¹ An earlier version of the part of this paper which deals with Manyika was read at the summer meeting of the Linguistic Society of America in 1960. The research on Manyika was carried out in Southern Rhodesia in 1956–7 under a Foreign Area Training Fellowship granted by the Ford Foundation, and with the cooperation of the Methodist Church in that country.

² See for example the very full discussion in L. W. Lanham, A study of Gitonga of Inhambane (Johannesburg, 1955); also Clement M. Doke, A comparative study in Shona phonetics (Johannesburg, 1931), and G. Fortune, An analytical grammar of Shona (London, 1955). Special phonetic symbols used in the present article include and (voiced and voiceless ‘whistling’ fricatives); b and d (voiced implosive stops); fi (Voiced h'); ç (un-grooved voiceless palatal fricative); ł and (voiced and voiceless lateral fricatives); q, gq, and nq (voiceless, voiced, and nasal apicopalatal clicks); (labiodorsal nasal).

³ Kenneth L. Pike, Phonemics 131 (Ann Arbor, 1947).

⁴ The phrase is C. F. Hockett's, A manual of phonology 174 (Bloomington, 1957). The influence of Hockett's thought is evident throughout this paper, and is gratefully acknowledged. I have also benefitted from correspondence with Hockett concerning the Manyika data, though the analysis set forth here is quite different from the one suggested by Hockett.

⁵ Shona is spoken in the northeastern two-thirds of Southern Rhodesia and in adjoining parts of Mozambique. Manyika covers an area of about 10,000 square miles, of which the center is approximately at Umtali. The number of speakers was estimated by Doke at 113,000. The subdialect which is documented here is spoken near the kraal of Chief Mutasa, in the vicinity of Watsomba, about 30 miles north of Umtali.

⁶ The classification of ‘implosive’ and ‘explosive’ b and d used here, and their relation to the voiced labial and apical stops which occur in clusters, differs from the treatment accorded the same sounds elsewhere. For discussion of this matter, and of the phonetics of Constituent H, see my article ‘The explosive implosive contrast in Manyika’, African studies 19.88–94 (1960).

⁷ Doke 109–24.

⁸ Fortune 13–7.

⁹ For discussion of this point, see B. Bloch, ‘Phonemic overlapping’, American speech 16.278–84 (1941).

¹⁰ Doke 75–81.

¹¹ Doke 45–6.

¹² Hockett 162.

¹³ The data for Tsonga were presented in approximately the form of Table 1 by D. T. Cole in a course in Bantu linguistics given at Georgetown University in 1961–2, and were supplemented by conversations with P.-D. Beuchat.