In the century since its decipherment, the phonology of Hittite has been much studied, but primarily from a diachronic and comparative perspective (see H.C. Melchert, Anatolian Historical Phonology, Amsterdam, 1994; S. Kimball, Hittite Historical Phonology, Innsbruck, 1999; and A. Kloekhorst, Etymological Dictionary of the Hittite Inherited Lexicon, Leiden and Boston, 2008, pp. 15–102 i.a). In the book under review, Sylvain Patri deliberately breaks from this tradition, systematically describing and analysing Hittite phonology qua synchronic system.
The core of the book consists of nine chapters, which can be divided into two parts. The first part consists of: an introduction to the language and the corpus (ch. 1); an overview of the cuneiform script (ch. 2); and a discussion of the phonological interpretation of this script (ch. 3). These introductory chapters are on the whole exemplary, full in detail and – importantly – accessible to non-specialists. Laudable, in particular, is Patri's exposition of the challenges that arise in interpreting the cuneiform script which, as is well-known to researchers in Anatolian linguistics or cuneiform studies, is limited in its capacity to represent certain phonological properties of the language. Thus, e.g., intervocalic triconsonantal clusters are written with an extra vowel sign, the purely orthographic character of which can be deduced by variable spellings of the type < kar-aš-nu-er>~<kar-ša-nu-er> (=[karsnuer] “they cut off”). Patri describes the diagnostics used to interpret these spellings and other ambiguities in the script, often employed with minimal explanation in specialist scholarship, with exceptional clarity, thereby enabling phonologists without such training independently to assess the primary data that provide the basis for his own (and alternative) phonological interpretations.
The second part then treats specific aspects of Hittite phonology: the segmental inventory (ch. 4); phonotactics (ch. 5); syllable structure (ch. 6); word stress (ch. 7); phonological processes (ch. 8); and clitics (ch. 9). Overall, these chapters are careful and thorough, with clear argumentation and a rich collection of relevant data (which is, moreover, organized into neat tables). Original claims are not lacking. Notable, in particular, is Patri's treatment of word stress: adapting tools from Slavic linguistics, he accounts for the phonologically unpredictable distribution of primary stress through the interplay of accented (i.e. lexically stress-preferring), unaccented, and unaccentable (i.e. stress-rejecting) morphemes. Patri's analysis raises interesting questions – e.g. about the status of the inherited distinction between thematic and athematic inflection, which in contrast to the historical situation plays no role in determining word stress under Patri's account (thus, e.g., historically thematic nouns like peda– “place” and athematic like ḫaran– “eagle” belong to his accentual class A). The synchronic and diachronic implications of this analysis will need to be addressed by future research.
It is far beyond the scope of this review to assess all of the individual analyses advanced in these six chapters, which provide (over 500+ pages) effectively exhaustive coverage of topics in Hittite phonology. As such, they will doubtless inform all subsequent research in this domain. For this reason, however, I note also that some of Patri's claims are seriously problematic; two of the most consequential are addressed below:
(i) Vowel inventory (ch. 4): According to Patri, Hittite has four phonemic vowels, /i,e,u,a/, with no underlying length contrast (see (ii) below). He thus rejects the hypothesis of an additional phoneme /o/, spelled with <u> vs. /u/ with <ú>. However, his claim that these spellings do not reflect, at minimum, a difference at the phonetic level is untenable (for the evidence see Kloekhorst 2008, pp. 35–60). For instance, the rounded vowel adjacent to <ḫ(ḫ)> is spelled almost without exception with <u> rather than <ú> (the form <ḫu-ú-ni-ik-zi> cited by Patri on p. 124 is singular in this respect). Diachronically, this distribution can be attributed in part to lowering of */u/ to [o] by the adjacent consonant, a phonetically natural development under the now widely held view that <ḫ(ḫ)> continues a uvular obstruent. Synchronically, [o] in this context could be analysed as an allophone of /u/ under the assumption that <ḫ> and <ḫḫ> remain uvulars in Hittite (although P opts for velar fricatives), but an independent phoneme /o/ is justified by the existence of the same spelling contrast in contexts in which no phonological conditioning factor is evident – e.g. <ku-u-uš> “these.c.acc.pl” vs. <ku-ú-ša-an> “son/daughter-in-law.c.acc.sg”.
(ii) Vowel quantity (ch. 4/7): At the phonetic level, Hittite has a contrast between short and long vowels, the latter optionally marked by plene spelling, i.e. the repetition of an identical V sign after a CV sign or word-initially before a VC sign. Observing that vowel length and word stress are closely correlated in Hittite (i.e. stressed/long vs. unstressed/short), Patri argues against positing phonemic length: surface long vowels are derived via stressed vowel lengthening. However, this hypothesis fails to account for several Hittite facts. One problem is posed by words with what appear to be multiple long vowels, such as Hitt. mūgā(i)– “incite” (e.g. 1sg.npst.act <mu-u-ga-a-mi>), since only one can be stressed. Patri's suggestion that plene spelling of the other is purely orthographic (perhaps residue of an erstwhile length contrast) is simply ad hoc, and he offers no principled criteria for determining which of the two is which. Another is that Hittite has morphemes with qualitatively identical vowels that show differing quantitative behaviour in stressed closed syllables – e.g., the participle suffix –ant–, which lengthens ([-áːnt-]) vs. 3pl.npst.act ending –anzi, which does not ([-ánt͡si]). Patri again attributes the virtual absence of plene spellings of the latter to orthographic convention, but given that this behaviour has a straightforward diachronic explanation – viz., lengthening /a1/ from prehistoric *o vs. non-lengthening /a2/ from *e – there is little reason to doubt the synchronic phonological reality of the contrast, which must reflect a phonemic length contrast of some kind (see A. Yates, UCLA PhD diss. (2017), ch. 3 for discussion).
The issues above have systemic consequences, problematizing (e.g.) Patri's claims about the syllabification of intervocalic consonant clusters (ch. 6) or the representation of geminate consonants (ch. 4/7), which cannot, however, be discussed further here. While caution is thus in order in accepting Patri's conclusions, the book remains an invaluable resource which will stimulate debate about Hittite phonology for years to come.
