The meaning of Homeric εὔχομαι through its formulas

Chapter I. Introduction

[In this on-line version, the page-numbers of the printed version are indicated within braces (“{” and “}”). For example, “{69|70}” indicates where p. 69 of the printed version ends and p. 70 begins. These indications will be useful to readers who need to look up references made elsewhere to the printed version of this book.]

Il faut bien dire cependant que notre connaissance du vocabulaire homérique est encore dans l’enfance. Nous avons reçu de l’antiquité un système d’interprétation auquel on continue de se tenir et qui marque nos lexiques et nos traductions. Tandis qu’un effort considérable a été employé à restaurer un texte sûr et à définir les caractéristiques dialectale de la langue épique, nos interprétations restent largement celles d’une époque où les conventions esthétiques primaient le souci de l’exactitude. Mieux on étudie les textes homériques, plus on aperçoit la distance entre la nature réelle des concepts et l’image qu’en donne la tradition scolaire.

Benveniste [1]

The primary purpose of this study is to explain the meaning of εὔχομαι in Homer. For an initial frame of reference, here are the accepted definitions of εὔχομαι and its cognates:

The article in LSJ9 s.v. εὔχομαι provides a convenient point of departure for an introduction to the problems which surround this word. Its order of presentation implies that the sacral meaning ‘pray’ is primary, that this was transformed by way of the notion ‘wish’ to ‘vow’ and then secularized to ‘boast’, ‘brag’, and finally ‘profess’. The progression from sacral significance to secular has intrinsic probability and many parallels. However, the opposite sequence can also occur, e.g. Myc. te-me-no, Hom. Gk. τέμενος ‘piece of land held by kings or public officials’ designates ‘temple land’ or ‘sacral precinct’ from the 6th and 5th centuries onward. [4] Another alternative is that the sacral and secular meanings both developed from an original sense ‘say’ which is unattested in Greek but does occur in Avestan aoj-. Furthermore, it is conceivable that ‘vow’ is primary and ‘pray’ secondary, especially in view of the Latin cognate of εὔχομαι. Finally, in the secular lemma, why not suppose that neutral ‘profess’ comes first—if necessary as a secularization of ‘vow’ or ‘pray’—followed by unjustifiable ‘boast’ and then ‘brag’? Is the direction of semantic development inevitably from expressive to neutral significance? Contrast συμφορά ‘conjunction, event’ which came to mean solely ‘disaster, disease, offense, etc.’ [5] or ἀρά 1. prayer 2. vow 3. curse. [6] There are two common ways to resolve such historical and conceptual problems, but neither is convincing. The first is to make a chronology of attestations, a dubious process for this word, since all its meanings are supposedly Homeric. It suggests still another alternative. Perhaps the legal sense (‘declares solemnly’) attested in Mycenaean and Homeric Greek is the original meaning from which the others derive. A second rule of thumb, that the statistical majority of attestations determines a word’s primary sense, is extremely unreliable. Archaisms are generally rare.

Clearly, a schema of this sort, which is mainly intended to be useful to readers of texts, relies on probabilities. But that is the basic difficulty with {10|11} this word, the improbability of having a single word which can mean both ‘boast’ and ‘pray’, and it threatens the veracity of the whole definition. Furthermore, similar objections arise less excusably with regard to more technical investigations of εὔχομαι. Five detailed studies of this word appeared between 1965 and 1972, and each of them supports and defends one of the four alternatives just suggested to the definition in LSJ9. Benveniste devoted a chapter in his latest book to the contention that ‘vow’ is the primary meaning of the Indo-European root of εὔχομαι. [7] The author of a dissertation on Greek words for prayer, André Corlu, argues on the basis of comparative evidence that the sacral and secular senses of εὔχομαι are both developments from an original sense ‘say’. [8] On the other hand, Citron, in a dissertation on εὔχομαι and σπένδω, and Adkins, in a long article on εὔχομαι and its Homeric derivatives, both consider the social or secular sense of εὔχομαι its oldest sense. [9] Finally, the most recent discussion of this word, by J.-L. Perpillou, plumps for priority of the legal sense attested twice in Mycenaean and once (Σ 499) in Homer. [10] Apparently all the alternatives have been exhausted. And each defense involves not just a different point of view on the special problem of this word’s prehistory, but radically different interpretations of its synchronic meaning in the text of Homer. Adkins, for example, believes that εὔχομαι never means ‘pray’ in Homer and always has one, basic secular sense. [11] Similarly Perpillou denies that it is a specifically religious word, yet to him its consistent, basic meaning is always juridical. [12] Other treatments adhere to the same {11|12} definitions of εὔχομαι but conflict about its etymology. [13] The value of etymology thus becomes academic.

There are two features of method which are shared by all these studies. The first is that their results are mutually exclusive and never concur. [16] The second is their total neglect of the most important research on Homeric language carried out in this century. The theses of Milman Parry, [17] which are now almost fifty years old, point the short way to a more rigorous procedure for semantic analysis, and it is a procedure which is not difficult to execute or original to invent. Begin with a systematic survey of the Homeric attestations and a minimum of a priori notions as to meaning or semantic development. Objective determination of both can be maximized by the study of individual contexts, where other, known words guide and control intuition, and by the accumulation of such contexts and their disposition into objectively homogeneous classes. The formulaic nature of epic language, which Parry described once and for all in studies that are models of clarity and rigor, places words in fixed metrical and lexical contexts and thus facilitates the processes of objective classification and internal analysis. And for the semantic analyst to take account of the nature of the medium in which his words appear, especially when it can maximize the objectivity of his research, is certainly good methodology. Once the attestations of the word have been formally classified, the best guides to its function are analysis of the constraints—grammatical, stylistic, contextual—on each usage group, formal comparison of one usage {12|13} group to another, and comparison of a given formula with its transformations. But from then on, the viable principle to follow is, with Benveniste, good sense informed by “le contact et l’inspiration des emplois vivants.” [18] The result should be compatible with etymological evidence which confronts text with text, not simply definition with definition.

In addition, I have found another Indo-European poetic collocation in the series inaugurated by Adalbert Kuhn’s discovery in 1853 [21] of an exact {13|14} Sanskrit cognate to Homeric κλέος ἄφθιτον. There are by now enough examples of poetic expressions attested in common by the Indo-European languages to validate the existence of an Indo-European poetic language and to show that Greek poetic language, both Epic and Lyric, participates in its tradition. [22] There is also a strongly supported case for the Indo-European origin of Greek Lyric meters, for parallels to them have been found in the Slavic, Vedic, and Celtic poetic traditions. [23] Furthermore, Homeric scholarship now has a working hypothesis for the origin of the dactylic hexameter as a prehistoric development from a Lyric meter with Indo-European prototypes. [24] This reconstruction by G. Nagy, which is inspired by the work of Parry and embodies bold advances in securing the methodology of comparative metrics, at last explains the preservation in Greek epic of Indo-European poetic language. The result is that Homeric scholarship can no longer ignore the perspective granted by comparative Indo-European evidence. If we wish to understand the meaning of these traditional poems as well as their technique, we must learn what can be learned of their prehistory. I hope that the concrete evidence I have presented will bring home forcefully to the reader the significance and value of historical research. The close contextual, formal, and functional correspondences between attestations of Homeric εὔχομαι and its cognates in the poetic traditions of Indic and Iranian cast doubt on the view that inherited archaisms in Greek epic are merely antique curiosities.

The Homeric past, both Mycenaean and Indo-European, and the transmission of the text are two important concerns of the present work. More fundamentally, I intend it as a contribution to basic methodology in the study of what words mean in Homer. At the head of this chapter stands a statement by Emile Benveniste that our knowledge of Homeric vocabulary is in its infancy. That knowledge will mature more rapidly, I submit, if investigators take advantage of the formulaic nature of Homeric language. In fact the search for an adequate method has made this thesis essentially an attempt to relate the discoveries of Parry, Lord, and their followers in the study of epic technique to the study of Homeric vocabulary. To support his theory that the formulaic diction of Homer was traditional, Parry concentrated on a provocative problem which faces the semantic analyst of Homer, the relative contextual independence of ornamental epithets. Yet his discoveries are of great importance for the meaning of Homeric words in general. For example, when a poet’s units of expression are not single words but metrically fixed groups of words, and when these groups of words have existed before him in a tradition, any single word (with only the possible exception of some particles) has maintained or acquired in time a sense and context which is more rigid, resonant, and intricate than it might be for a poet who lacks such a medium. This is what one critic articulately responds to as “the shining specificity” of epic diction. [29] Even the complex lemma in LSJ9 for εὔχομαι is a crude indicator of its specializations, an inheritance of that scholastic tradition which suppresses “le souci de l’exactitude.” Another important and contrasting aspect of the meaning of single words in formulaic diction is the apparent blurring of semantic distinctions between them. It is at times impossible to apply the assumptions of semantic economy which a natural language validates to the Homeric Kunstsprache. To take an extreme case, consider the difficulty in showing a difference in meaning between προσέφη, προσεφώνεε, προσέειπεν and προσηύδα in their Homeric attestations. In Homer different words can be semantic but not metrical synonyms, while in non-formulaic poetry, neither kind of synonymy is possible. Obviously semantic analysis cannot be effective without an awareness of such considerations of technique. And at times the alternatives to adjudicate between are more complex. How can one account for differences in contextual distribution between words which are apparently formulaic coefficients, which otherwise differ only in the immediate prosodic environments to which they are adaptable? And how can a classification scheme based on formulas account for and describe language which is not precisely paralleled and does not fit strict or flexible definitions of a formula? The language of enjambment is a case in point which will be of some concern. Another difficulty: how to approach doublets, expressions which recur only once in the Homeric corpus? {16|17}

These important questions are left unraised in the large bibliography of εὔχομαι and also in general studies of epic technique, where the principal concern is form, not semantics. To fill this void, my investigation of the meaning of εὔχομαι has become a case study in how the traditional poet uses a repertoire of formulas and themes to express his meaning, in short, an investigation of poetics. I hope the result will be useful for further research by attaining a level of general applicability, and that in the process we may deepen our understanding of the beautiful Homeric poems in some particulars.


[ back ] 1. Benveniste 1969: t. 2, 58.

[ back ] 2. For the terms Theme I and Theme II, and the convention of *w (vs. *), see Benveniste 1935:147–173. The sources of the definitions given are as follows: Greek/Liddell and Scott 1940 (hereafter abbreviated as 9): s.vv. Mycenaean/Palmer 1963:421 s.v. e-u-ke-to Avestan/Bartholomae 1904, s.v. aoj-. Vedic Sanskrit/Grassmann 1873, s.vv. ūh-, vāghát-, Latin/Lewis and Short 1879, s.vv. Umbrian/Vetter 1953:446 s.vv.

[ back ] 3. Grassmann 1873, s.v. vāghát- also offers a definition ‘betend, opfernd’ which is to be rejected on formal grounds. The word is a substantive, not a participle: see Wackernagel and Debrunner 1954:160.

[ back ] 4. For the meaning of Myc. te-me-no, see Palmer 1963:456 s.v., and on τέμενος see LSJ9 s.v.

[ back ] 5. LSJ9 s.v.

[ back ] 6. ibid. s.v.

[ back ] 7. Benveniste 1969: t. 2, 233-43.

[ back ] 8. Corlu 1966:17–18. He does not develop the idea, and his supporting hypothesis that the Indo-European roots of ἔπος (*əwekw-) and εὔχομαι (*ə1ewgwh– / *ə1wegwh-) were related (so also Perpillou 1972:182) should be treated with caution. If Hittite ḫwek– ‘beschwören’ is related to ἔπος, then the initial laryngeals in these two roots were different in Indo-European. ḫwek-presupposes *ə2wekw-. In any case, the reconstructions show that *əwekw– and *ə1ewgwh- 2wegwh– were lexically distinct in the protolanguage. (I owe this information to Professor Calvert Watkins.)

[ back ] 9. See Adkins 1969 and Citron 1965:115, who offers this “Grund-definition”: “zu seiner Ehre und bei seiner Ehre erklären.”

[ back ] 10. Perpillou 1972. The second, third, and fourth chapters of my work appear largely in the form they had before the publication of this important study. However, I have tried to take it into consideration wherever it was possible to do so without making major alterations in those chapters.

[ back ] 11. With exceptions such as ‘vow’, which he does not account for. Adkins does not specifically discuss the word’s etymology, but tries to prove his point of view by exegesis of texts and to justify it with generalizations about the differences between Homeric society and ours. He postulates the absence of a distinction between sacral and social speech in the Homeric world and thus solves the problem of the divergent meanings ‘pray’ and ‘boast’.

[ back ] 12. See Perpillou 1972:170, 177.

[ back ] 13. For example, Citron 1965 and Corlu 1966 give the word essentially similar meanings synchronically.

[ back ] 14. So Corlu 1966 and Citron 1965. See the previous note.

[ back ] 15. He does not know the crucial evidence in Linear B; see Perpillou 1972:169, and below, Ch. IV.

[ back ] 16. It should be said that these authors are, with only one exception, ignorant of each other’s work. Perpillou cites two (Corlu and Benveniste) of his four immediate predecessors. The others cite none at all.

[ back ] 17. Parry 1928a, 1928b.

[ back ] 18. Benveniste 1969: t. 2, p. 250, and, for ‘le bon sens’, Benveniste 1966:289–307.

[ back ] 19. Van der Valk 1964: 267. See also Davison 1962: pp. 258–9. He considers it a necessary assumption for the Unitarian critic that “the poems were preserved in writing from the moment of their composition” (p. 258). Van der Valk and Davison overestimate the freedom of rhapsodic transmission, it seems to me, both in its consequences for scholarship and in its effects on the unity of the poems.

[ back ] 20. For an intolerant approach, see Finley 1957:133–159, especially p. 159: “Homer is not only not a reliable guide to the Mycaenean tablets; he is no guide at all.” The study of Mycenaean language and society can be valuable for Homeric scholarship freed of euhemeristic or literal-minded predilections. As an example, Nilsson 1933, although Linear B had not been deciphered.

[ back ] 21. Kuhn 1853:467.

[ back ] 22. See the excellent compilation of bibliography in Schmitt 1967.

[ back ] 23. See Meillet 1923, Jakobson 1952, Watkins 1963.

[ back ] 24. Nagy 1974.

[ back ] 25. Clader 1973.

[ back ] 26. Boedeker 1975.

[ back ] 27. Frame 1971.

[ back ] 28. Shannon 1975.

[ back ] 29. Finley 1966:12.