________________ REVIEWS 237 111 confusion even worse, when R comes to the vibrants (pp. 35-36), he follows the usual Dravidianist practice, using the underbar for the alveolar flap, but no diacritic for the more fronted trill. Pp. 20-21: R here describes alveolar it and nd sequences which, in my experience, are rarely heard. In colloquial Tamil, they generally merge with dental (tt) and alveolar [nn), respectively; in some literary pronunciations, they are rendered as alveolar (ttr] and (ndr).15 P. 26: R lists fricatives s and s but says nothing about s, a common alternant not only of s and s, but also of c.16 P. 35: R distinguishes flap [r] from trill (r] - a literary distinction which many Tamil speakers claim to make, but which few actually carry out.17 P. 40, middle: in the passage"-C2 is always [c]", c is an error for j. Pp. 65-6: the phonemic sketch assigns not only (P TK), but also [b d g], to /p tk/, in spite of the existence of contrasting examples like palam'a measure', balam 'strength' and kisi 'parrot', gili 'fear' (pp. 53-5). R simply notes that the initial voiced stops occur in loanwords; he seems not to realize that his analysis will only work if these loanwords are somehow marked as such. For unexplained reasons, R keeps c and j as separate phonemes. P. 77: the heading "/n/ (n)" (where the underbar means "dental") occurs with examples of initial n, which seems to contradict R's statement (pp. 28-29) that initial n is alveolar. The confusion probably arises from the Tamil orthography, since the symbol used for initial n historically designated a dental. P. 80, bottom: /p/ (n) is a misprint for /n/ [n]. 5. Venkateswara Sastry's Telugu phonetic reader struggles with fewer analytic problems; in general, it presents a clear picture of Telugu phonetics, with some valuable comments on regional and social variation. The comparisons with Hindu phonetics will undoubtedly be useful for North Indian students. Since my own experience with Telugu is relatively slight, my comments will be few: P. 35: VS describes m as having a variant [n] when intervocalic; other writers have also noted this pronunciation in final position, e.g. in koopam 'anger'.18 P. 38, sec. 3.5.6: [n] is a misprint for [n]. P. 48, middle: [u] is a misprint for [w]. P. 70: VS's phonemic sketch states that "In case of word-final vowels, the feature of length does not serve as a distinctive characteristic of words. In monosyllabic words the word-final vowel is always long. In case of other words the final vowels are highly variable in length, generally short when the word is not final in phrase, longer when it is in phrase-final position". This corresponds to the generalization which I stated for Dravidian in $2 above; but in fact, VS gives Telugu examples elsewhere with final i vs. ii, e.g. pilli 'cat', kislii 'betel leaf' (p. 13), and with a vs. aa, e.g. paata 'song', paagaa 'turban' (p. 58). It seems that the distribution of vowel length in Telugu still awaits an accurate statement. P. 71: VS states "The alveolar affricates [ts] and [dz] are not phonemic. They are grouped with palatal affricates in this phonemic analysis." But elsewhere (pp.28-31) he gives examples which show apparent contrast between alveolar and palatal affricates: khartsu 'expenditure vs. carca 'discussion', and kuudzaa 'a pot' vs. puuja 'worship'. He observes that ts and dz are replaced by c and j in some dialects, but does not clarify the background of these sounds: namely, that in coastal Andhra, Dravidian 15 Cf. Arden, Tamil, pp. 50-51. 16 Cf. Fowler, pp. 364-5, where the symbol (c) is used. 17 The frequent merger of these sounds is mentioned by Arden, p. 50, and Fowler, pp. 363-4. 18 Cf. Bh. Krishnamurti, Telugu Verbal Bases (Berkeley, 1961), p. 5.
