Catherine P. Roth: “Mixed Aorists” in Homeric Greek: Introduction

Introduction

The Homeric poems provide some of the easiest reading in Greek literature, as well as some of the most rewarding, and so we are introduced to them at an early stage in our study of the language. But when we learn more, we discover that Homeric Greek is not so simple after all. Some of its phenomena remain unexplained after two millenia of scholarship. For instance, we come across imperatives like οἴσετε, ἄξετε, ὄρσεο, λέξεο, δύσεο, βήσεο and secondary tense forms like ἄξοντο, ἐδύσετο, ἐβήσετο, ἷξον. When we look in the usual grammar books, Smyth for example, [1] we find these forms labeled “mixed aorists,” and are told that they combine the sigma of the first aorist with the thematic vowel of the second aorist. That description is an admission of ignorance. How could such a “mixture” happen? What kind of process is being attributed to the grammar of the epic language? What is Homeric grammar anyway? How can one tell what is grammatical and what is not?

In order to answer such questions, we must start by understanding the peculiar character of the epic language: its basic units are not so much single words as formulae ready-made to fit various parts of the hexameter verse.

This approach to Homeric diction was chiefly developed by Milman Parry. [2] Earlier scholars had seen that the language is very much affected by the meter, and for that reason differs from a natural language. Kurt Witte (1913) declared that “die Sprache der homerischen Gedichte ein Gebild des epischen Verses ist.” [3] Karl Meister (1921) used the term Kunstsprache to indicate the artificiality of the epic dialect. [4] The meter produces both conservatism and innovation. It causes the preservation, as fixed formulae, of archaisms which have long been dead in the natural Greek language. Thus, for instance, the lost consonant digamma is often represented by hiatus in a formula, e.g.μελιηδέα (F)οἶνον. Furthermore, archaic formulae can be inflected as the syntax may require, and the inflexion may violate the conditions of their original formation. [5] Thus the genitive μελιηδέος οἴνου no longer allows for digamma (the irresolvable contraction in οἴνου is also relatively late in the Homeric language). Inflexion may even produce a form which is “wrong” from both diachronic and synchronic (in the natural language) points of view, as the dative εὐρέϊ πόντῳ leads to the accusative εὐρέα πόντον (the regular accusative being εὐρύν).

Is εὐρέα synchronically grammatical in the epic language?

Doubts about the concept of synchronic grammaticality, as applied to the epic, are even more seriously raised when οἴσω and ἄξω are given imperatives, when ἄξω, δύσομαι, and βήσομαι are given imperfects, or when such morphological oddities as ὄρσεο and ΐξον are produced. In each case we can trace the chain of analogy,which led to the observed result. The processes are not at all arbitrary: nothing is invented solely metri gratia. And yet one suspects that such analogical chains could be extended indefinitely without running into any limits of grammaticality. In a natural language, the native speaker has command of a grammar which enables him to judge the grammaticality of a given sentence; but nobody was a native speaker of epic. Every poet was imitating preceding poets. [6] There were two basic rules: one could say anything which had been said before, and one could create new expressions by analogy with those that already existed. If a base formula was synchronically unmotivated, an analogical adaptation of the formula was not subject to any control. So the epic language consisted of a large stock of unmotivated formulaic expressions augmented by unrestrained analogical development. In brief, there is no such thing as a synchronic Homeric grammar.

The method which leads to such a conclusion may illustrate both the necessity and the difficulty of using philological evidence to solve a linguistic problem. Clearly, the philological evidence cannot be ignored; as Watkins says, “auch eine morphologische Form in ihrer Besonderheit nur im Kontext gewertet werden kann. Momente wie die Stellung einer Form im Vers, ihr Vorkommen in wiederholter Formel, auch das genre der Quelle – kurz, philologische Daten – können an sich Indizien für sprachliche Altertümlichkeit seine.” [7] Or, of course, they may show that a particular form is not ancient. Those who have neglected this kind of information (or who have used it badly) have often confused poetic innovations with archaic inheritances, and have then misapplied comparative evidence to posit non-existent Indo-European categories. Both the thematic sigmatic aorist and the secondary future may be relegated to the histories of the attested languages.

For the literary scholar, we may extend our view of Homeric language to the scenes and stories from which, on a larger scale, the epics are constructed. When we do this, the “Homeric question” loses its significance. Each poet was working within a tradition of language and legend, and was developing it in traditional ways, by expanding, elaborating, and combining the given materials. He tried to tell the stories better than his predecessors had done, and so changes inevitably crept in. This being the case, the notorious inconsistencies in the epics cause less difficulty. If a poet invented everything for himself, he ought to be consistent; but if he is using a great mass of material worked out by many different predecessors, perfect consistency can hardly be expected. The analytical hypothesis thus becomes unnecessary. We can allow the overwhelming sense of coherent artistic purpose and faithful observation of character to persuade us that we have the work of one mind shaping the tradition in an exceptionally successful way.