¹ KUB XXXIII, 103, ii 1–2.

² KUB XII, 65, iii 5.

³ The whole Kumarbi cycle appears to be metrical, and other mythological texts could be also. Magical and ritual texts may well contain metrical recitations. One such is the Old Hittite stanza utilised in this article.

The Kumarbi cycle is of Hurrian origin, and the metre visible so far matches that of Mesopotamian tradition. However, the existence of the Old Hittite stanza, isolated though it is, suggests that a nearly identical metre formed part of native Hittite tradition. It is possible that this simple metrical structure could arise independently in languages with a dominant stress accent. A similar pattern appears, for instance, in the Germanic poetry of a much later age.

It is not the history of the genre that concerns us here however, but the light cast by the metre onto Hittite sentence stress.

⁴ For instance, in H. G. Güterbock's Ullikummi, Fragment c, line 4 of Text C has nu arunas Impaturiya appa memiskizzi, while Text B has memiskiwan dais as the verb. Text C is known to be less reliable, and memiskizzi leaves the verse short of a stress.

⁵ Examples are: Syncope in Hieroglyphic, particularly of initial a-, in Lycian and in Lydian; Some of the vowel differences between Hittite and Luwian, e.g. between Hit. i- and Luw. a-; Intervocalic consonants in Lycian appear to undergo stress-conditioned voicing.

⁶ The accentual variations found among the dialects of Serbocroat, for instance, confer great caution upon the extension of Hittite features to languages, however closely related, which are separated from it by hundreds of miles and hundreds of years.

⁷ Though, of course, comparison with other IndoEuropean languages can indicate where word-accent may once have stood. Syncope, as mentioned in note 5, plays a visible part in the development of some Anatolian languages. Again, comparison may be useful, but the nature of the cuneiform script acts to obscure the effects of syncope.

⁸ The evidence for the unstressed dependent genitive, discussed below, does not provide more than a hint. Further instances are needed of genitives in other syntactic environments, with their stress-values apparent, before a pattern of contexts can be deduced.

⁹ JAOS, LXXXIV.1, 1964Google Scholar.

¹⁰ From this context, and from its use as a component in personal names, uwa is assumed to be a term of family relationship, probably female, like anna “mother”, to which it stands parallel. In Lycian A a noun uwe may exist, not to be confused with the particle uwe, which seems also to stand parallel, in some contexts, to relationship terms such as χnna “grandmother”.

¹¹ MHE omits the genitive noun in this formula, but quotes an example of the full form which is repeated above in the third paragraph: Mukisanus Kumarbiyas uddar aruni appa memiskiwan dais.

¹² The stressed:unstressed contrast is that which is significant for the metre. In actual speech degrees of stress would be expected. The behaviour of uddar (see below) may be evidence of a secondary stress.

¹³ The dative object can be replaced in soliloquy by istanzani-si “to his own mind”.

¹⁴ MHE does not bracket uddar, and indeed the clause does not appear in Ullikummi without this word. However, there is an instance in the Divine Kingship myth, which seems to have the same metrical structure as Ullikummi, Text 1a, Col. 1, line 37, ma-a-an ^DA-nu-uš

¹⁵ As in Nesas, nominative, “Nesa” and Nesas waspas “clothes of Nesa”.

¹⁶ A clear statistical count of dependent genitives following their noun is obstructed by the frequent use of ideograms with a word order differing from that used with their syllabically written equivalents.

¹⁷ E.g. Skt. rāja-putrá “king's son”.

¹⁸ Lehmann, W. P.: “On Earlier Stages of the Indo-European Nominal Inflection”, Language, XXXIV, 2, 1958Google Scholar.

¹⁹ On the other hand, genitives formed with other terminations, such as Celtic and Latin -ī, are held to be identical with suffixes used elsewhere to form secondary nouns and adjectives, such as Skt. ratha “chariot”: rathī “charioteer”. The growth of paradigms employing these various endings belongs to a relatively late stage of IE, as is shown by the different patterns evolved by the attested languages.

²⁰ E.g. *wik-póti-s “head of clan/household” – Skt. viśpáti, Lithuanian viẽšpats “lord”. (Burrow, T., The Sanskrit Language, 1965Google Scholar).

²¹ This may be stated differently. The first member was affected by grammatical concord and acquired the -s of the second member. (For various reasons I believe the -s termination to have been the morpheme most frequently found at the end of such “compounds” in early IE). This situation is continued into Hittite after various secondary levellings.

When compounds remained, they had normally developed a special meaning and contrasted with the newer genitive constructions, which became the norm for ad hoc formation without additional overtones. The ability to form compounds was never lost, however; in fact a wide variety of unrelated languages form comparable units and show that such formations stem possibly from the general properties of language.

In the Indo- European languages there are isolated instances of compounds whose first member is in the genitive, but with the accent still shared. Examples are Skt. góṣpada “small puddle” (lit. “cow's footprint”) and Gk. Διόσκοροι “Castor and Pollux” (lit. “Zeuṣ's lads”), both with a specialised meaning. Incidentally, Skt. rathas-páti “chariot master” appears to show the same gen. ending as Hittite, instead of the usual rathasya (cf. Note 19).

A further stage still can be found as a rarity in Vedic Skt. Here a genitive first member appears and both halves are accentuated, e.g. the divine name , lit. “Lord of Prayer”. The compounding of separate words could affect any group when habitual usage led to such words being used together idiomatically. Other declensional cases are also involved, but only genitival constructions have been mentioned as being relevant to the Hittite evidence.

²² Literally, “X, my vizier, / / which words I speak to you,

To my words thine ear / / hold forth inclined”.

²³ Also plural:

V. nu nuntarnuir liliwahhir

VI. nat I-anki sarr-attair