THE e-PERFECT IN HITTITE

E. H. STURTEVANT

YALE UNIVERSITY

The Hittite documents show many instances of an interchange of the vowels e and a. Since the variation is not uniform, and neither vowel of the pair is restricted to any phonetic surroundings or morphological categories, we must apparently assume several causes, and Sound method requires the separate treatment of groups of words which show parallel phenomena. In this paper I propose to discuss the variation in monosyllabic verbal roots which end in a consonant. That the matter needs elucidation appears from these typical examples:

ešzi 'he is'

ešir 'they were'

ašanzi 'they are'

ari 'he has arrived'

erir 'they had arrived' aranzi 'they have arrived' Hrozný1 discussed the variation in the radical syllable of verbs, but without coming to any satisfactory conclusion. Forrer2 groups toget her paradigms of a number of verbs which exhibit the phenomenon, under the captions: 'Umlaut von Verben der mi-Konjugation' and Umlaut von Verben der hi-Konjugation'. Tenner3 supplements the forms given by Forrer, and remarks quite justly: 'Mit dem nur vor i eintretenden germanischen Umlaut hat dieser hethitische Vokalwechsel jedenfalls nichts zu tun'.

Tenner confines his attention to three verbs of the hi-conjugation, and precisely these verbs, I think, suggest the true source of the variation in most of the words discussed by Hrozný and Forrer. The verbs are citable as follows:

1 Die Sprache der Hethiter 169, 170f. (1917).

2

3

Zeitschrift der Deutschen Morgenländischen Gesellschaft NF 1.213f. (1922). Ein Hethitischer Annalentext des Königs Mursilis II 18f. (Leipzig, 1926).

With the exception of the forms printed in italics, these verbs show the vowel a in the singular, the third person plural present and imperative, and in the participle, while the vowel e appears throughout the plural, except the third person present and imperative. The few forms which vary from this scheme are of course to be explained by analogy.

=

I have pointed out that some of the personal endings of the Hittite hi-conjugation are remarkably similar to those of the Indo-European perfect tense (2nd sing. ti tha, 3rd sing. i = e, 3rd pl. pret. ir beside Skt. ur, Lat. ĕre, s frequently inserted before second personal endings as in Lat. vidisti, vidistis, etc.). If, then, the hi-conjugation is in part identified with the Indo-European perfect, the stem-vowel e in the plural of our three verbs is to be equated with Germanic plurals like Gothic bērum 'we bore' beside bar 'he bore'. The fact that there is usually no indication of length in the e-forms cited above does not constitute an objection. If the original vowel quantities survived in Hittite, at any rate they cannot be inferred from the orthography of our documents; e-eš-zi =tori 'he is' and e-šá-riñora 'he sits' are typical examples.

The correctness of our equation becomes obvious upon a closer scrutiny of the three verbs. Other Hittite words clearly related with

LANGUAGE 2. 33f. (1926).

šakk- 'know' are šagaiš 'omen's, šakiya- 'give an omen", and šakuwa 'eyes'. It follows that the original meaning of the word was 'see', and so we have a preterito-present like Greek olda, Skt. veda, Gothic wait 'I know' beside Lat. vīdī 'I have seen'. We must therefore connect Hittite šakk- with Gothic saihan, Anglo-Saxon séon 'see', from IE *segu-8.

=

Hittite šakuwa 'eyes' preserves the labialization of the qu, as do kuiš Lat. quis 'who', and kuen- 'strike': Gk. Oeivw 'strike', óvos 'murder'. The inconsistent representation of the sound-group kw is due to the limitations of the cuneiform writing. Since there was only one available sign containing the consonant w, namely wa, the groups we, wi, etc., had to be written otherwise. Hence we find ku-en-zi for kwenzi 'he strikes', ku-iš for kwiš 'who', etc. On the other hand it was impossible to write two consonants together at the beginning of a word, and so we find ku-wa-áš-ki-nu-un for kwaškinun 'I frequently struck' (preterit of the iterative stem from kwen- 'strike'), ku-wa-at for kwat 'why', and ku-wa-pi for kwapi 'when', 'where'. To be sure, it would have been easy to write the group kwa in the interior of a word (*šá-ak-wa); but a clumsy orthography which is sometimes necessary is frequently used where it could be avoided. For šanhzi 'petit' it was necessary to write either šá-an-ah-zi or šá-an-ha-zi; hence beside the accurate šá-an-hu-un 'petii' we find also šá-an-ah-hu-un'. An extra vowel had to be written in kar-ap-zi for karpzi 'he musters', but not in kar-ap-an-zi for karpanzi 'they muster'. We should therefore write the Hittite word for eyes šakwa, and derive it from PreIndo-European **soqua. Quite possibly it is the same as IE *ogu(Greek ooσe, Church Slavonic oko-, etc.) 'eye' with an initial s from contamination with *sequ- 'see'.

The labialization was, no doubt, regularly lost before consonants as in Latin. Hence we have the verb-forms: šakti, šekteni, šakta, šekta, šekten, šakdu (cf. also šak, šekkweni, šekkwen). Analogy carried the simple k through the rest of the paradigm and into the derivatives, šagaiš 'omen', and šakiya- 'give an omen'.

* See Weidner, Archiv für Keilschriftforschung 1. 10 (1923); Sommer and Ehelolf, BoSt. 10. 99.

• Friedrich, Zeitschrift für Assyriologie NF 3. 198 and fn. 7 (1926).

7 Friedrich, Staatsverträge des Hattireiches in Hethitischer Sprache 35f. (1926). This is Friedrich's transcription of the word; a better transcription is suggested below.

For further connections in IE, see Walde-Pokorny, Vergleichendes Wörterbuch der Indogermanischen Sprachen 2. 477–80 (1927).

• Sommer, Boghazköi-Studien 7. 45.

The preterit of Goth. saihan is sah 'he saw', sehun 'they saw', the stem-vowel of the singular representing IE o, and that of the plural IE ē. It follows that in our Hittite verb also šakki corresponds to IE *soque, while šekkir is equivalent to IE *sēqure, *sequer, or the like. We have, then, fairly clear evidence of qualitative ablaut in Hittite. This harmonizes with my suggestion10 that the Hittite ablative ending ts (az, za, etc.) is the nil-grade of the suffix *tos, which forms ablatival adverbs in the IE languages; we may now conclude that Hittite did not break away from the parent stock until after the ablaut changes. The verb ak- ‘die' does not usually have the meaning of the IE perfect. Examples from the law code are these:

§72 (Hrozný): ták-ku GUD-áš A.ŠAGHIA -ni ku-el-qa a-ki BE. EL A. ŠAG II GUD pa-a-i, 'If an ox dies in anyone's field, the owner of the field gives two oxen'. §197 (Hr.): (6) ták-ku LU-áš SAL-an HAR. SAG-i e-ip-zi LÙna-áš wa-áš-tul na-áš a-ki (7) ták-ku E-ri-ma e-ip-zi SAL-na-áš waáš-ta-iš SAL-za (8) a-ki, ‘If a man siezes a woman on a mountain, <it is the man's crime, and he dies. If he seizes <her> in a house, <it is the woman's crime, and the woman dies.'

The participle of ak- does, to be sure, fall into line with the IE perfect; for D.MEŠakkantes 'Di Manes' is parallel to Greek oi TEOVηKÓTES 'the dead'. In general, however, the verb corresponds to άлойvýσк rather than to révηкα. We have, in short, the same situation as in a large proportion of the verbs of the hi-conjugation. There is no lack of hi-verbs denoting a state or condition of the subject (šuppi 'he is [ceremonially] pure'12, zinir 'man ist fertig'13, šalli ašeššar appai 'der grosse Gottesdienst ist aus'14), but a majority of them have the meaning of the IE present tense formations (e.g. arri 'he washes', peddai 'he hastens, flees', španti 'orévde'). From the point of view of Hittite we must continue to speak of the hi-conjugation, rather than of the perfect tense; for in general it is parallel in meaning to the miconjugation. I would therefore see in the participle akkantes 'the dead' a survival of the original force of the verb, while the more common aki 'he dies' is a part of the far reaching Hittite assimilation of the original perfect to the present.

10 Journal of the American Oriental Society 47. 182-4 (1927).

11 Hrozný, BoSt. 3. 1397; Sommer and Ehelolf, BoSt. 10. 47.

12 Sommer and Ehelolf, BoSt. 10. 7.

13 Götze, Hattusilis 99.

14 Friedrich, ZA NF 2.2942.

If we may assume that Hittite aki once meant 'he is dead', there is no difficulty in connecting it with Latin egit, perfect of ago 'drive, pass, do'. The connecting link is suggested by Latin phrases such as aetatem agere 'pass one's life' and qui tum agebant 'who were alive at that time'. If we put the verb into the perfect in this meaning we get something parallel to vixit 'he has lived'; i.e. 'he is dead' (cf. Plautus Bacchides 151: vixisse nimio satiust iam quam vivere). Possibly, however, we should start from the meaning 'do, perform'; in which case the development of meaning was similar to that seen in Greek kekunótes 'those who have finished their work; the dead'.

Latin ēgĕre 'they have done' may, then, be the precise phonetic equivalent of Hittite ekir 'they died'. That the stem-vowel è was IE is shown by Greek xa, Skt. āja 'I drove". Icelandic ōk must be analogical; possibly the ō is due to a contamination of o in the singular and è in the plural (compare below). The vowel è of this and similar Latin perfects has often been discussed 15, and widely divergent sources have been suggested. It has not usually been connected with Germanic è in plural preterits beside a in the singular, partly because Germanic shows the alternation of a and è only in roots with an initial consonant, and partly because the a which thus alternates with è represents IE

It is, however, possible that the Germanic restriction to one type of root is a secondary development, and it is also possible that in Latin ēģī, etc., the ĕ which intruded from the plural supplanted an earlier o. The parallelism of Hittite aki: ekir with šakki: šekkir and the identity of the latter pair with Goth. sah: sehun makes some such development plausible. At any rate there is little doubt that ekir is to be identified with Lat. égère.

16

There are several different verbs from the Hittite root ar-, whose forms and meanings must be kept apart. Friedrich distinguishes between arnuzi 'he brings' (originally 'causes to come'), artari 'he rises, comes, takes his stand, stands', arai (also araizzi) '<the wind, the enemy> rises', and our verb ari. This last Friedrich takes to mean 'he arrives', but as far as my observations go it always permits a perfect meaning, 'he has arrived, is come'. Typical examples are these:

Keilschrifttexte aus Boghazköi 4.4.3.27: nu-mu KI. KAL. BAD I. NA URU Har-ra-na an-da a-ar-áš (28) nu-za A.NA KI. KAL. BAD 15 See Sommer, Handbuch der Lateinischen Laut- und Formenlehre1 551, Brugmann, Grundriss der Vergleichenden Grammatik der Indogermanischen Sprachen 22. 3. 27, 4271.

16 ZA NF 2.41-5; cf. Götze, ib. 2.18. The meanings in the text are not precisely those given by either of these authors, but accord with my own observation.