Swahili phonology reconsidered in a diachronic perspective by Fidèle Mpiranya (review)

674 LANGUAGE, VOLUME 74, NUMBER 3 (1998) > ? as support for this as a sound change seems inappropriate since the Russian change took place in a phonetically motivated context (following consonants which were 'hard' at the time); Russian I'll later velarized under similar conditions. A comment is due on L's typological proposal that as verb inflections, reflexives and reciprocals are ordered further from the root than modal expressions (199). Bantu contradicts this notion, e.g. Swahili wazungumz -an-e (they-converse-REcrpROCAL-suBJUNCTTVE ) 'they should talk to each other' . In any case, this book is informative and thoughtprovoking on virtually every page. A superior book is unlikely to appear on this topic and yet be as comprehensible to the general linguistic readership. [Benji Wald, Los Angeles, CA.] Swahili phonology reconsidered in a diachronic perspective. By Fidèle Mpiranya . Köln: Rüdiger Koppe Verlag, 1995. Pp. 87. In this scrappy little book, subtitled "The impact of stress on morphophonemics and syllable structure', Mpiranya presents a great deal of detailed information about standard Swahili morphophonemics, well organized according to reconstructed sound changes, with due attention paid to lexical reanalyses in certain sets of examples. M's discussion of the literature on Swahili synchronic and historical phonology is well informed, and his criticisms of some of the literature are informative and well-taken. For basic data and various issues in the literature, this book will serve the interested reader well. On the other hand, some of M's more original proposals are problematic, and his discussions do not always do justice to the issues involved. Among these, I was struck by M's proposal that Swahili's penultimate word stress derives from Persian final closed syllable word stress, which he attributes to Shirazi settlers shifting from Persian to Swahili (or its ancestor) at the end ofthe first millennium (13-14). Linguistic argument is minimal, only based on preservation of the original Persian stress in Swahili loan words, without considering that some kind of penultimate syllabic prominence is widespread among East Bantu languages, even among those which remain highly tonal and for which Swahili phonological influence is implausible. Although M'sdescription ofthephonetics underlying Swahili morphophonemics is usually accurate to aninteresting levelofdetail, hisdiscussionofthe Swahili nasal syllables (his term) ismisleading atbest(38). M's claim that mpya mbwa 'dog' donotphoneticallyindicate thatthe source ofthe syllabic nasal is N ratherthanm«, that only the syllabicity ofthe nasal distinguishes the sequence mu-b > mb from N-b > mb in unstressed contexts. This false assumption is reinforced by M's discussion ofN-bovu > m-bovu (class.9/10-rotten) as distinct from mu-bovu > m-bovu (class.1/3-rotten) only in terms of a difference in syllabicity of the unstressed nasal. He never mentions that Swahili syllable -initial IbIis imploded while prenasalized IbIis not (hence not syllable-initial). Thus, mbovu < mu-bovu is distinct from mbovu < N-bovu, notonly in its syllabicity but also in the implosion of IbI. Ifthis had been discussed, the reader might realize that words like mbwa reflect an initial N- rather than mu-, since the following IbI is not imploded. It remains unclear to me whether M overlooked this or whether he tacitly chose to take advantage of the fact that there are no Swahili words where m(u) happens to precede a final syllable beginning with IbI, necessary to expose the contrast. M's critical talents are most impressively displayed in his chapter discussing an obscure theory of Herman Batibo's that Swahili has 'phonemic' vowel length. He soundly shows that there is no connection between Batibo's Swahili data, and Proto-Bantu lexical vowel length (long lost in Swahili), that what Batibo calls 'phonemic' is predictable by grammatical conditioning, that Batibo's empirical tests of his theory fail all reliability criteria, and that to the extent that they suggest anything, it is that Swahili speakers cannot generally recognize the distinctions of length that Batibo claims to exist. By the end of the discussion M exceeds necessary argumentation by dismissing as 'far-fetched' possible distinctions across sentences which play on homophony between certain nouns and pronouns, e.g. zao 'crop' or 'their (possessions )' . A regular or preferred distinction in the phonetic length of the stressed vowel according to the grammatical function is worth investigating, however the results may be interpreted. In sum, this book is most useful for the level of detail to which its Swahili data are organized in connection with the general synchronic and diachronic issues it raises even if M's omissions sometimes compromise the validity of his conclusions. [?e?p Wald, Los Angeles, CA.] Mungaka (Bali) dictionary. Rev. and trans, byJohannes Stöckle. Köln: Rüdiger Koppe Verlag, 1992. Pp. xi, 439. This volume makes available in English a dictionary , originally compiledin Germanby themissionary ...