No CrossRef data available.
Published online by Cambridge University Press: 22 April 2026
[An attempt to show that the Hittite ai-verbs are not continuants of IE āyo-verbs, but an analogic modification of IE denominatives in -ā-mi thru association with forms of the thematic verbs.]
1 Forms like those in -iyasi show that
was retained between some unlike vowels without loss of either of the surrounding vowels, so that there is a strong presumption against
having become -ai-.
2 If -enzi here is IE *-enti it does not concern us in this connection, but is merely another instance of the encroachment of the athematic conjugation in Hittite.
3 From the variation between -i-ya-zi and -i-ye-iz-zi and the like in the third person singular, and between -i-ya-an-zi -i-an-zi and -i-en-zi in the plural, it is safe to conclude that there was an actual variation between e- and o-grades. The other alternative, that of Sturtevant, Hitt. Gram. 54f., is to assume that the sign for -ya- was also used with the value -ye-. Against this is the fact that e and a were otherwise carefully distinguished as well as the improbability that when to designate
unambiguous combinations like -i-en- were available and actually in use, the Hittites nevertheless preferred the ambiguous -ya- which usually designates
and not
. It is hard to believe that the conservatism of the Hittite scribes would overcome the tendency to use the clearer and better means of representation which they themselves occasionally used. Furthermore, Sturtevant's positive reasons for assuming the use of -ya- for -ye- are etymological and historical, e.g. because -ya- is found where -ye- existed in IE times, as in above -yazi, it must here be pronounced -ye-. This is an extremely precarious argument in view of the ease with which such forms are explained as actual variations. Unless therefore the use of -ya- for -ye- is established by instances in which there is no possibility of analogical change or ablaut variation, Sturtevant's assumption is merely a distant possibility and does not dispose of the heavy burden of proof resting upon it.
4 Forms like da-ski-mi, in which i took the place of a < IE o, as opposed to *da-ska-mi presupposed below by the theory there presented, do not favor the assumption of a late origin of hatrami < *
, for in that case we would expect the 1 sg. to be *hatra-ye-mi or -yi-mi, which could not be fitted into Sturtevant's hypothesis.
5 Kurylowiez, Symbolae Grammaticae in honorem Joannis Rozwadowski 1.102, connects with the IE denominatives in -ā- rather the Hitt. h-class (Sturtevant compares Hitt. newah- 'renew' with Gr.
, Lat. novāre). However, to make credible a hypothesis which compares zero in all other languages with an extant sound in one, we need evidence of the previous existence of such a sound in at least one other language. In this instance there is no such evidence, and considerations of probability are decidedly against the assumption of the loss of h in the ā-denominatives, for the use without change of the nominal stem in -ā- as stem of denominative verbs is so decidedly a primitive category, that the theory of the previous existence of the h will scarcely be taken seriously in this instance. Moreover, the identity of the root part of newah- and Lat. novāre etc. does not mean anything as long as the suffixal part is not obviously identical.
6 Cf. Schulz-Sieg-Siegling, Tocharische Grammatik 350 ff.
7 To be accurate we should not perhaps use the sign k̂ to refer to IE palatals as represented in the centum-languages, since it is probable that the palatals were a purely dialectical development of the satem-languages, and never existed in the remaining IE territory. Cf. Hirt, BB 24.218 ff.; Idg. Gram. 1.233 ff.
8 Forms like daskisi in which i consistently represents IE e instead of alternating with e, are probably due to the assimilative influence of the i of the final syllable, although there must have been some spreading by analogy. In favor of this view is also the fact that in the 2 preterite we find the thematic vowel as e in such forms as daske-s, IE *-sk̂e-s, where there was no following syllable with i.
9 The 2 pl. imperative is also found with the o-grade of the thematic vowel, cf. daskatten.
10 A sporadic occurrence of an IE o in its original domain is tar-si-ga-u-e-en (1 pl.) in KBo. 3.19.2.19.
11 Occurrence of assimilated beside unassimilated forms is not surprising, both because of the complicated conditions under which instantaneous sound changes like distance assimilations took place, and because of the disturbing effect of analogy.
12 To those who insist on a proportion before they can accept an analogy I submit the following possibility. For the similar arnu-izzi, following Sturtevant, this proportion may be invoked: arnu(w)-izi : arnu(w)anzi = dask-i-zi : daska-nzi. If this is accepted hatraizzi can follow thus: arnu-izi : arnu-mi = hatra-izi : hatra-mi.
13 Also the verbs which vary between the mi- and hi-conjugations could have played an important part. The endings of the 3 sg. present of the two conjugations, -zi and -i, were readily contaminated to -izi (spelled -iz-zi). Thus in our a(i)-class ara-i 'rises again' may account for ara-izzi in place of original *ara-zi.
14 Possibly also uizzi, classified as an old thematic form above, may be merely an analogical formation of the type here discussed, the i then spreading to uit and uiddu. Only certain knowledge of the etymology and stem-formation of this verb could decide which alternative is correct. It is in fact difficult in many instances to decide between old thematic forms and analogical extension of the thematic conjugation to originally athematic forms, for even Hittite, with its general recession of the thematic type, doubtlessly shows its analogical extension in some instances.