ABSTRACT
Such information as is available on Swahili intonation is scattered in various text-books, monographs and papers. Most of it cannot stand in direct comparison with the analysis presented here, simply from the point of view of the provenance of the data. In the past informants have been in the main Zanzibaris: this is the case with Tucker and Ashton (1942), Ashton (1944), Polome (1967: 'Zanzibar and along the Mrima coast'), and Closs (1967: one informant here was 'from Tabora, but long resident in Bagamoyo'). Watkins (1958) used non-native informants, one a Luo speaker, one a Kibena speaker. Neither Polome nor Closs specify whether their intonational analyses are based on the utterances of one of their informants or represent generalisations derived from the speech of all speakers observed. We assume the latter in the case of Polome: cf. p. 52: 'It is possible to define at least the main features of a commonly acceptable Swahili intonation system'. The data in Tucker and Ashton and in Ashton derive from one and the same speaker, a Zanzibari. Whilst there is no objection to the investigation of the phonetic and phonological structure of a language as used by second language speakers as a primary language, in the sense of Whiteley (1969), the outcome cannot be called a 'Swahili Phonology', as Watkins baldly titles his monograph. We have ourselves purposely left out of account in the present analysis the utterances contributed to text N by a Swahili L2 speaker (native language Chagga) , on the grounds that such data cannot for our purposes be brought together for examination with data from Ll speakers. From this point of view alone, then, it seems that Watkins' analysis must stand somewhat apart from the main body of accumulated data on the intonation of Swahili, and we do not accept his submission that 'although in the KiSwahili spoken by these two persons there is something more than mere individual differences, it is quite evident that they speak a single language in which they converse with each other without difficulty. This language may be analysed, therefore, in terms of uniform phonological and morphological features.' (p. 28, our italics) ... Even were this so, and it is a large assumption, particularly at the surface level at which Watkins' analysiS is made, the 'language' would still be L2 Swahili. As he himself observes, ' "KiSwahili", as the word is employed here, must be defined as the language of the speech community that comprises only the two members named above' (p.28). Additional doubt is cast on the validity of Watkins' analysis in the remarks contained in the prefatory note to the tone-marked text which concludes it. This text is meant to be exemplificatory, and does not represent he data on which the analysis is based. The text was read
reading, but there was not consistency as regards prosodic phonemes, particularly juncture, and sometimes stress and intonation, either between the two readings or between the two recordings of each person. Neither of them was satisfied with his performance. The text presented is a revision of the recorded versions. This revision was made with the assistance of Mr. -. It corresponds more closely to "normal" speech than do the recordings'. (p.39). The problem of variation is well-known, but the answer is not to normalise texts in this way. The details of the variation and the method of normalisation are not revealed. On the basis, then, of the status of the informants' Swahili and of the methodological approach, we venture to suggest that Watkins' treatment of Swahili intonation leaves something to be desired and should be consulted with caution. We feel justified in excluding it here from further consideration. The Swahili intonation data presented in Tucker and Ashton is of a different standing from that in Ashton in that the former do not pretend to comprehensiveness or to the status of a formal analysis of the intonational system, but are rather a collection of more or less random observations arising from a study of Swahili phonetics. Closs's statements, likewise, whilst they do reprQsent a formalisation of part of the system, relate to only a fragment of the language, with no indication of how general she holds her categories to be, or of whether, indeed, they are held to apply at all outside the 'copula constructions' under discussion in her paper. Ashton and Polome both present analyses that are claimed to be substantially complete, that is, to have isolated, at the highest order of abstraction, the components of a unit of Swahili phonology and the classes of items operating within the components. Both admit that refinement of their statements will be possible, the addition of subclasses,perhaps, cf. Ashton p. 11: 'There are four main intonation patterns in Swahili ... ' (our italics),and Polome p. 57: 'A more detailed analysis would show ... ' The two differ in that Ashton is based, as mentioned above, On one named informant, Polome on unspecified sources. Polome, in his remark quoted above, p. 61, uses the curious expreSSion 'the •.. features of a ... acceptable ••• system' ,which raises moreover the question of exactly what degree of generalisation over the body of his informants his analysis represents.
