¹ The e-signs occurring in Hittite texts are e, el, en, eš, ḫé, me, ne, še, te, zé; the i-signs are therefore i, il, in, iš, ḫi, mi, ni, ši, di, ti, zi. The other signs commonly transcribed with i (iḫ, ik, im, ip, ir, it, iz, gi, ki, li, pí, ri, wí, kit ₉, kir, piš, lik, lim, lis, miš, mit, nir, pít, bil, fir, šir, tim, tin, dir, tir) may represent e or i indifferently.

² Goetze has recently maintained in conversation that even in Hittite documents the cuneiform signs usually read as vowels alone may represent glottal stop plus vowel, so that Hitt. e-eš-mi should be read [ˀesmi] or the like. I do not know his reasons for this opinion in detail, but I must justify my neglect of it in this paper.

No one could maintain that the signs a, e, i, u, and ú always stand for glottal stop plus vowel in Hittite. The personal ending of the pret. 3 pl. is frequently written -e-ir as in da-aš-ki-e-ir, but if this stands for [daskiˀer] such a form as pí-eš-kir (KBo. 6.3.1.17, 22, etc.) is inexplicable. Just so the personal ending of the pret. 2 pl. is often written with the sign -tin while the corresponding primary ending is often written -te-e-ni. The thesis can therefore be maintained only by assigning double values to each of the five signs concerned: e = [e] or [ˀe], etc.

Even in the initial position similar difficulties are extremely frequent. For example, I have record of the verb akk- ‘die’ 41 times with an initial vowel sign (e.g. a-ak-ta) and 46 times without (e.g. ak-ta); either a in the first group does not indicate the presence of a glottal stop or ak in the second group must be assigned an alternative value [ˀak]. Similarly e in e-eš-ḫar ‘blood’ (passim) must denote [e] on account of iš-ḫar (KBo. 3.16.1.16; 3.67.2.12; HT 1.1.37) or else the sign iš must be assigned the value [ˀes].

Of course many of the cuneiform signs are known to function in two or more values each; a single sign means either ri or tal, etc. A peculiarity of the Hittite use of the syllabary, however, is the elimination of many of these alternative values; in fact, the scribes avoid, in writing their own language, some secondary values that they employ in writing Akkadian. Under these circumstances proof that a particular sign has alternative values in Sumerian, Akkadian, or any other language cannot establish even a probability that it had the same two values in Hittite.

It is certain that the signs a, e, i, u, and ú were sometimes used in Akkadian and in Hittite to indicate vowels unaccompanied by a glottal stop; I do not know of any evidence that in Hittite they ever indicate the presence of a glottal stop.

³ It is thought that in Hurrian cuneiform writing the sign gi was specialized in the values [ge, ke] and the sign ki in the values [gi, ki]; see Speiser, Introduction to Hurrian 21 and references. In the Hittite texts, at any rate, it is impossible to interpret the sign ki as evidence for the vowel i, in view of such spellings as these (some of them very frequent): ak-ki-iš-ki-e-[it], a-ri-iš-ki-e-nu-un, ḫu-iš-nu-uš-ki-e-it, ki-e, ki-e-el, ki-e-it, ki-eš-šar, ki-eš-ši-ra-aš, li-in-ki-eš-ta, ḫu-uk-ki-e-ir, me-ik-ki-eš, na-ak-ki-e-eš, ni-in-ki-e-ir, da-aš-ki-e-ir, da-aš-ki-eš, zi-ik-ki-eš.

⁴ See Sturtevant, Lang. 12:183 (1936).

⁵ See Ehelolf, OLZ 32.322–8 (1929).

⁶ Lang. 12.189 f. (1936).

⁷ See Friedrich, OLZ 2.358–65 (1934).

^7a It has been suggested to me that some unknown feature of accent or vowel quantity may be the cause of the observed alternation, or that there may have been three or more front vowel phonemes (since original diphthongs such as ei and oi appear in Hittite as e or i). I cannot see how any such linguistic features can account for a variation between e and i in identical forms; but if any scholar can present an alternative hypothesis to account for the facts described above, I shall have to answer it or withdraw my present conclusion.

⁸ For the most part KBo. 3 and 6 contain early texts, and so the reader may be interested to note the relative frequency of citations from these volumes in the list of correspondences above. It must be remembered that the archaic material comprises only a small fraction of our Hittite texts.

⁹ Studien zur hettitischen Sprachwissenschaft 2–13, 31–6 (1917).

¹⁰ Caractère indo-européen de la langue hittite 101–14 (1919).

¹¹ ZDMG 76. 199–202 (1922).

¹² Speiser, Introduction to Hurrian 13 f. (1941) and references.

¹³ Op.cit. 22–5.

¹⁴ See Ungnad, Materialien zur altakkadischen Sprache 5–21 = MVAG 20.2 (1916); A. Poebel, Studies in Akkadian Grammar 117 fn. 1 (1939). I am indebted to Goetze for calling this evidence to my attention.

¹⁵ BoSt. 1.194–9 (1917).

¹⁶ See Sommer AU 106.

¹⁷ These words are cited for the personal ending of the 1 pl.

¹⁸ These words are cited for the personal ending of the pret. 1 sg.

¹⁹ I am not here discussing the question of vowel quantity. As far as I can see we have no way of determining whether or not there were phonemic differences of quantity. At any rate that question lies outside the scope of this paper.

²⁰ Sturtevant, Lang. 16.82 f., fn. 10, and references.

²¹ Hrozný, Die Sprache der Hethiter 186 f.; Marstrander, Caractère indo-européen de la langue hittite 127–30; Sturtevant, HG 93–5; Couvreur, De Hettitische Ḫ 140–3, 251 fn. 1; Pedersen, Hittitisch und die anderen indoeuropäischen Sprachen 164.

²² Op. cit. 143–6, 251.

²³ Cf. Couvreur op.cit. 251 and fn. 1; Bonfante, IF 55.133 (1937); Pedersen, op.cit. 164.

²⁴ Cf. Sturtevant, Lang. 14.15 (1938).

²⁵ Op.cit. 41.

²⁶ IF 55.133 (1937).

²⁷ See my forthcoming monograph on the Indo-Hittite Laryngeals, Chap. 3 §36 1.

²⁸ HG 54.

²⁹ HG 102.

³⁰ AJP 53.202 (1932).