If we try to summarize our findings on the “mixed aorists,” the most evident fact is their heterogeneity. The problems appear to be morphological; and yet the first set of forms treated (οἴσετε, ἄξετε, ὄψεσθε, etc.) could not be explained on the morphological level. It was necessary to move to the syntactical level to show how οἴσετε was created by a transformation of a phrase with the future participle and a verb of going:

This redistribution of inflexional morphemes evidently represents an effect of the verse medium on the language, a pressure for economy of expression with special emphasis on the initial position of the verse, so we would not expect the same thing to happen in natural language. We might compare, in English verse, the permissible distortions of normal word order (word order in English having similar functions to inflexional endings in Greek). For example, in Milton’s Paradise Lost 1.237f: “Such resting found the sole/ Of unblest feet,” the subject and object are inverted in a way which would be impossible (without changing the meaning) in prose.

With ἐδύσετο and ἐβήσετο, on the other hand, we have a phenomenon, the secondary future or conditional tense, which has developed in natural language in various Indo-European languages. The Homeric evidence allows us to detect an original secondary future or past desiderative sense under the prevailing aoristic or imperfect usage, but it does not tell us whether this secondary future was ever used in natural Greek or was always restricted to poetry.

The process which created ὄρσεο and λέξεο is basically one which operated quite generally in Greek, namely the thematization of athematic forms. The only reason why the creation of ὄρσεο and λέξεο is peculiarly poetic is that it started from forms (ὄρσο and λέξο) which were archaic survivals in poetry, of which the original segmentation was no longer apparent. In natural Greek the process of thematization would not have started from isolated forms but would have worked through a whole paradigm. A similar case, where regularized inflexions have been based on an obsolete form, may be seen in archaizing English: the verb “to wit” (I wot, thou wost, he wot) has received such forms as “wottest” and even “wotted.”

The creation of ἷξον follows a process which, though not otherwise known in Greek, did occur in Sanskrit. An original thematic aorist becomes “sigmatized” on the model of other verbs which show s before a vowel opposed to no s before a consonant. The condition for this process in Greek is the existence of formulae with ἷκε and *ἷκον which metrically prevent the substitution of the normal aorist ἵκεο or ἵκοντο. Thus it is limited to the language of epic poetry.

From these diverse phenomena, we can see that the differences between the poetic and natural languages are real but not simple to define. Rarely, poetic style may systematically alter normal grammar; most often the poetic features represent the preservation of synchronically unmotivated archaisms and their further use as bases for analogical innovation. When the base form is unmotivated, even if the analogical process is the same as in the natural language, the resulting form will often differ from the natural development.

Additional notes:

p. 45 η 135 εἴσω with genitive is late: Witte Zur homerischen sprache 30

p. 48 cf. Mimnermus 10.11

p. 49 η 189 The sun has to be up for a while longer, as Odysseus meets Nausicaa and goes to the city. Aristarchus reads δείλετο “was afternooning” (a word otherwise unattested) because he did not understand δύσετο “was about to set.”

p. 65 Sappho θυρωρωι ; contraction implies that it was s , not w , that fell: Lobel Σάπφους μέλη xxxiv