SOAS Research Online

Abstract

This thesis provides a comprehensive phonological and morphological description of the KiNgome dialect of Swahili as currently spoken in the Northern Division of Mafia Island District on the south-east coast of Tanzania. Unlike other Swahili dialects, which are succumbing to the encroachment of Standard Swahili (SSW) forms and only found in small pockets of older generation speakers, KiNgome does not face extinction at present. It stands up to be a major means of daily communication among northern inhabitants of Mafia Island. The thesis has two primary goals: The first is to provide a synchronic linguistic description of this undocumented main variety of Mafia Swahili (designated as G43d under Guthrie's classification) at the phonological and morphological levels. The second goal is to examine the KiNgome data in a diachronic perspective by comparing its phonological and morphological systems with the reconstructed Proto-Sabaki (PSA) as proposed by Nurse and Hinnebusch (1993). Where necessary I have referred to a more distant proto-language, namely Common Bantu (CB) as proposed by Guthrie (1969-71) and which we assume to be the earliest common ancestor of the Bantu languages including PSA. Relative to the rest of the Swahili dialects, KiNgome is a conservative dialect, retaining older forms presumably from a proto stage prior to the massive influence of KiUnguja (Ung) and SSW. Most of these forms survived unchanged in KiNgome and nearly so in the surrounding rural non-Unguja Southern Swahili dialects (SD). The study shows that the contemporary KiNgome forms by and large can be linked to PSA and CB: Phonologically, KiNgome has kept more transparent forms of a prenasalised voiced/voiceless series segments than any other Swahili dialect. It has largely preserved stops, both with restricted and unrestricted distribution which link directly to CB proto-segments. However, changes have also been noticed: KiNgome has a 5-vowel system when compared to the PSA/CB 7-vowel system and has undergone fully the Bantu spirantization process. It has lost an earlier tone system. Aspiration is a regular feature of KiNgome. It has no /r/: /l/ distinction; instead it has flap [r]. It also attests certain palatalised consonants in the form of /ky-/ and /ry-/, and there is a palatal /n/ which has derived from /ni-/ that expresses 1sg. A homorganic moraic nasal (N) has resulted from syncopation in classes 1, 3, and 18 where the nominal prefix was /mu-/. In congruence with other Non-Unguja SD, KiNgome displays a full 5-suffix vowel copying (CV) phenomenon and a common VHH process. In addition, it uniquely displays a ku-/ko- prefix VHH determined by verb root vowel. Morphologically, it has retained a simple verbal system characterised by an unmarked perfective/past tense with VC operating in the final vowel suffix. The most significant morphological aspect of the KiNgome nominal system is a clear opposition between classes 11 and 14. It also joins other non-Unguja SD, Comoro and Pokomo in displaying sets of 2sg and 3sg persons which are mainly determined by certain tense/aspect formatives, polarity and finite form (subjunctive). Although I have not subjected KiNgome lexical data to a lexicostatistic analysis, the thesis demonstrates the existence of a set of a peculiar lexical items that are also found in the neighbouring KiMwani, KiEkoti, Makonde, KiMakunduchi and in Old KiUnguja. The results of my field fieldwork largely corroborate Nurse and Hinnebusch's reconstruction of many features of PSA. In some cases, I have proposed a reconsideration of some aspects of their reconstruction in the light of the new data from KiNgome. The case of KiNgome is of interest to the field of Swahili dialectology and Bantu linguistics in general. Undoubtedly, it will offer valuable material for systematic comparison with other Swahili dialects as well as for the contemporary study of Swahili phonology, morphology and lexicon.