Epilogue. The etymology of εὔχομαι

The semantic analysis of Σ 499 has definite implications for the problematic prehistory of Greek εὔχομαι. This is a subject of consequence to the internal analysis of εὔχομαι in Homer, which is now complete. True, etymology is, from the viewpoint of method, secondary to understanding texts. But the validity of that understanding is in jeopardy if external etymological evidence cannot be related to it by common sense. Furthermore, the research I have undertaken is potentially more productive if its results apply to other Indo-European languages.

First there is the problem of internally reconstructing a plausible semantic antecedent for the various usages of Homeric εὔχομαι. The analysis of Σ 499 in Chapter IV has provided us with a point of departure. A formal link actually exists between the legal and secular usages of εὔχομαι, namely the collocation of εὔχομαι and φημί. This formal link suggests that the prehistoric functional link between secular and legal εὔχομαι was the notion ‘say (in a functionally marked context)’. Such a meaning for εὔχομαι is implicit in the collocation where εὔχομαι is the functionally marked, φημί the functionally unmarked word for ‘say’. [1] Thus we can make the simple hypothesis that an original meaning ‘say (in a functionally marked context)’ became specialized in two functionally marked contexts, one secular, resulting in the meaning ‘say (proudly, contentiously, accurately)’, the other legal, perhaps resulting in a meaning ‘state, plead, allege’.

The foregoing hypothesis is a straightforward application of the comparative method with results that adequately account for the attested phenomena. Alternatively, one could suppose that legal εὔχομαι is derived from secular εὔχομαι, or vice versa. But it is not necessary to do so. Both usages could have developed independently of each other. And there is no evidence in Greek as to which of these usages of εὔχομαι, if either, did come first. [2] No evidence has come to light, in this investigation, of phraseological borrowing in Greek from one usage to the other. Instead, the εὔχομαι—φημί {107|108} collocation, which is the only instance of phraseological correspondence between the two, argues for a meaning which is conceptually anterior to both, not one which was originally appropriate to one or the other.

The question arises whether this view is reconcilable with evidence from the other Indo-European languages and with the sacral sense of εὔχομαι. But more immediately, it should be reconcilable with the usage and sense of εὖχος, the only nominal derivative of the root εὐχ- which does not seem to occur in a sacral sense. There is disagreement about the meaning of this word: ‘Ruhm’ (Frisk, Citron), ‘gloire’ (Chantraine); ‘I. object of prayer, II. boast, vaunt’ (LSJ⁹); ‘le nom institutionnel de l’avantage recherché [en combat]’ (Perpillou). [3] The confusion appears needless, for εὖχος is in complementary metrical distribution with κῦδος, as formulaic analysis reveals:

I.

Ia.

II.

IIa. These lists match all the attestations of εὖχος with representative κῦδος attestations. From a comparison of I and Ia, it is clear that εὖχος and κῦδος are simply prosodical variants which are never metrically interchangeable. In II vs. IIa, the situation is less clear. The formula systems are functionally complementary, and κῦδος could not substitute for εὖχος in II., but εὖχος could substitute for κῦδος without violation of meter in IIa. In fact, εὖχος in IIa would improve the rhythm by alleviating the overlengthening in Ζεὺ̅̅ς̲ | κ̲ῦδος. [6] But the situation is exactly paralleled elsewhere in the Homeric corpus, cp. ποδάρκη̅ς̲̅ | δΐος Ἀχιλλεύς # (vs. ὠκὺς Ἀχιλλεύς), or πολύτλᾱ̅ς̲̅ | δ̲ῖος Ὀδυσσεύς (vs. ἐσθλὸς Οδυσσεύς). These cases of ‘uncorrected’ overlengthening are probably prosodic and phraseological archaisms embedded in the poetic repertoire. It is possible to rationalize them by considering εὖχος, ἐσθλός, and ὠκύς as metrical and functional alternatives to κῦδος and δῖος but not vice versa. Consider the large statistical predominance of κῦδος (64 x) vs. εὖχος (19 x). But the real point is that Ζεὺς κῦδος ἔδωκεν # was an immutable phrase to the traditional poet, and the substitution of εὖχος for κῦδος never occurred to him even in spite of the rhythmic anomaly. Neither does it prove that εὖχος and κῦδος differed from each other in meaning. As the formulaic analysis suggests, the two words occur in plainly identical contexts. [7] Moreover, the phraseology surrounding εὖχος and κῦδος is paralleled in only one other epic word, κλέος.

Compare the following:

{109|110} I conclude that εὖχος, κῦδος, and κλέος belong to a common semantic field. κλέος, however, has certain archaic epithets (ἐσθλόν, μέγα, εὐρύ, ἄφθιτον) which separate it prosodically from κῦδος and εὖχος, and it has separate functions as well. [8] The epithets (μέγα, ὐπέρτερον), verbs (δίδωμι, ἄρνυμι, ὀρέγνυμι, ἀπηύρα), and contexts of εὖχος and κῦδος are interdependent and imply their semantic equivalence. If κῦδος means ‘glory’, ‘fame’, then so must its formulaic variant εὖχος.

This is not to deny that εὖχος can on occasion have thematic and secondary (i. e. on a level below the actual formula) dictional associations which differ from those of κῦδος. In general, such associations are crucial aspects of a word’s synchronic poetic effect in Homer, but they can be only synchronic and thus misleading in a diachronic analysis such as we are now concerned with. I propose a minimalistic axiom for diachronic semantic analysis in Homer: the testimony of thematic and dictional associations which are secondary to a word’s formulas is inadequate to controvert the testimony of that word's inherited formulas.

For example, there is one instance of εὖχος in which κῦδος could fit the metric just as well: Formally speaking, this line has the characteristics of doublet composition (sic, in spite of its three attestations), namely improvisational x-sector phraseology and two (E 654, Λ 445) repetitions in a fixed context of three lines. [9] The poet’s choice of εὖχος instead of κῦδος has been motivated by thematic and secondary dictional associations in the context. First, it cannot be a coincidence that all three attestations occur as the last line in the challenge speech of one hero to another, speeches which commonly contain statements of a hero’s γένος or γενεή introduced by secular εὔχομαι. [10] Moreover, one of the attestations of this line (Λ 445) is followed four lines later (Λ 449) by secular εὔχομαι introducing a speech of death exultation. [11] One might well be tempted to translate εὖχος in this line, with LSJ⁹, ‘boast, vaunt of superiority’, or, with Perpillou, to consider it ‘le nom institutionnel de l’avantage recherché [en combat]’. [12] However, it is also clear that this line is associated in the poet’s mind with another line which uses sacral εὔχομαι: {110|111} The parallelism in compositional technique, [13] diction, and arrangement (# εὖχεο … κλυτοτόξῳ # vs. # εὖχος … κλυτοπώλῳ #) cannot be coincidental, especially in view of the fact that κλυτόπωλος is attested only here in all Greek literature. It is now tempting to translate εὖχος in this line, again with LSJ⁹, ‘object of prayer, thing prayed for’. But this translation leaves aside the word’s associations with secular εὔχομαι as well as the most immediate associations of all, those with κῦδος and κλέος, which imply a meaning ‘fame, glory’. For # εὖχος ἐμοὶ δώσειν |, compare | κῦδος ἔδωκεν #, || δίδωθι δέ μοι κλέος, etc. as cited above, and for meaning and context compare: vs.

Does εὖχος in this line, then, mean ‘fame’, ‘vaunt’, or ‘object of prayer’? It means all three from a synchronic point of view, but the meanings ‘vaunt’ and ‘object of prayer’ are secondary and due to a relatively new formal association of εὖχος with εὔχομαι in both of its senses. Because of the phraseological conservatism of its poetic medium, εὖχος has not lost the synchronically incomprehensible sense ‘fame’ and has not been partitioned, under the influence of εὔχομαι, into a sense ‘vaunt’ and a sense ‘object of prayer’, as has the newer nominal derivative of εὐχ-, εὐχωλή. [14] But these two meanings are subliminally present here and perhaps in other attestations of εὖχος. [15]

{111|112} If ‘fame’ is the diachronically significant meaning of εὖχος, then how can we reconcile it to the hypothesis that the root εὐχ - originally meant ‘say (in a functionally marked context)’? In fact an archaic form εὖχος ‘fame’ makes good sense as a direct development from this hypothetical archaic meaning. Just as Latin fāma ‘rumor, fame’ and Greek φήμη ‘divine utterance; rumor; fame’ are derived from the verbal root *bheə ₂- ‘say’ (Gk. φημί; Lat. fārī, fās [legal]) [16] so εὐχ- ‘say’ produces εὖχος ‘fame’. Note that the two roots *bheə ₂- and *ə ₁ weg ^w h- show the parallelism in meaning and semantic development implicit in the Homeric and Mycenaean collocation of εὔχομαι and φημί. Furthermore, a symmetrical parallel to this sequence exists in another word identical with εὖχος in its formation and associated with it in epic diction: κλέ(ϝ)-ος ‘fame’ from the verbal root κλυ-/κλεϝ- ‘hear’. The notions ‘hear’ and ‘say’ are reciprocal, and they produce a semantically unitary derivative, ‘fame’.

It is plausible, then, to hypothesize that ‘say (in a functionally marked context)’ is the semantic antecedent of legal εὔχομαι, of secular εὔχομαι, and, finally, of εὖχος. The question remains, however, whether this is also the semantic antecedent of εὔχομαι in sacral contexts, as has been suggested {112|113} by Corlu. [17] Can this be supported formally and by typological parallels? What of the conceptual gap between ‘boast’ or ‘say’ and the complex, holy utterances designated by sacral εὔχομαι, in Homer?

Again, the formulaic repertoire of the Homeric texts provides formal evidence for exactly this contextual bridge. The most common expression with sacral εὔχομαι is a line in which a form of εὔχομαι occurs with reciprocating ἔκλυε (cp. secular εὖχος vs. κλέος) and parallel ἔφατ’ (cp. legal and secular εὔχομαι … φημί). The same collocation of εὔχομαι and κλύω is attested elsewhere in transformations. [18] Nor are typological parallels lacking. The development ‘say’ => ‘pray’ is a contextual specialization of the same order as ‘say’ => ‘boast, ‘speech’ => ‘fame’, etc. A word often connected with εὔχομαι in sacral contexts reveals the pattern again: ῥέζω (root *werg-/wreg- ‘do’), ‘do’ => ‘sacrifice’ (both meanings Homeric).

Finally, postulating ‘say (in a functionally marked context)’ as an antecedent of ‘pray’ gives an excellent account of the structure of Homeric prayers. In its pristine form, the Homeric prayer’s structure (invocation, grant of favor, request of favor) is essentially physical, formal communication. Only if our notion of such a prayer is conditioned by cultural prejudice as basically ‘vow’ or basically ‘request’ does the neutral and concrete sense of εὔχομαι ‘say a prayer’ escape us. This concreteness is echoed on the one hand by the concreteness of the god’s response: he ἔκλυε, ‘heard it’, says the formula which concludes prayers, not ‘accepted it’, or ‘agreed’ or ‘granted it’; and on the other hand by the significant pairing in ritual narrative contexts of εὔχομαι and σπένδω, ‘pour a libation’. To cite Marcel Mauss once more, “La prière est évidemment un rite oral." [19] Moreover, within the formulas of the epic, the concrete sense of εὔχομαι is still discernible, for the adverb μεγάλα, twice attested with sacral εὔχομαι in the formula occurs elsewhere with the following verbs: [20] Evidently, prayer is not only functionally but also physically marked speech in Homer.

On the evidence of the Homeric attestations, then, it is plausible to suppose that a pre-Homeric sense of εὔχομαι, ‘say (in a functionally marked context)’ has become specialized in three social spheres: 1) secular, or heroic, resulting in the specialization ‘say (proudly, accurately, contentiously)’, 2) legal, preserving the meaning ‘say’ or perhaps developing a special sense ‘state’, 3) sacral, with the specialization ‘pray’.

These three specializations had already become fixed in Indo-European and were inherited as such by the dialects. That is perhaps clear from a cursory review of the cognates (above, p. 9f.), but the survey which follows establishes this in detail. In addition it searches for contextual evidence and archaisms to test the hypothetical meaning that I have reconstructed for εὔχομαι. In fact, it repeats, with intentional similarities in method and surprising similarities in content, the process of synchronic analysis and internal reconstruction just concluded.

Vedic Sanskrit preserves a cognate of εὔχομαι in the verb óhate, [21] root ūh- [2], attested in intransitive (18 times) and transitive (11 times) usages. The transitive usage of a medium tantum verb, as are όhate and its cognates in Greek (only intransitive) and Avestan (only intransitive), must be a secondary development, and we shall return later to this usage. [21a] To begin, I offer a formal classification of all the intransitive usages of ūh- on the basis of their meters:

I. In octosyllables: II. In dodecasyllables (Jagatī): III. In hendecasyllables (Triṣṭubh):

In octosyllabic verse, the attestations show an inviolate restriction of present indicative singular forms to the last three syllables of the line. Now according to Nagy, [22] the Vedic octosyllable is divisible into a four-syllable opening and a four-syllable closing on the basis of a statistical contrast between the unfixed metrical pattern of the first four syllables (# ⏒ ⏒ ⏒ ⏒) and the fixed pattern of the closing syllables (⏑ – ⏑ ⏓ #). Accordingly he suggests that the Jagatī (dodecasyllabic) verse was originally composed of a complete octosyllable (opening + closing) following by an additional four-syllable octosyllabic closing, or, alternatively, of a four-syllable octosyllabic opening followed by a complete octosyllable (opening + closing). Nagy is able to support this hypothesis with statistical and phraseological evidence, and it appears that the attestations of όhate in II above agree with the theory as well. The four-syllable closing segment in the Jagatī verse 10.65.10: yá όhate # is also attested as the four-syllable closing of a Mahāpaṅkti octosyllable, 8.40.11: yá όhate #. This Jagatī verse may be a compound of the first type, while the other, 8.59.2, appears to be one of the second type, with octosyllabic opening (# yayόḥ śátrur: 4 syllables) + octosyllable … nákir ā́deva όhate #. No exact parallels to this octosyllable exist independently, but compare:

In any case, the fixity in form, meter, and metrical position of óhate/ohase/ohiṣe is disrupted in the Triṣṭubh verses. This occurrence can be interpreted in terms of a traditional Vedic fixation of óhate, etc. at the end of octosyllables and the metrical etymology of Triṣṭubh proposed by Watkins. [23] If, as he suggests, this hendecasyllabic verse is simply catalectic Jagatī, then a traditional dodecasyllabic close … óhate # could not adapt to the erosion of its final syllable without grammatical alteration: thus the perfect with present sense [24] ūhé # in 1.181.4 and 5.3.9. However, the participial forms, which appear only in Triṣṭubh verses and have different metrical values and positions, offer no basis for speculation as to their metrical history. {115|116}

If this tentative formal analysis brings to mind that the Hymns of the Rig-Veda, whatever the technique and circumstances of their composition and however alien they may have been to those of Homeric epic, have an archaic, traditional verse-form which is also capable of the preservation of archaic phraseology and theme, [25] then it has fulfilled its purpose. As a detail supporting this contention, we note that the grammatical restriction of intransitive ūh- to present indicative singular, “present” perfect, and tenseless participles is reminiscent of the grammatical restriction of secular εὔχομαι to the present indicative (singular or plural) and tenseless participles in Homer (as against four [of 84] instances of present tense sacral εὔχομαι). But the contextual correspondences between precisely this secular εΰχομαι and intransitive ūh- offer more dramatic testimony to the archaism and conservatism of both poetic media, Homeric and Vedic. In our analysis of secular εὔχομαι for Chapter II, we divided the attestations into four groups on formal grounds, as follows:

1. γένος εὔχομαι εἶναι and its transformations
2. εὔχομαι εἶναι + comparative/superlative
3. εὔχομαι εἶναι + social relationships
4. Secular εὔχομαι introducing, concluding, and reporting speech.

The attestations of intransitive ūh- in the Rig-Veda are faithful to this classification in detail, though they are not susceptible to such rigorous formulaic analysis. For εὔχομαι specifying γένος [26] compare this genealogy of the Aśvins: This passage corresponds in etiquette to the Greek γένος εὔχομαι εἶναι passages, in which a personage specifies the name of his father but omits {116|117} his own. Furthermore, the second genealogy, divás putrá ‘son of Dyaus’ (divás = Διός) is attested in Greek with εὔχομαι [28] : Greek πά(ϝ)ϊς and Sanskrit putráḥ are cognate (Frisk 1960 s.v. παῖς, Ernout-Meillet 1959 s.v. puer, Pokorny 1959: p. 843) and for possible thematic relationships between Paris, Hector, the Dioscuri, and their Vedic equivalents, the Aśvins, see D. Frame, The Myth of Return in Early Greek Epic (forthcoming).

The same collocation of putráḥ and ūhé exists in another hymn, this time with a metaphorical extension which does not occur for εὔχομαι in Homer except, perhaps, in the line just cited (N 54): [29] A priest is here speaking of himself as Agni’s son, qua seeker of his protection. [30] Two other Vedic attestations seem to preserve the genealogical context for óhate:

The first passage is difficult to understand. [31] It comes in a list of gods being invoked in a Hymn to all the gods, but the varying interpretations of it which have been offered consistently suggest genealogy. The second belongs in this classification if we appreciate the connotations of śátru- in {117|118} connection with the war-god Indra. It does not mean ‘überlegener Gegner’, as Geldner translates it, but rather ‘gleich starker, ebenbürtiger Gegner’ (Grassmann 1873: s.v.). Compare Indra’s other epithets pratimá̄na- ‘id.’ 6.18.12, 1.102.6; ájāta-śatru- ‘dem kein Gegner gewachsen ist’, 5.34.1, 8.93.16; cf. also 1.176.1, 10.28.6, 10.54.2, where Indra’s birth and strength against his enemies are correlated. This attestation belongs to the same context of genealogical comparison in battle confrontations as do several attestations of γένος εὔχομαι εἶναι in Homer (see above, pp. 73 f., 93). Further attestations of ūh- in relation to Indra’s arch-enemy, the Vṛtra, will be discussed below.

The second category in our previous analysis of secular εὔχομαι was εὔχομαι εἶναι + comparative/superlative (above, pp. 79ff.). Each of these constructions are attested once in the Rig-Veda: In the first, the poet addresses a pregnant compliment to his patron. The second refers to Indra. The syntactic parallelism between these attestations and such Homeric lines as cannot be coincidental. From a semantic point of view, it suggests that the contentiousness and pride associated with secular εὔχομαι in Greek were associated with its ancestor in Indo-European (as well as Indic). On the other hand the absence of a form of the verb ‘to be’ in the Vedic attestations, [33] coupled with the unqualified comparative in 8.5.39 [34] suggests that the accuracy and objectivity associated with Greek εὔχομαι has weakened slightly in favor of more emotive and subjective connotations, or, alternatively, that the Greek poetic tradition developed and strengthened the word’s objective connotations. Actually both processes may have occurred. {118|119}

Precise and detailed correspondences also exist in Vedic to the third use of secular εὔχομαι, for specifying as marks of privilege and identity social relationships which range from the possession of hospitality rights to the hieratic profession of θυοσκόος: [35] A further parallel to Homeric θυοσκόος εὔχεαι εἶναι (χ 321) exists in the thrice attested collocation sūrí-/-óhate. Two instances have already been cited, 1.181.4, where the mortal Aśvin is called sūrír and 8.5.39, concerning the poet’s generous sūrír. In the third, Indra is being invoked to crush an enemy, a man who does not extract Soma: The bard is aiming his hostility at a niggardly patron, for the word sūrí- is elsewhere applied either to gods [37] or men in their role as wealth-giving patrons of Vedic hymn and ritual. [38] Its semantic sphere is consistent with ṛ́ṣi- and θυοσκόος, a word attested alongside μάντις and ἱερεύς at Ω 221. Finally, there are these verses, again addressed to Indra: {119|120} An exact formal parallel to the expression pátir … ohase. exists in one of the attestations of εὔχομαι, specifying a social function. Hector, in the prayer to his horses, [39] speaks of himself in relation to Andromache: Vedic páti- and Greek πόσις are exact formal cognates, [40] but their functions in these attestations diverge: πόσις only survives in the sense ‘lord, husband’ in Greek, and it is applied in Homer to Zeus (H 411 et passim) and an assortment of heroes whose spouses, goddesses and heroines, are always specified. [41] Vedic páti- can also have this sense, [42] but here, as often, it is a hieratic term, restricted to divinities, with the sense ‘lord, master’. However, the social connotations of their common root *pot- are well established for Indo-European, [43] particularly Indo-Iranian, so that the hieratic applications of Vedic páti- are to be understood as an instance of the archaic use of prestige social words and concepts for divinities, such as Ζεῦ πάτερ/dyáuḥ pitár-/Juppiter, etc. or ἄναξ Διὸς υἱὸς Ἀπόλλων, Ζεῦ ἄνα. The situation in Greek represents a later development. Though the specialization of *potis to ‘lord, husband’ must be old, the broader social and hieratic sense ‘lord, master’ is older still. In Greek only the former has survived (contrast the less restricted and hieratic use of the feminine form πότνια ‘lady, mistress’ in Homer and Mycenaean [44] while the older meaning ‘lord, {120|121} master’ has been displaced by such words as ἄναξ. [45] Given the survival of πόσις only in the sense ‘lord, husband’, it remains possible that Homeric πόσις εὔχομαι/Vedic pátir … ohase reflect the combination of these words in the poetic language of Indo-European, and that the traditional phrase was preserved in the Greek epic tradition even though its meaning had altered. We are, then, postulating the preservation—only in connection with Hector—of a phrase which might once have been hieratic, and this can be paralleled by other still hieratic expressions we have found associated with him. [46] Note that the prosody of πόσις εὔχομαι, which is the same as that of γένος εὔχομαι and κλέος ἄφθιτον, is compatible with its Indo-European origin in terms of Nagy’s metrical hypotheses. [47]

The last category in our analysis of secular εὔχομαι was its function to introduce, conclude, and report the warrior’s speech of triumph. [48] In the Rig-Veda there is one attestation of intransitive ūh- to report such a speech. It is the war-god Indra announcing his triumph over the dragon of drought, Śuṣṇa, and his brood: {121|122} Moreover the participial form of ūh- is twice applied to Indra’s chief enemy, the Vṛtra, though without any syntactic complement: To introduce contrasting claims about victory and genealogy, Homeric heroes constantly use the verb εὔχομαι, as we have seen. Just such claims, we can suppose, did the Vṛtra utter before his slaughter by the heroic Indra.

Except for one further example, which occurs in an altogether different context, this completes our survey of the attestations of intransitive ūh-. We can summarize its results by saying that Vedic intransitive ūh- categorically and in detail corresponds to Homeric secular εὔχομαι. Although a tertium comparationis for some of the usage categories will be lacking when our survey of other cognates of εὔχομαι is concluded, it is fair to say even at this point that the correspondences between these two words in context, syntax, and also accompanying phraseology are too close and complete to be a coincidence or the result of common innovation. The whole series of usages of secular εὔχομαι = Vedic intransitive ūh- apparently belongs to the Indo-European poetic tradition, of which Greek epic and the lyrical Vedic hymns are both faithful conservators. On a semantic level, this implies that the meaning we deduced from the attestations of secular εὔχομαι applies to intransitive ūh- as well: ‘say (proudly, contentiously, accurately)’, though the final adverb should be softened for the Vedic word.

Let us postpone for a moment our consideration of the last attestation of Vedic intransitive ūh- and test these conclusions by an overview of the contexts and meanings of other cognates of εὔχομαι. In this connection, two of the three [49] attestations of the athematic verb aoj- (= εὐχ-) in the oldest Avestan texts, the Gāthās or hymns of Zarathuštra himself, merit detailed attention. At the conclusion of Yasna 50, Zarathuštra describes his relationship to Ahura Mazdā in the following way: {122|123} In grammatical form and usage we rejoin in this passage secular εὔχομαι εἷναι and Vedic intransitive ūh-. The verb aojāi (1st sing. pres. conj.) occurs with a predicate nominative və̄ staotā “eure Lobsänger” and is coordinate with a finite form of the verb ‘to be’, aŋhā (1st sing. pres. conj.) “ich will es sein., which is cognate with εἰμί [51] ). Compare: This inheritance of a secular usage for *ə ₁ ewg ^w h- is as expected. [52] Again, in Yasna 43, Zarathuštra meditates before the fire-altar, Ahura Mazdā’s mystical dwelling. He sees himself encounter, for the first time, Ahura Mazdā himself, who greets and questions him: The first two questions which Ahura Mazdā asks Zarathuštra are ciš ahī kahyā ahī “Wer bist du? Wem gehörst du an?” As Geldner [54] long ago pointed out, this combination of questions has precise etymological correspondences in both Greek and Indic epic: (In Ο 247 Hector is asking Apollo his identity, in Ω 387 Priam is asking Hermes’, and in α 170 Telemachus is asking Athena/Mentes’s.) The Indo-European pedigree of this anaphoric combination of questions has thus been established. But the correspondence between the answers to these questions by a man to a god and vice versa in Greek and Avestan has not been noticed before. Zarathuštra answers Ahura Mazdā as follows: And Athena/Mentes thus answers Telemachus: We have, then, a piece of Indo-European poetic conversation, a question and its answer. [59] While a precisely parallel answer or naming attestation [60] of Vedic intransitive ūh- does not actually occur, the comparison just made confirms in general the Indo-European archaism of secular εὔχομαι and Vedic intransitive ūh -, and it strengthens the possibility that specific usages can be inherited phenomena rather than the result of common innovation. Such parallelism between form, function, and formal and functional context, cannot be considered a coincidence, especially in view of other evidence to be given for parallel phraseological inheritances by εὔχομαι and its cognates.

Younger Avestan aoxta (= εὖκτο) is consistent with Gathic aogədā. For example, the verb occurs at Y. 41.5 with a predicate nominative in a phrase parallel to Gathic staotā aojāi … aŋhācā. It is frequently used to {124|125] introduce the direct speech of Zarathuštra to Ahura Mazdā, as at Y. 9.16 āat̰ aoxta zaraθuštrō ‘also sprach Zarathuštra’, where it introduces a prayer, and in secular, heroic contexts as well. [61] Avestan aoj- is in fact the only cognate of εὔχομαι, whose speech connotations are overt enough to be recognized lexicographically. Thus Bartholomae’s definition praedicāre, which coincides closely with our hypothesis that the original meaning of εὔχομαι was ‘say’ in a functionally marked context. In Avestan the semantic derivation we are proposing is transparent and needs no proof.

If the Indo-European origin of the secular usage of εὔχομαι is confirmed by Indic and Avestan evidence, these languages offer no testimony to support or deny the specialization of εὔχομαι in legal contexts at so early a date. There exists, however, one piece of evidence that this usage was inherited in Homeric and Mycenaean. The Latin cognate of εὔχομαι, voveō, occurs in the phrases vōtī reus and vōtī damnātus. [62] These are without doubt archaic legal formulas which have been reinterpreted as sacral expressions. [63]

The one remaining attestation of Vedic intransitive ūh-, and the most important one for our purposes, is as follows: Geldner’s translation of ohiṣe in this passage as a transitive verb with the meaning ‘empfehle’/‘praise, commend’ is initially dissatisfying. To our {125|126} way of thinking, at least, messengers do not so much ‘commend’ their speeches as speak them. There is evidence within the Rig-Veda itself to sustain this objection. Comparison to a messenger is conventional elsewhere in the hymns for the most powerful of all speech, the hymn (vā́c-·, stóma-, etc.) itself, as, for example, [65] To reconcile the persistent speech associations of this simile with our passage is not difficult. Formally, it is just as possible that ohiṣe is an intransitive verb and that vā́cam is its ‘cognate’ accusative, i. e. that the syntax of 8.5.3c is a figura etymologica instead of transitive verb + direct object. Given that Vedic ūh- as an intransitive verb has the meaning of secular εὔχομαι, ‘say (proudly, contentiously, accurately)’, we can translate as follows: As such, the line embodies a transference of the messenger simile from the hymn to its poet, and there is a close parallel to this transference elsewhere in the Rig-Veda: Here again, the hymn (vā́cam) is like a messenger (dūtám iva), but the poet also sends (iṣya) it, as though, paradoxically, he were a messenger and the hymn his message. This, precisely, is the image in 8.5.3c if it is construed as a figura etymologica.

Actually, this way of construing 8.5.3c has already been suggested by Benveniste on the basis of diachronic evidence. [66] He has found a parallel to the line in Avestan. Here are the two passages, Indic and Iranian, with his translations: {126|127} Y. Av. aoxta is a cognate of εὔχομαι and Vedic ūh-, while vā́cam and vācim are also cognates. [67] The grammatical construction in Yt. 13.90 is doubtless a figura etymologica, since aoxta is an intransitive verb only. The grammatical construction in the Vedic line should be the same. That is to say, it was probably inherited as the same, since the attestation of these cognate phrases in both Indic and Iranian poetic texts constitutes evidence for a hypothesis that their prototype existed in the common Indo-Iranian poetic tradition. [68] The grammatical construction of such a prototype could only have been a figura etymologica, since transitive ūh- is a secondary development proper to Vedic alone. At this point, then, there is both synchronic and diachronic evidence to show that ohiṣe is still an intransitive verb in R. V. 8.5.3c.

We can carry the historical analysis of this common Indo-Iranian etymological figure a step further, however, and by doing so enhance our understanding of the etymology and meaning of Homeric εὔχομαι. For Homeric Greek also attests one and only one figura etymologica with the intransitive verb εὔχομαι in the formula which occurs seven times (Λ 379, Ν 619, Ρ 537, Φ 183, Ξ 500, Υ 424, and Κ461). [69] The accusative noun with εὐχόμενος is none other than (ϝ)έπος (<*əwek ^w-), functionally equivalent and related to Vedic vā́cam and Avestan vācim (<*əwοk ^w-). [70] However, the exact cognates of Greek (ϝ)έπος ‘word’ are Vedic vácaḥ, Avestan vacah- ‘id.’, while the formal but not functional equivalent of Vedic vā́cam, Avestan vācim, is Homeric (ϝ)όπ- ‘voice’. [71] The formal difference is slight, and the comparison of these three etymological figures, as evidence for the existence of their prototype in the poetic tradition of Indo-European [72] is prima facie justifiable: {127|128}Yet it is tempting to ask the reason for this variation between reflexes of *əwek ^w- and reflexes of *əwok ^w-. It is a fact which supports our comparison that two other Indo-European poetic phrases have been found which feature the comparison of Homeric (ϝ)έπος and the Indo-Iranian reflexes of *əwek ^w- and *əwok ^w-. Here are the texts cited in a comparison made by Bartholomae: [73] Here are those in Darmesteter's comparison: [74] It is clear from the attestations of svādmā́naṃ vācáḥ/vácaḥ svādóḥ and vā́cam … atakṣiṣuḥ/vácā ṃsi … takṣam that vā́c- (<*əwok ^w-) and vácaḥ (<*əwek ^w-) are functional alternates in Indic. I cite here the remarks of Meillet on the meaning of vā́c- (n. sing, vā́k): [75] We can use the Vedic line 8.5.3c, which is the focus of our discussion, as an example of this. It is a variation on lines which consistently compare the poetic vā́cam, the sacral hymn itself, to a messenger. Although it transfers {128|129} this simile to the poet, the line directly attests to the religious value of vā́c- (= hymn) and, by latent allusion to the other application of its simile, it calls to mind the animate force of vā́c-: We can now see that this simile, which may appear banal enough to us, is instead a bold instance of the poet identifying himself with his hymn’s affective power.

To transpose Meillet’s analysis of vā́c- into linguistic terminology, we can say that it is the functionally marked word for ‘speech’. By contrast, the neuter noun vácaḥ (= ἔπος) is the functionally unmarked alternative to vā́c- (feminine). vácaḥ, like ἔπος, signifies speech either with or without the animacy and affectivity characteristic of vā́c-. [76] Now the Greek cognate of vā́c-, ϝόπ-, has lost its affectivity and come to mean simply ‘voice’, with the root *əwok ^w- retaining its original, Indo-European sense, as Meillet points out, [77] only in the ‘personified’, feminine Homeric noun ὄσσα (<*əwok ^w -ya). [78] On the basis of Greek and Indic evidence, then, we can hypothesize that the relationship between *əwek ^w-, functionally unmarked, and *əwok ^w-, functionally marked speech, once belonged to Indo-European. [79] Given the semantic change undergone by Greek ϝόπ- ‘voice’, it is understandable that from Indo-European poetic phrases consisting of *ə ₁ ewg ^w h- (> Gk. εὔχομαι, Ved. ūh-, Av. aoj-) or *sweə ₂ du- (> Gk. ἡδυ-, Ved. svād-, Av. hud-) or *tek- (> Gk. τεκ-, Ved. takṣ-, Av. taš-) plus the functional {129|130} alternates *əwek ^w- and *əwok ^w-, Greek has only retained those with ἔπος ( *əwek ^w-), whose meaning has survived intact in Greek as in Indic and Iranian.

A final piece of evidence strengthens the formal reconstruction of our Indo-European poetic phrase as *əwek ^w- or *əwok ^w- plus *ə ₁ ewg ^w h-: Here vácas (<*əwek ^w-) is the direct object of ohase, [80] but in spite of the functional innovation in this transitive usage of ohase, the collocation of vácas and ohase is formally equivalent to Homeric εὐχόμενος ἔπος. If this collocation were not actually attested in Vedic, it would have been reasonable to hypothesize its existence on the basis of comparative evidence from other Indo-European poetic phrases.

Having compared the form of the three etymological figures, we are ready to consider their contexts and functions. What can we reconstruct as the meaning and usage sphere of their Indo-European prototype? Does this agree with our earlier hypothesis that εὔχομαι was the functionally marked word for ‘say’ which was specialized, in pre-Homeric times, in sacral, secular, and legal contexts ?

For our translation of ohise in the line: we invoked the fact that intransitive ūh- had inherited and preserved the same meaning as secular εὔχομαι, ‘say (proudly, accurately, contentiously)’. It is important that this synchronic meaning provides a satisfactory sense in context, but this should not obscure us to another fact, that the context of ohiṣe in 8.5.3c has nothing to do with any other contexts of intransitive ūh -. It specifies neither genealogy nor superiority nor social function nor victory in battle. Instead, the context of ohiṣe recalls us to the beginning of our study, to the use of sacral εὔχομαι to introduce or conclude speech by a man to a god: [81]

1. Sobald die rötlichscheinende (Uṣas) von fern her, als ob sie hier wäre, aufschimmert, so hat sie allenthalben ihren Glanz ausgebreitet.
2. Auf dem männerfahrenden, gedankengeschirrten, breitgestaltigen Wagen folget ihr Meister Aśvin der Uṣas.
3. Euch, ihr Gewinnreichen, haben sich die Loblieder vorgestellt. Wie ein Abgesandter, usw.

The relationship between artistic and religious values in the Rig-Veda is the subject of some scholarly disagreement, but not to the extent of making it doctrinaire to say that the hymns are fundamentally religious. [82] As its messenger simile makes clear, then, vā́caṃ dūtó yáthohiṣe refers to the utterance of sacral speech as formal, affective communication by a man with a god. They are hymns evolved from prayers, meant to accompany sacrifices, containing invocations to divinities and making requests of them. I reiterate that this context for ohiṣe is not predictable from the other contexts of intransitive ūh- in the Rig-Veda.

In Younger Avestan, the context of vācim aoxta is entirely comparable to that of R. V. 8.5.3c. Here is a fuller citation: The antecedent of yō and subject of aoxta is Zarathuštra himself, and vācim refers to the Gāthās, his prayer hymns. [84] The vācim which Zarathuštra {131|132} aoxta ‘aufsagte’ contrasts with the miθaoxtəm vācim ‘falschgesprochene Wort’ mentioned elsewhere in Younger Avestan, as Y. 60.5: vainīt̰ … aršuxdδō vāxš miθaoxtəm vācim “überwinden soll … das wahrgesprochene Wort das falschgesprochene Wort." [85] The latter are the liturgical hymns of inimical religious sects, presumably resembling the Vedic hymns in style. [86]

In both Indic and Iranian attestations, then, the collocation applies to sacral language. But it is also clear that the expression of sacral language is presented as a matter of claims and counter claims: vācim . . . vīdōyum ahurō. t̰kaēšəm “das gegen die Daēva’s gerichtete, Ahura’s Lehre enthaltende (heilige) Wort." Note also that forms of the root aoj- are applicable to both acceptable (vācim aoxta) and unacceptable (miθaoxtəm vācim) sacral language, just as Indra and his arch-enemy, the Vṛtra, both óhate. [87] Most likely the same nuance is conjured up by ohiṣe in R. V. 8.5.3c, in view of the connotations of contentiousness carried by the other uses of intransitive ūh- and the well-documented “joute littéraire., as Renou describes it, [88] in which the Vedic poets participated, and which is signalized by vā́c-. Prayer and ritual are decidedly agonistic in Indic and Iranian, a characteristic which can be paralleled for institutions based on reciprocity in other archaic cultures, and which should probably not be considered a recent, isolated, or ‘secularizing’ tendency in these. [89]

{132|133} As we have already seen, the Homeric parallel to these Indic and Iranian phrases, εὐχόμενος ἔπος, is attested six out of seven times to introduce a speech of triumph by a warrior over his dead enemy. [90] These speeches are companion pieces to the genealogical statements introduced by εὔχομαι which precede single combat, and both are examples of secular εὔχομαι in the meaning ‘say (proudly, contentiously, accurately).’ The seventh attestation is a follows: Here εὐχόμενος ἔπος occurs in a line with two finite verbs (ἀνέσχεθε … ηὔδα), as is typical of the language of ritual narrative. [91] It introduces a prayer of the regular tripartite structure: 1. invocation (χαῖρε, θεά) 2. grant of favor (σέ . . . ἐπιδώσομεθ’) [92] 3. request of favor (ἀλλὰ καὶ {133|134} αὖτις . . . εὐνάς). I conclude that εὐχόμενος ἔπος in Κ 461 means ‘speaking the word (of prayer)’, as do its Indo-European cognates vācim, aoxta “das (heilige) Wort aufsagte” and vā́cam . . . ohiṣe ‘I speak the word (of the hymn)’. In previous discussion of Κ 461 [93] we left open the possibilities that its breaking of the sacral-secular split in the formulas of εὔχομαι, was an archaism predating the rigorous implementation of that split or a slip of the tongue. But this line constitutes the only one of five instances of cross-over between the sacral and secular formulas of εὔχομαι which has neither a textual variant nor a simple explanation of the presence of εὔχομαι as filling a variable slot in an otherwise fixed expression. [94] In view of its textual security and the simple explanation for its presence offered by comparative evidence, I view this attestation of εὐχόμενος ἔπος as a contextual and formal archaism.

This establishes the viability of our hypothesis that the etymological figure *əwek ^w-/*əwok ^w- plus *ə ₁ ewg ^w h- belonged to the Indo-European poetic tradition. The formal collocation and its function are preserved in poetic texts of three Indo-European dialects. What does this tell us about the meaning and usage of the prototype of εὔχομαι in Indo-European? First of all, it provides dramatic evidence that the root originally meant ‘say’. In an inherited phrase attested in three dialects its reflexes can successfully be translated as intransitive verbs of saying (ohiṣe ‘I speak’, aoxta ‘er sagte auf’, εὐχόμενος ‘saying, speaking’). Secondly, we have evidence that the usage of the root *ə ₁ ewg ^w h- in sacral contexts is at least as old as its usage in secular contexts, for it is attested in a fixed phrase in sacral contexts in three Indo-European dialects. [95] Finally, the contexts of the Indic and {134|135} Iranian cognates of εὐχόμενος ἔπος add a new dimension to our definition of *ə ₁ ewg ^w h- as the functionally marked word for ‘say’ in Indo-European. The overtly agonistic quality of sacral speech in these cultures in general and in the contexts of their reflexes of the Indo-European poetic phrase *əwok ^w- *ə ₁ ewg ^w h- may be a purely local phenomenon. But it gives life to the possibility that prayer was agonistic in its Indo-European setting as well, that speech, in order to communicate with divinity, had to win out over the speech of competitors. [96] Moreover, if we specify that functionally marked ‘say’ in Indo-European implied the utterance of words which win out over a competitor’s, then we can understand in a more concrete way the reasons behind the diverse acceptations of Homeric εὔχομαι. Then its use for legal claim and counter-claim, on the field of battle, in specifying one’s place in a prestige-conscious society, and for the rite oral as well, makes good sense.

We are in a position to provide further evidence to confirm the second of these conclusions, that the sacral as well as secular usages of εὔχομαι were inherited by the Greek epic tradition. Other cognates of εὔχομαι in Indic and in Latin have sacral usages as well, and meanings which are {135|136} compatible with the etymology we have proposed. The words we are about to discuss are derivatives of an Indo-European transformation of the root of εὔχομαι, Ved. óhate and Av. aogədā. In Benveniste’s terms, [97] the root of εὔχομαι is Theme I: *ə ₁ ewg ^w h-. The root of Latin voveō and Vedic vāghát- is Theme II: *ə ₁ ewg ^w h-.

Dictionaries define vāghát- (<*ə ₁ ewog ^w h-) [98] as a cult official with unspecified functions: ‘Beter, Opferer; Veranstalter des Opfers’. [99] But in several passages the word clearly refers to singers of the Vedic hymns, and Geldner’s translations of it as ‘Sänger, Priester’, which Citron [100] ably defends, are justifiable: It is the poet who traditionally expects material reward from his patron, to whom he gives in return ákṣiti śrávah “unvergänglichen Ruhm." [101] The first person plural yajāmahe ‘wir erbitten’ in the following stanza makes plain the reference of vāgháte. Plainer still is the following: Here Agni is conceived of as the winner of the wealth which the poet’s patron bestows. Other attestations support the definition ‘Sänger’. [102] And it is worthy of special note that the vāghát- is twice attested as complementary to the hótṛ (< juhóti ‘pour (a libation)’, = Gk. χέ(ϝ)ω), [103] who is the most important official in Vedic rites: {136|137} Perhaps these passages attest to an Indic reflex of the pairing of σπένδω and εὔχομαι in ritual contexts of Greek epic. [105] Citron has already suggested such an inheritance on the basis of the functions and etymologies of hótṛ and vāghát-, and he explains the prominence in Vedic of hótṛ at the expense of vāghát- as due to the great importance of the Soma offering in Vedic ritual. [106] In any case, there is good evidence to show that vāghát- refers to a human speaker of sacral language to divinities, to the man who vā́cam óhate.

Likewise the Latin derivative of Theme II *ə ₁ weg ^w h-, voveō, which features iterative-causative morphology [107] and is a transitive verb, occurs exclusively in sacral contexts, with a legal usage attested vestigially. [108] This word has, however, undergone semantic development to ‘pledge, vow’ or ‘wish’. As was previously suggested, [109] we can understand this meaning for a cognate of εὔχομαι by the parallel but relatively transparent semantic development of Latin spondeō, cognate of σπένδω. spondeō, which is morphologically identical to voveō, means ‘pledge, guarantee’ and is restricted to legal contexts. Yet it is clear from Indo-European evidence that its root did not originally refer to pledges, but to a concrete, sacral act: cf. σπένδω ‘pour a libation’, Hitt. šipand- ‘spenden, Gußopfer darbringen; {137|138} opfern’. [110] As Benveniste has shown, [111] in their Homeric contexts such libations are intended to guarantee security in the face of danger, e. g. when Achilles dispatches Patroclus into battle (Π 225–256) or when Priam sets out to ransom Hector (Ω 302–316). This explains the use of σπένδομαι for treaties, which are mutual guarantees of security. We can say that in Latin, the word spondeō has developed from concrete to abstract, from sacral act with symbolic content to the symbolic content of the sacral act: ‘pledge, guarantee’. The same development occurs later within Greek, for the compound ἐπισπένδω is attested in the Gortynian Code with a meaning ‘guarantee’, [112] and in a recently published 5th century Cretan inscription the simplex form σπένδω means ‘solemnly pledge, promise’. [113]

On this basis we can hypothesize that voveō ‘pledge, wish’ was a word originally designating a sacral act whose symbolic content was a pledge and a wish. But this is precisely the case with Homeric prayers designated by the cognate of voveō, εὔχομαι. They contain a statement that a gift has been, will be, or is being given to the god, and this gift, which can even be a libation, symbolizes the existence of a contract between god and man. [114] On the basis of its existence the person praying expresses a wish or request, often for his own security in the face of danger: To this passage and others [115] we may compare a Plautine attestation of voveō. Jupiter himself is speaking, disguised as Amphitruo, to Alcumena: vōta here refers to sacral pledges made in battle whose fulfillment was contingent upon the fulfillment of a wish or request. The structural elements are the same as those in a Homeric prayer, but they have been arranged in a tightened temporal hierarchy, and voveō ‘pledge’ has become specialized {138|139} and also abstract, in that it no longer refers overtly to speech. [116] Specialization to the other element of prayer, the wish or request, also occurs for voveō and vōtum: In this usage voveō has almost returned to the meaning ‘pray’, but its specialization to the wish or request element of prayer is clear from numerous attestations, [117] and it is not to be confused with the sacral usage of εὔχομαι, in Homer, where the verb refers to the totality of a prayer’s structural elements. [118] Nevertheless, the existence of such a meaning may have perpetuated the existence of archaic sacral phraseology to which the old, concrete sense of voveō was appropriate. A possible example is the collocation vōtīs vocāre, attested in sacral contexts no less than seven times [119] in the corpus of Virgil, as in the concluding line of a prayer to the deified Caesar Augustus in the Georgics (I. 40ff.): This fixed phrase may be the reflex in Latin of the Indo-European etymological figure *əwek ^w -/*əwok ^w- + *ə ₁ ewg ^w h-, for vocāre is cognate with Ved. vā́c-, Av. vaxš, and voveō with εὔχομαι, Ved. ūh-, Av. aoj-.

To sum up, our survey has demonstrated the Indo-European origin of the three contextual specializations of εὔχομαι. For the sacral usage of its root we can compare εὐχόμενος ἔπος (Κ 461), vācim aoxta, vā́cam … ohiṣe, as well as Ved. vāghát-, Lat. voveō and some attestations of Avestan aoj-; for the secular usage, Vedic intransitive ūh- and other attestations of Avestan aoj-; for the legal usage, Latin vōtī reus and vōtī damnātus. In the process of this survey, we have repeatedly encountered evidence of continuity between the traditional, formulaic diction of Homer and Indo-European poetic diction. The result in terms of semantic reconstruction is confirmation of our internally motivated hypothesis that the root of εὔχομαι originally meant ‘say (in a functionally marked context)’. This is compatible with the meaning or semantic developments of all its cognates, {139|140} and we have been able to suggest more specifically that functionally marked ‘say’ in Indo-European implied ‘say to win out over the speech of a competitor’. More abstractly, we take our cue from Meillet, and describe the ancestor of εὔχομαι as the mot-force [120] for ‘speak’, ‘say’. It referred to words not as signs distinct from deeds, but as forces with immanent power in sacral, legal, or heroic and aristocratic settings. [121]

Of the four languages which preserve this archaic word, only Homeric Greek has preserved all three of its original usages intact, and it has done so even though at least two of these usages, which we have called sacral and secular, had lost their original semantic unity and become, in effect but not in fact, separate words. In the analysis of εὖχος, we saw that a single formulaic phrase can evoke a range of disparate meanings. The triple usage of εὔχομαι is to be understood as another dramatic instance of the power a rigorous formulaic medium has to implement and preserve semantic richness. Formulas are not cliché́s, receptacles of cant, or merely convenient phrases to help a faltering performer. They are metrical combinations of words in which the heritage of the primordial past could achieve its highest potential for the expression of living poetic meaning. They are the means by which that past survived, and the medium in which it first crystallized. It is time for Homeric research to realize the creative potential in diachronic, synchronic, and aesthetic understanding of them.