσ-σα, νάσ-θη)
μαίεσθαι to feel one's way (future μάσ-σεται)
λιλαίεαι desire (λι-λασ-)
καίω (for κᾱϝ-ιω, cp. aorist ἔκηα for ἔ-κηϝ-α)
κλαίω (for κλᾱϝ-ιω)
δαῖε kindled (δᾱυ-)
ναῖον swam (cp. ναῦ-ς)
γαίων rejoicing (γαῦ-ρος, Lat. gau-deo)
ἀγαιόμενος indignant (cp. ἐ-κέρασ-σα, ἠγάσ-σατο, but the σ in these words is not original, § 39.2)
Perhaps also φθαίω (if παρα-φθαίῃσι in Il. 10.346 is present subjunctive, see K. Z. xxiii. 298). δαίω divide forms its tenses from two roots:
2. δατ-, perfect δέδασ-ται, present δατ-έομαι (cp. πατέομαι, πεπάσμην).
πενθείε-τον (probably for πενθεσ-ιε-τον) mourn
τέλειο-ν brought to pass
ἀκειό-μενοι being healed
νεικείῃ-σι shall quarrel
ὀκνείω I shrink
When the diphthongs αι, ει come before a vowel there is a tendency to drop the ι.
ἀγα-ίο-μαι, 2nd plural ἀγάα-σθε (for ἀγά-ε-σθε, § 55)
κερα-ίω, 2nd plural κεράα-σθε
Also τέλε-ο-ν; ναῖον swam also νά-ει, νά-ουσι; perhaps also δάηται shall be destroyed (root δαι-; see Schulze, K. Z. xxix. p. 258). Where this tendency does not show itself, as in παίω, πταίω, σείω, it will usually be found that the diphthong belongs to the whole verb, not merely to the present stem. So perhaps
ἐράασθε you loved
ἕλων drove (participle ἐλάων)
- -υιω: ὄπυιε had to wife(for ὀπυσ-ιω).
Most of the presents in -υω are of this class (original -υιω), as φύω (Aeolic φυίω), θύω (ἔθυιεν Hesych.), λύω, δύω, ἰθύω, ἠπύω, ὀϊζύω. The vowel is doubtful, but only because it comes before another vowel (as was noticed in the case of verbs in -ιω).
ἰθύω generally has ῠ; but ῡ in ἐπ-ιθύουσι (Il 18.175), which ought to be so divided, not ἐπι-θύουσι. It is a denominative from ἰθύς (ῡ) aim.
The verbs in -ευω, -ουω are probably also of the Ι-Class (for -ευίω, -ουιω). For, as Curtius points out (Verb. i. 360), they are chiefly denominatives, and it is contrary to analogy to form a verb by suffixing the thematic ε (ο) to a noun stem.
-νω: μαίνο-μαι, φαίνω, βαίνω (βαμ-ιω), and with reduplication, τι-ταίνω, παμφαίνω.
-ρω: αἴρω, σκαίρω, ἀσπαίρω, μαρμαίρω, καρκαίρω, χαίρω.
αἴρω (for ἀρ-ιω) is distinct from ἀείρω, which by contraction would become ᾄρω: cp. ἀείδω, ᾄδω (Brugmann, K. Z. xxvii. 196).
This class includes also the numerous denominatives in -αινω, -αιρω : see § 120. The stem is in the weak form.
53. Assimilation of ι. Examples:
-λλω : ἅλλο-μαι, βάλλω, πάλλω, στέλλω, τέλλω; from nouns, ἀγγέλλω, ναυτίλλομαι; with reduplication ἰάλλω, ἀτιτάλλω I rear, tend, cp. ἀτάλλω I cherish.
Epenthesis (instead of assimilation) is found in ὀφείλω I owe.
-σσω: ὄσσο-μαι (ὀκ-), πέσσω (πεκ-), ἑλίσσω (ἑλικ-), πτύσσω (πτῠχ-), λίσσο-μαι (λῐτ-), κορύσσω (κορυθ-), πτώσσω (πτωκ-).
-ζω: for -διω in κλύζω, φράζω, χάζο-μαι; for -γιω in ἅζο-μαι, ῥέζω, τρίζω; with reduplication, μιμνάζω I loiter, βιβάζω I cause to go, ἐλελίζω I make to quiver (Il. 1.530).1
54. Compensatory lengthening. Examples: -εινω (for -εν-ιω), in τείνω, κτείνω, θείνω.
-ειρω (for -ερ-ιω), in εἴρω, κείρω, μείρομαι, πείρω, σπείρω, τείρω, φθείρω, ἀγείρω, ἀείρω, ἐγείρω, ἐθείρω.
-ῑνω (for -ιν-ιω), in κλίνω, κρίνω, ὀρίνω.
-ῡνω (for -υν-ιω), in πλύνω, ἐντύνω. -ῡρω (for -υρ-ιω), in κύρω, μύρομαι, φύρω, ὀδύρομαι.
55. Verbs in -αω, -εω, -οω: Assimilation is applied to certain forms of the verbs in -αω, in which, instead of contraction, we find assimilation of one of two concurrent vowels to the other, as ὁρόω for ὁράω, ὁράᾳς for ὁράεις. The chief varieties are as follows.
(a) Forms with simple assimilation, the vowel being long.
μνᾱό-μενοι gives μνωό-μενοι
ἡβάο-ντες gives ἡβώο-ντες
μενοινάω gives μενοινώω
ἠγά-εσθε gives ἠγάασθε
μνά-εσθε gives μνάασθε
μνάῃ gives μνάᾳ (2nd sing. mid.)
(b) With shortening of the first vowel.
ὁράω gives ὁρόω
ἐάῃ-ς gives ἐᾰ́ᾳ-ς
αἰτιάε-σθαι gives αἰτιᾰ́α-σθαι
Cp. δεδάα-σθαι from δεδαέ-σθαι (§ 35) and ἀγάα-σθε from ἀγάε-σθε; future ἐλόω, κρεμόω from ἐλᾰ́ω, κρεμάω.
(c) With lengthened second vowel.
ὁράο-ντες gives ὁρόω-ντες
ὁράοι-τε gives ὁρόῳ-τε
ὁράει-ς gives ὁράᾳ-ς
This is the commonest form of assimilation: cp. δηϊόω-ντο, δηϊόῳ-εν from δηϊόω, ἀρόωσι (Od. 9.108) from ἀρόω, κατ-ηπιόωντο (Il. 5.417), ἐστρατόωντο (Il. 4.378), ῥυπόωντα (Od.).
(d) With lengthened second vowel (the first being also long), in very few forms.
δράουσι gives δρώωσι
μαιμάουσι gives μάιμώωσι
ἡβάουσα gives ἡβώωσα
μενοινάει gives μενοινάᾳ
Other isolated examples are: μενοινήῃσι (Il. 15.82); ἀλόω (Od. 5.377), 2nd singular imperative of ἀλάομαι (for ἀλάεο, ἀλάου); κεκράανται, κρηῆναι, κραιαίνω; φαάνθη (for φαέν-θη); σόωσι (subj.), σόῳς, σόῳ (optative, cp. § 83), σώοντες (σαόω). Similar phenomena may be seen in φόως for φάος (or φᾶος), σόος for σάος, φαάντατος for φαέντατος, νηπιάας for νηπιέας, πρώονες (Il.) for πρήονες, ἀστυβοώτης for ἀστυβοήτης: also in a form Αἰνείωο (for Αἰνείαο) read by Zenodotus in Il. 5.263, 323.
- These forms were regarded by the older grammarians as the result of a process called "distraction," (the exact reverse of contraction), by which a long vowel, ᾱ or ω, could be separated into two distinct vowels (ᾰᾱ, οω, etc.). The first attempt to account for them in a more rational way was made by L. Meyer (K. Z. x. 45 ff.). According to him they represent an intermediate stage in the process of contraction. The order, he argued, is ὁράω—ὁρόω—ὁρῶ: i.e. in ὁρόω the α has been assimilated to the following ω, but is not yet uttered in one breath with it. In the forms ὁρόωντες, ὁρόωσι, etc., he pointed out that the long vowel is never wanted for the meter, and accordingly he wished to read ὁρόοντες, ὁρόουσι, etc. To this last proposal exception was taken by G. Curtius (Erläuterungen, p. 96), who made the counter supposition that, as the α of these verbs was originally long, the successive steps might be ὁρᾱ́οντες, ὁρώοντες and (by metathesis of quantity) ὁρόωντες. The stage -ωο- is exemplified in μνωόμενος.
- The main objection to this theory lies in the circumstance that the forms ὁρόω, ὁράᾳς and the like are exclusively "epic," that is to say, they are confined to Homer, Hesiod, and their direct imitators. If they had been created by any natural development of Greek sounds, we should expect to find them in other dialects. But neither in Ionic nor elsewhere is there any trace of their existence in living speech. It must be admitted, too, that neither Meyer nor Curtius has given a satisfactory account of the long vowel in ὁρόωσι, ὁρόωντο, ὁρόωντες, etc. A form ὁρόοντες, as Curtius pointed out, would give ὁροῦντες, not ὁρῶντες. And if there has been metathesis of quantity, why do we never find ὁρόωμεν for ὁρᾱ́ομεν, or ὁράᾱτε for ὁρᾱ́ετε?
- An entirely different theory was put forward by J. Wackernagel (Bezz. Beitr.iv. 259). The true Homeric forms, in his view, are the original uncontracted ὁράω, ὁράεις, etc., and these have passed into the ὁρόω, ὁράᾳς, etc., of our Homer by a process of textual corruption consisting of two stages:
1. contraction, according to the ordinary rules of Attic, into ὁρῶ, ὁρᾷς, etc.—which would obviously give forms of different metrical value from the original words—and then
2. restoration of the meter by a kind of "distraction" (in the old sense of the term), i.e. the insertion of a short vowel before the new contracted -ῶ, -ᾷς, etc. Thus οὐχ ὁράεις first became οὐχ ὁρᾷς, and then metri gratia οὐχ ὁράᾳς.2
- Paradoxical as this may seem, there can be little doubt that it is substantially right. The forms in question, as Wackernagel justly argues, are not a genuine growth of language. They are the result of literary tradition, that is to say, of the modernizing process which the language of Homer must have undergone in the long period which elapsed before the poems were cared for by scholars. The nature of this process is excellently described and illustrated in his dissertation. In many cases, too, he shows that when the later form of a word ceased to fit the meter, some further change was made by which the metrical defect was cured, or at least disguised. Corruption of this latter kind may often be traced in the various readings of MSS.
But must we suppose that ὁρόω, etc., went through the two changes which Wackernagel postulates?
- The case is unique, not only from the large number of forms involved, and the singularly thorough and systematic way in which they have been introduced into the text, but also from the circumstance which he has himself so well pointed out, viz. their unreal conventional stamp. They are hardly more "modern"—in the sense of being familiar through contemporary speech—than the forms which they have displaced. Wackernagel has shown how ἕως and τέως supplanted the original ἧος and τῆος, even where the result was absolute ruin to the verse; as in Od. 19.367, where nearly all the MSS. have ἕως ἵκοιο. Similarly the loss of the old genitive in -οο (§ 98) has produced the forms Aἰόλου, Ἰφίτου, Ἰλίου, etc. scanned - - - . These examples, however, prove too much; for if such unmetrical forms could remain in the text without further change, why do we never find the slightest trace of an unmetrical ὁρῶ?
- It is a further objection to this part of Wackernagel's theory that in several words the original -αω, -αεις, -αουσα, etc., have been retained. The instances are, ναιετάω, -άει (Hes. Th. 775), -άουσι, -άων, -άοντα, ὑλάει, -άουσι, ἀοιδιάει, -άουσα, ὁμοστιχάει, γοάοιμεν, -άοιεν, κραδάων, ἐλάων, ἱλάονται, τηλεθάοντας; with ᾱ, ἀναμαιμάει, πεινάων, -άοντα, διψάων. (The forms which have lost a ϝ, as λάε, φάε, ἔχραον, do not concern us now.) A third variety is exhibited by the form ναιετάωσαν (-σης,-σῃ, -σας), which occurs in MSS., usually as a variant along with -άουσαν and -όωσαν. These facts are enough to show that the causes which produced the Homeric -οω, -αᾳς, etc., were not of universal efficacy.
- Is there, then, any way from ὁράω, ὁράεις to ὁρόω, ὁράᾳς except through the contracted ὁρῶ, ὁρᾷς? We have to deal with a time when ὁρῶ, ὁρᾷς were the forms of ordinary speech, while ὁράω, ὁράεις were only known from the recitation of epic poetry. Under such conditions it is surely possible that the poetical forms were partially assimilated to the colloquial forms—that ὁράω, ὁράεις were changed into ὁρόω, ὁράᾳς by the influence of the familiar ὁρῶ, ὁρᾷς. Similarly ἑήνδανε for ἑάνδανε was doubtless due to the presence of the later ἥνδανε, not to any process of contraction and distraction. The principle is constantly exemplified in language; cp. the change of φρασί, the original dative plural of φρήν, into φρεσί through the association of the other case forms.
- With this modification of Wackernagel's view it is easier to account for the occasional retention of the original -αω, -αεις, etc. If ὁρόω, ὁράᾳς are due to the presence of ὁρῶ, ὁρᾷς in everyday language, we may expect to find a different treatment of words which went out of use in post-Homeric times. Thus ναιετάω does not pass into ναιετόω because there was no ναιετῶ alongside of it in common use. Similarly ἐλόω, ἐλάαν are accounted for by the Attic ἐλῶ, ἐλᾶν; but the Homeric present participle ἐλάων is unaffected. Two instances call for a different explanation, viz. πεινάω and διψάω, since they are not rare or poetical words. But these are exceptions which prove the rule. As is shown by the Attic contraction (πεινῇς, etc.), they are not really verbs in -αω. Whatever may be the origin of the ᾱ in the Homeric πεινάων, διψάων, etc., they do not belong to the group with which we are now concerned.
- An example of the process supposed by Wackernagel may be found in the Homeric τρωπάω, τρωχάω, στρωφάω, πωτάομαι (as to which see Nauck, Mél. gr.-rom. iv. 886). The forms which occur are always contracted, but in every instance except one (Il. 13.557 στρωφᾶτʼ) the uncontracted form can be restored if at the same time the root vowel is shortened. Thus in Il. 15.666 μηδὲ τρωπᾶσθε φόβονδε we may read μηδὲ τροπάεσθε φόβονδε. The verb πωτάομαι only occurs once (Il. 12.287 λίθοι πωτῶντο θαμειαί), while the form ποτάομαι is well attested. In the other cases the restoration is supported by etymology (τροπάω from τροπή, etc.), and by the considerable traces of τροπάω, τροχάω, στροφάω in our manuscripts (see Leaf on Il. 15.666). The process must have been that (e.g.) original τροπάεσθε became τροπᾶσθε (which is also found in MSS.), and then τρωπᾶσθε.
- In the imperfect active assimilation is unknown, mainly because the meter generally allows contraction. We find however (1) several uncontracted forms, viz. οὔταε (Od. 22.356), πέραον (Il. 16.367), ὕλαον (Od. 16.5), κατεσκίαον (Od. 12.436): ἐχράετε, ἔχραον (for ἐχράϝετε ἔχραϝον) do not belong to this head. Also (2) some verbs show the New Ionic -εο- for -αο-, viz. ὁμόκλεον, ὁμοκλέομεν, ποτέονται, μενοίνεον, ἤντεον, τρόπεον.
For φάος we find the two forms φόως and φώως (Il. 16.188 ἐξάγαγεν φώωσδε), but never φόος or φῶος.3 The exclusion of φῶος is remarkable, since it is related to φᾶος as μνωόμενος to μνᾱόμενος. The reason doubtless is that φᾶος came under the influence of φῶς (cp. ὁράᾳς and ὁρ-ᾷς). On the other hand σάος became σόος owing to the later σῶος. The change of πρηόνες to πρώονες is similarly due to πρῶνες. In the case of ἀστυβοώτης (for -βοήτης) there is no evidence of a form -βώτης, but such a form would be according to the rules of Ionic contraction (βώσας for βοήσας, etc.).
56. Verbs in -αω, -εω, -οω: Contraction. The extent to which contracted forms of verbs were admitted in the original text of Homer is a matter of much dispute. In this place we are properly concerned only with verbs of the Ι-Class (-αω, -εω , -οω, for -α-ιω, -ε-ιω, -ο-ιω), not with those in which a different spirant has been lost (as τρέω for τρέσ-ω, πλέω for πλέϝ-ω).
- In the verbs in -αω contraction is frequent. If the resolved form were written wherever the meter admits it, we should still find that in about half the whole number of cases the contraction must remain. It is worth notice too that contracted forms are often used in phrases of a fixed type, as ἔπεα πτερόεντα προσηύδα (or προσηύδων)—τόδ’ ὀφθαλμοῖσιν ὁρῶμαι—ὁρᾷ (ὁρᾶν) φάος ἠελίοιο—ἀνείρεαι ἠδὲ μεταλλᾷς—ἐξαύδα, μὴ κεῦθε, and the like.4 It has indeed been noticed that there is an apparent preference for the resolved -αον of the 1st singular and 3rd plural imperfect;5 but this must be accidental. We must conclude then that contracted and uncontracted forms of verbs in -αω were used in the language of Homeric times with equal freedom, or at least—if this be thought improbable—that they subsisted together as alternative forms in the poetical dialect.
- Verbs in -εω rarely contract -εο or -εω, except in the participle (-ευμενος for -εομενος). This rule is confirmed from New Ionic inscriptions (Erman, Curt. Stud. v. 292), as well as the MSS. of Herodotus. For ευ in ποιεύμην (Il. 9.495), θηεῦντο (Il. 7.444), ὀχλεῦνται (Il. 21.261), ἐγεγώνευν (Od. 9.47, etc.) and a few similar forms we should write -εο (see § 57).
The contraction of -εε, -εει is established by the large number of instances6 in which it is required by the meter. Moreover it is not merely a license, necessary for the sake of admitting certain forms into the hexameter (such as ταρβεῖς, νεικεῖν, τελεῖται, ἡγεῖσθαι, σμαραγεῖ, ἐφίλει, οἰνοχόει). Among the instances of contraction in the last foot we find 29 of -ει for -εε (as χόλος δέ μιν ἄγριος ᾕρει), and 16 of -εῖ for -έει (as καί με γλυκὺς ἵμερος αἱρεῖ); also the forms φιλεῖ (Il. 2.197 τιμὴ δʼ ἐκ Διός ἐστι. φιλεῖ δέ ἑ μητίετα Ζεύς, also Il. 7.280, 10.245, 552, 16.94, Od. 15.74), δοκεῖ (Od. 2.33, and six times in the phrase ὥς μοι δοκεῖ εἶναι ἄριστα), τελεῖ (Il. 4.161), καλεῖ (Il. 3.390, Od. 17.382), φοβεῖ (Il. 17.177). On the other hand the uncontracted form has the support of the meter in about a hundred places, and against the instances now quoted of φιλεῖ, etc., we have to set about thirty of the corresponding uncontracted φιλέει, δοκέεις, -ει, τελέει, καλέει, φοβέειν. The uncontracted form therefore seems to have a slight preference, when the meter allows either.
In the MSS. of Homer contraction is generally introduced as far as possible, according to the tendencies of Attic; but the open forms occasionally survive, chiefly in the fourth foot (in such forms as προσεφώνεε θεῖος ὄνειρος—καὶ ᾔτεε σῆμα ἰδέσθαι—κατὰ δʼ ᾕρεε Πηλείωνα). And the meter clearly points to the open form in several other places.
Il. 11.553 (= 17.663) τάς τε τρέει ἐσσύμενός περ.
Il. 21.362 ὡς δὲ λέβης ζέει ἔνδον κτλ.
Il. 16.201 ἀπειλέετε Τρώεσσιν.
Od. 10.548 ἀωτέετε γλυκὺν ὕπνον.
3. Verbs in -οω generally contract; χολοῦμαι, κορυφοῦται, γουνοῦμαι. For the "assimilated" forms δηϊόωντο, κατηπιόωντο, ἐστρατόωντο, ῥυπόωντα (§ 55) we ought, on the analogy of the verbs in -αω, to substitute δηϊόοντο, etc.
57. Synizesis. The vowel ε sometimes coalesces with a following ο or ω, so as to form one syllable for the purpose of the meter; e. g. ἀελπτέοντες, ἠλάστεον, ἠγίνεον, ἐπόρθεον (at the end of a verse), οἰκέοιτο, εἰλέωσι, χρεώμενος. Whether the pronun-ciation of these words differed from that of the contracted forms is a question which perhaps there are no means of determining.
- 1. Cobet (Misc, Crit.), following Bentley, has sought to show that the forms of ἐλελίζω belong in reality to ἑλίσσω (ϝελίσσω). He is doubtless right in substituting ϝελιχθέντες for ἐλελιχθέντες wheeling about: but it seems necessary to retain ἐλελίζω where the meaning is to set trembling (with intensive reduplication, like ἀκαχίζω, ὀλολύζω, etc.).
- 2. This theory was criticised by Curtius in the Leipziger Studien, iii. pp. 192 ff.
- 3. φόωs may represent an ancient plural φάως (Joh. Schmidt, Plural. p. 142).
- 4. Mangold, Curt. Stud. vi. 194.
- 5. Menrad, pp. 122-124.
- 6. About 160 according to the list in Menrad, pp. 132-142.