century A.D., all major coastal towns had sprouted. ...... These patterns; and the list is by no means exhaustive,. 31 ...... Indiana University Press, pp 106-125.
CHAPTER I Introduction: Preliminary Issues Concerning This Research
1.1
Introductory Remarks
Kiswahili language, the subject of this study, has acquired the status of a national language in at least two countries: Kenya and Tanzania. According to Ruo (1994: 14), Kiswahili was declared the national language of Kenya by the country‘s first President Jomo Kenyatta in 1964. Ten years later, he directed that the language be used in parliament alongside English, which was hitherto the only parliamentary language. Bwenge (1994:18) states that Kiswahili was declared the national language of Tanganyika in 1963. Moreover, Kiswahili is also the de facto official language of the larger Tanzania [Whiteley (1969), Abdulaziz (1971), Bwenge (1994), Mreta (1998), Mohammed (2001)]. In fact, Mreta (1998:8) puts the advancement of Kiswahili into perspective when he argues thus:
The rapid growth and expansion of Kiswahili constitutes a major threat to the ethnic languages … Kiswahili is used both as a medium of instruction and a subject (in schools) … (It) functions as a means of mass communication (and its use) has expanded so much that it is gradually extending its functions as a language of every day interaction. In addition, as Mathews (1997:365) states, the language is widely used in Kenya, Tanzania, Uganda and the Eastern part of Democratic Republic of Congo (DRC) in general communicative and interactive discourses. He also states that the language is
1
increasingly being used in cities as a first language. It is also used widely for commercial and mercantile purposes. The language is also gaining greater acceptance in Rwanda and Burundi.
A number of scholars, inter alia Habwe (1999) and Okombo (2001) view Kiswahili as a lingua franca in the African region. In fact, Okombo (2001) observes that with the exception of Afrikaans, Kiswahili is the most privileged indigenous language in subSaharan Africa. This study holds Okombo‘s observation as correct. It is this study‘s view that given Kiswahili‘s widespread use, it is clearly and evidently a lingua franca. As far back as 1968, Meinhof asserted that Kiswahili was a lingua franca which serves the purposes of trade and government, spoken in the East Coast but used deep into the interior - in Burundi, Rwanda, DRC, Zambia and Malawi according to Chiraghdin & Mnyampala(1977:83-84)1. Bakari (1982:5) is even more forthright when he remarks thus:
(Ki)swahili is one of the best known African languages in the African continent. Its utility as a major lingua franca in East Africa was recognized by the Euro-Christian administrators ... This view is not only held by Kiswahili scholars but also by many other pragmatic and renowned scholars as exemplified in Chimerah (2000) 2. Chimerah quotes such researchers as Whiteley, Ali Mazrui and even Wole Soyinka from West Africa (an area where Kiswahili has not really established itself) arguing the case for Kiswahili as African lingua franca.
However, one of the most steadfast proponents of Kiswahili as a world language is Ngugi wa Thiong‘o. Even as he argues the case for all indigenous languages, he holds the view
2
that Kiswahili is the ―all Kenya national language‖ (1986:29, 1993: xiv) and goes on to suggest that Kiswahili should be made one of the languages of the United Nations (U.N) (1993:38). In fact, as at 2005, the UN itself was considering approving Kiswahili as one of its working languages.Thiong‘o further states that Kiswahili is the unifying language of culture and commerce (1993:161). However, it is in the following remark that Thiong‘o (1993:41) makes the most radical of his suggestions: But if Kiswahili … were to become the language for the world, this would symbolize the dawn of a new era …I for one would like to propose Kiswahili as the language for the world. It is, however, important to note that in terms of appropriation of the language, Thiong‘o displays discernable ambivalence.
According to Chiraghdin & Mnyampala (1977), Mutahi (1980), Bakari (1982), Nurse (1985), Abdulaziz & Osinde (1997) and Chimerah (2000), Kiswahili is a mobilizing tool that is spoken in the East African Coast from Brava all the way to Mozambique. Encarta Africana [Ms Encyclopedia (2005)] remarks that this coastal strip measures about 2,000 miles (approximately 3,200KM)3. This study affirms that the language has its place now and in the future. In the year 2003, for example, Kiswahili was declared as one of the working languages of the African Union. It has also been adopted as a language for the East African Community by the Heads of State Summit of member countries.
It is in the light of the above discussion that it has been deemed necessary to undertake a study of Standard Kiswahili morphophonology. As noted by Bakari (1982), most of the studies carried out on Kiswahili are focused on its morphology and, to a lesser extent, on
3
its syntax. The emphasis of most of the works is on grammar. Some examples include Ashton (1944), Loogman (1965), Kapinga (1973), Nkwera (1978), Gibbe (1983), Mbaabu (1985, 1992), Waihiga (1999), Kihore et al (2001), Mohammed (2001), Habwe & Karanja (2004) and Njogu et al (2006) among many others. However, at the phonological level, there has been very little research. A language that is so widely used, and one with so much potential, undoubtedly deserves much more in-depth and systematic research at all its levels.
This research has deliberately opted to study Standard Kiswahili because this is the dialect that has largely given Kiswahili international status (the language is taught in most major world Universities) and which is bound to take it to greater heights. Most of the earlier work on Kiswahili phonology, including Polome (1967), Meinhof (1968), Bakari (1982, 1985) and Maganga (1990) concentrated more on the various coastal dialects of Kiswahili and dialectology. Standard Kiswahili is the dialect that is taught in schools and colleges, used in formal trade and official circles and which therefore calls for research. Understandably, studies in other dialects are also necessary especially for comparative purposes. As stated by Chimerah (2000) Standard Kiswahili is the mainstream Kiswahili 4, a view also held in this study.
1.2 Classification of Kiswahili Language This study starts from the premise that Kiswahili is a Bantu language [Guthrie (1947-67, 1970/71), Nurse (1985)].5 Greenberg (1966) states that the African region has four main language families namely: Niger-Kordofanian, Nile-Saharan, Afro-Asiatic and Khoisan.
4
Under Niger-Kordofanian there is the Benue-Congo sub-family from which Bantu languages emanate. One of the major distinctions of the Bantu languages is that their noun-class systems portray concordial agreement.
Guthrie (1947) and Greenberg (1966) suggest that the origin of the Bantu speaking people was the area around Chad/Cameroon. This conclusion is arrived at via the use of historical comparative linguistics. Greenberg (1966) states that the Bantus migrated to Congo through valleys like Logone, Chali and Sanga up to the banks of River Mweru. According to both Guthrie (1947) and Greenberg (1966), further migrations occurred with some Bantus heading to the West, others North, some South and yet others to the East. Of those who headed East, some went North-East wards, hence the North-Eastern coastal group was formed. Guthrie (1947) states that the North-Eastern coastal group was mother to the Sabaki, Taita, Ruvu and Pare.
The Sabaki group is further divided into the Swahili, Pokomo, Elwana, Comoro and the Mijikenda. The Swahili are further divided into several sub-groups (dialects). The migration from Chad/Cameroon began around 40 B.C., reaching the East Coast around 1st century A.D. Nurse & Spear (1985) are of the opinion that the spread of the various Sabaki groupings from the North was very gradual and took about seven centuries. The first socio-economic differences emerged when the speakers of proto-Swahili and protoComoro moved to the Coast and started a coastal way of life. 6
5
This exposition seems to concur with the views of Reusch (1954) and Heine (1970) who separately state that Kiswahili emerged as a language between 700 and 800 A.D. By the 9th century A.D., the Waswahili had spread between 800 and 1,000 km of the East Coast of Africa. By the 12 th century A.D., all major coastal towns had sprouted.
Nurse and Spear (1985) further observe that the Mijikenda and the Pokomo settled at the Somalia valleys up to around 16 th century A.D. when they migrated to the area around Tana River as a result of a conflict between them and the Orma. This division of the Sabaki languages was completed when the Mijikenda migrated to the Kenyan Coast. This spread may be illustrated diagrammatically thus:
6
(1)
The Origin of the Waswahili
Niger - Kordofanian
Benue- Congo
Bantu
Eastern Bantu
North-Eastern Bantu
North-Eastern Coastal Bantu
Sabaki
Elwana
Taita
Swahili
Ruvu Pare
Comoro Pokomo
Mijikenda
The Waswahili continue to live on the East Coast to this day. More divisions occurred and subgroups and clans were formed. According to Abdulaziz (1979), Mombasa alone has over 12 major Swahili clans, a view also held by Mbaabu (1985). As already stated in section 1:1, it is evident that the Waswahili are residents of the East Coast all the way from Brava, Somalia to the Comoros and Mozambique.
7
In his 1967 classification, Guthrie places Swahili in zone G at No. 42 with the following subdivisions: 42a- Amu, 42b-Mvita, 42c-Mrima, 42d-Unguja. At No. 43 in the same zone is Pemba with three subdivisions: Phemba (43a), Tumbatu (43b) and Hadimu (43c). Komoro is also in zone G at No. 44 and its subdivisions are given as Ngazija (44a) and Njuaji (44b). Of all the Kiswahili dialects, Standard Kiswahili is so far the last to emerge.
1.3
General Remarks on Standard Kiswahili
Mohammed (2001) notes that the earliest forms of Kiswahili language are unknown. According to a number of sources7 , by the second half of 15 th century, Kiswahili was already a well established language. As this study has already noted 8, the language spread along the East African Coast and consequently split into different dialects. The emergence of these dialects contributed heavily to the need of standardising the language later in the first half of the 20 th century.
According to Whiteley (1969), standardising a language involves developing it at the lexical, idiomatic and structural levels. New words are coined to cater for all fields of life9 and its grammar is streamlined. Standardisation also involves the choice of one dialect to be used as the basis for standardisation and which would be developed at all its levels.
Ironically, the standardisation of Kiswahili was not the initiative of linguists or educationists.
Kihore et al (2001) and Mohammed (2001) observe that it was the
8
colonial administrators who felt the need to have one language to be used in administration in the East African colonies. It was as a result of that need that the colonial Governor of Tanganyika called for the holding of an educational meeting in Dares-Salaam in 1925. The meeting had the express aim of choosing one local language to be used as a medium of instruction in the whole of Tanganyika.
Chiraghdin &
Mnyampala (1977) observe that Swahili was the natural choice since it had spread far and wide in the region. However, one small problem arose; there were a number of varieties (dialects) of what was referred to as Kiswahili language. Naturally, a choice as to which amongst the many dialects was to be used had to be made. The chosen dialect would be the standard dialect.
By 1927, according to Chiraghdin & Mnyampala (1977), the need to standardize Kiswahili had arisen in Kenya. In June 1928, a conference was convened in Mombasa where Kenya, Uganda, Tanganyika and Zanzibar were represented and was attended by various experts, inter alia, Carl Meinhof, then of the International Institute of African Languages & Cultures. This respected linguist gave guidance on linguistic issues. In the meeting, it was suggested that Kiunguja, a Zanzibari dialect, be used as the base of standardisation as opposed to Kimvita, the Mombasa dialect which was preferred by some.
The Inter-Territorial Language Committee (ITLC) was established on 1 st January, 1930 with the aim of standardising Kiswahili. One of the core functions of the committee was to choose the dialect that would be used as a basis for the standardisation given the
9
competing interests between those who supported Kimvita and their counterparts who fronted Kiunguja. This had been a contentious issue since the 19 th century when Ludwig Krapf showed his preference for Kimvita. In 1870, Bishop Steere opposed this choice and gave a strong case for the choice of Kiunguja 10. Owing to administrative and trade influences, Kiunguja was chosen, leaving out Kimvita, although as a matter of fact the latter had a longer history, especially insofar as written literature and works go.
The second thorny issue involved orthography. The Arabic orthography which was used then lacked some Kiswahili phonemes such as / /, /p/, /v/, / / and / /. As a result, the Latin orthography was preferred. At the phonological level, a total of 31 phonemes which included 24 consonants, 5 vowels and 2 semi-vowels – were proposed. A system of open syllables was also proposed and agreed upon. Where two similar consonants followed each other, one had to be deleted /omitted. Work on identification of nominal classes as well as that on grammar began.
Since then, a lot of work has been done on Standard Kiswahili at various levels, although, as mentioned earlier not much has been done on phonology. Much of this work has been under the ambit of the Institute of Kiswahili Research, popularly referred to as TUKI (Taasisi ya Uchunguzi wa Kiswahili) based at the University of Dar-es-Salaam, Tanzania.
10
It is this standard dialect, one which was amongst the last to be realized, that this study proposes to survey at a morphophonological level. Of this dialect, Kihore et al (2001:2) observe that:
Presently, Standard Kiswahili is no longer Kiunguja. It is the official dialect of Kiswahili which is used on a massive scale in East and Central Africa. In contrast to Kiunguja, Standard Kiswahili has a wide range of vocabulary and technical terminologies which transcend many fields. It has a lot of literature and is used by many people in different parts of the world. Consequently, the standard dialect may capture many different environments as opposed to Kiunguja and the other dialects. (Our translation) These are views this study fully asserts. Similar views are held by Matthews (1997) who states that though Standard Kiswahili is based on the dialect of Zanzibar, it is not identical to it. A simple example would prove this point:
1.4
Kiunguja:
Mtu wenyewe ni huyu. (This is the person)
Standard Kiswahili:
Mtu mwenyewe ni huyu. (This is the person)
Statement of The Problem
The research problem of the present study stems form the premise that no comprehensive, systematic linguistic study of the sound segments and sound changes of Standard
11
Kiswahili has been carried out. In order to understand a language well, it is imperative to start from the level of the sounds of that language. This study offers a systematic study of the sound segments of Standard Kiswahili. It also investigates the relevant environments of these segments and, in addition, surveys how the various segments influence each other in different environments.
Morphophonology is closely related to phonology and morphology. As a result, in order to have a comprehensive study of Standard Kiswahili morphophonology, some aspects11 of Standard Kiswahili phonology and morphology have been explored in the next chapter. These are only those aspects that are relevant to this study.
On Standard Kiswahili phonology, this study recognises a lacuna insofar as a systematic analysis is concerned. The only systematic studies available include Mgullu (1999), Massamba et al (2004) and, to a lesser extent, Kihore et al (2001), Habwe & Karanja (2004) and Njogu et al (2006). Excepting Massamba et al (2004), the analyses in the other works is very brief and only forms chapters of larger works. The works of Bakari (1982, 1985) and Maganga (1990) deal more with dialectal issues and leave a gap in that they do not specifically and comprehensively deal with Standard Kiswahili. 12
This study intends to bridge this gap. It has surveyed Standard Kiswahili phonemes and uses them in the formulation and positing of both consonant and vowel matrices of Standards Kiswahili. It is the posited sound segments that are later subjected to a morphophonological analysis. It is also noted that no comprehensive systematic study on the phonology of borrowed words in Standard Kiswahili is, prior to this study, available.
12
As a result, it was not possible to conclusively describe the morphophonology of these words. That is why this study has treated it as part of its research problem.
Consequently, the issues that have been investigated may be summarized in one broad question:
What is the synchronic segmental morphophonology of Standard Kiswahili?
This broad question may then be split further into the following specific areas that have been covered:
(i)
Which aspects of Standard Kiswahili phonology are directly connected to Standard Kiswahili segmental morphophonology?
(ii)
Which aspects of Standard Kiswahili morphology are directly connected to the various sound changes?
(iii)
Which sound changes occur in Standard Kiswahili consonants?
(iv)
Which sound changes occur in Standard Kiswahili vowels?
(v)
Do these sound changes follow a particular pattern?
(vi)
Which sound adjustments occur in borrowed vocabulary entering Standard Kiswahili?
The fact that Kiswahili is a fast growing and spreading language and whose morphophonology of the standard dialect has not been the subject of any in-depth and systematic analysis has made this a justifiable research problem. This problem has thus been researched in this study using specific theoretical frameworks.
13
1.5
Objectives of the Study
This study has the following objectives:
(i)
To investigate some aspects of Standard Kiswahili phonology that are
directly
connected
to
Standard
Kiswahili
segmental
morphophonology. These are aspects related to vowel, semi-vowel and consonant inventories. (ii)
To find out some aspects of Standard Kiswahili morphology which are relevant to the sound changes central to this study. These include the noun and the verb morphology.
(iii)
To explore the various sound changes which occur in Standard Kiswahili consonants, to survey the environments in which they occur and to investigate whether they follow a particular system and pattern.
(iv)
To explore the various sound changes which occur in Standard Kiswahili vowels, to survey the environments under which they occur and to investigate whether they follow a particular system and pattern.
(v)
To explore the sound adjustment of borrowed vocabulary entering Standard Kiswahili and to investigate how these adjustments relate to Standard Kiswahili morphophonology.
14
1.6
Hypotheses
This study proceeds under the following hypotheses:
That:
(i)
Sound changes in Standard Kiswahili follow a specific consistent and predictable pattern.
(ii)
Sound changes in Standard Kiswahili consonants are brought about by other consonants when the affected consonants occur in neighbouring environments.
(iii)
Sound changes in Standard Kiswahili vowels are brought about by other vowels when the affected vowels occur in neighboring environments.
(iv)
The sound system of borrowed vocabulary in Standard Kiswahili is adjusted phonologically to fit in the Standard Kiswahili syllable structure.
1.7
Significance of the Study
As noted earlier, Kiswahili is a language that is growing rapidly and is increasingly gaining more recognition both regionally and internationally especially in political, social and academic spheres. As a result, this study plays a part in the understanding of the language. In order for a language to be applied in all spheres, it is important that it be
15
understood at all its levels and aspects. This study has discussed several aspects of Standard Kiswahili such as its phonology, morphology and, comprehensively, its morphophonology. To the best of this study‘s knowledge, this is the first comprehensive, systematic and fully morphophonological research with a critical analysis of the sound changes evident in Standard Kiswahili. This is yet another study in African linguistics and one which puts Kiswahili more conspicuously on the world linguistic map. It is in part a response to Bakari‘s (1982: 29) exhortation:
It is hoped that this (his work) would be the beginning of a more comprehensive study of the whole spectrum of the Swahili dialects13. (Emphasis added) Further, this study is a reaction motivated by Mberia (1993:230) who commented as follows:
This study (his) has the potential to inspire researchers to re-examine some of the phenomena here discussed or examine similar phenomena in other languages especially those related to Kitharaka. (Emphasis added) Kiswahili is one such language and true to Mberia‘s prediction, his work has turned out to ―have played a worthwhile catalystic role‖ in this research; a further step towards a better understanding of African linguistics.
This study will also assist in the comparative study of the Bantu languages, specifically as far as their morphophonology is concerned. Lastly, the study has demonstrated the validity of the two theories used (viz Natural Generative Phonology and Autosegmental
16
Phonological Theory) vis-à-vis Standard Kiswahili. The application of the autosegmental theory to segmental phonology is particularly enlightening since this had not been done before to Standard Kiswahili by any other study known to this researcher.
1.8
Scope and Limitations
This is a study on sound changes that occur to Standard Kiswahili phonological segments. The study gives an in-depth analysis of the various changes insofar as the consonants, semi-vowels and vowels are concerned. It is worth noting that the study does not delve into the suprasegmental part of Standard Kiswahili morphophonology. The main reason is that the suprasegmental part justifiably constitutes a whole research area on its own. This is so because some of the prosodic features of Kiswahili denote some interesting observation. Tone is a case in point. Most of the Bantu languages related to Kiswahili are tonal in nature. However, as many Kiswahili scholars have observed, the language does not seem to be tonal.14 This study holds the view that this is an area in need of extensive research. As a result, this research has left out the suprasegmental aspect (only mentioning it to explain segmental issues) and consequently is able to give a detailed account of the segmental aspect. It is this study‘s view that this is a better approach than giving a skeletal across-the board account.
The research uses a synchronic approach. However, once in a while, the proto-sounds as well as the undrerlying ones are made reference to. They have thus been, of necessity, important. All the same, it has not been within this research‘s scope to analyse the sound
17
changes diachronically. As a result, although this seems a diachronic aspect, telescoping is done such that many intermediary stages are omitted thus making the study more synchronic and less diachronic. However, it is noted, a diachronic study is another fertile area of research.
The study has limited itself to Standard Kiswahili as used mainly in Kenya and Tanzania. It is important to point this out since Kiswahili has numerous dialects, no less than 15 according to Chiraghdin & Mnyampala (1977:25)15. This study contends that as at the time of undertaking this research, Standard Kiswahili is amoung the latest (and may actually be the latest) of all Kiswahili dialects to emerge. It is a unique dialect in the language due to the fact that, unlike the others, this dialect was created out of deliberate efforts by groups of scholars, most of who were not even indigenous Kiswahili speakers.
Since the major theory (NGP) used in this research recognizes phonology and morphology as different entities; and proposes the marriage of the two in any study on sound change as the only way to give rise to a comprehensive theory of morphophonology, it is imperative to have an introductory chapter. The chapter gives background information concerning the classification and standardization of Kiswahili language. The chapter also covers two other important aspects. Firstly, it surveys aspects of Standard Kiswahili phonology. In this survey, vowel, semi-vowel and consonant matrices as well as the various distinctive features are laid down. These aspects are important in the later analysis of the various sound changes.
Secondly, the chapter surveys some aspects of Standard Kiswahili morphology. The aspects surveyed primarily involve the noun and the verb. These two word categories
18
form the bulk of the data and gloss used. The morphological aspects studied include the structure of nouns, noun types and nominal class systems. The verb structure as well as verb types have also been analyzed. Moreover, Kiswahili being a language that observes concordial agreement, with a good understanding of these two word categories, it has not been difficult to draw examples from other categories; this has been done occasionally as situations demand.
On borrowed words, the study has endeavored to establish whether the borrowed words adapt the same sound changes as native Standard Kiswahili words. The study has surveyed words borrowed from: Arabic, Persian, Indian, Portuguese, Turkish and English, and which words have been accepted as part of the vocabulary of Standard Kiswahili. These languages, though diverse, are by no means the only ones from which Standard Kiswahili has borrowed.
However, words borrowed from the languages mentioned above form a great number of the borrowed vocabulary in Standard Kiswahili. It is noted that Standard Kiswahili has borrowed, to a lesser extent, from languages such as Latin, Malay, French and German, among many others. All these are foreign languages and are not close to Standard Kiswahili either in cognate or geographical terms. The choice is deliberate. African, and more so Bantu, languages (from where Standard Kiswahili has also borrowed heavily) bare a lot of similarity to Standard Kiswahili, hence the preference to foreign languages in the study of this aspect. As has been indicated in the section on classification, Kiswahili is a Bantu language. This study theorises that the findings that have been found concerning the chosen languages are true for words from all other languages.
19
1.9
Theoretical Framework
This study has used two theories. The main theory used is the natural generative phonology theory (henceforth NGP) and the second, a theory used on a supplementary basis, is the autosegmental phonology theory (APT) which is also referred to as nonlinear phonological theory (NPT). While the NGP has been applied in the analysis of all the processes surveyed, APT has been used in the analysis of only two processes. It is necessary to use two theories since the major theory has been inadequate in the analysis of nasalization and vowel harmony. On the other hand, being a suprasegmental theory, APT cannot effectively help in the analysis of the core segmental aspects. It is also noted that APT is closely related to CV phonology [Clark & Yallop (1995)]. As a result, APT has been very helpful in the study of borrowed lexicon in Chapter V. The two theories are fundamentally not different as they are both offshoots of generative phonology (GP). Katamba (1993:154) states that:
Autosegmental phonology has several precursors, [as John Goldsmith, the originator of the model points out (1990:14)]. It is a direct descendant of the theory of generative phonology, given its fullest expression in The Sound Pattern of English (SPE) by Chomsky and Halle in 1968. Autosegmental phonology continues to regard as central the goals of phonological investigations set out in SPE, though it has rejected many of the original assumptions, formalisms and principles of SPE. Clark and Yallop (1995:402) state that NGP, on its part, does ―not claim to depart radically from the mainstream of generative phonology‖. For Massamba (1996), both NGP and APT are ―extended models of generative phonology‖ (:154 ff). The import of
20
all this is that the two theories have some ties that bind them together. It is, therefore, possible to use them both to study the same aspects. There are, however, some differences between the two theories as has been established in the sub-sections below.
In addition to these two theoretical approaches, this study has applied mutatis mutandis the distinctive feature model proposed by Chomsky & Halle (1968) in The Sound Pattern of English (SPE). These are GP features and they do, therefore, cope well with the two theories.
1.9.1
The Natural Generative Theory
The main theory used in this study is the NGP. This is one of the many theories that were developed to respond to some of the ―weaknesses‖ 16 evident in GP. GP is a line developed by linguists such as Noam Chomsky and Morris Halle from the mid-1950‘s17 culminating in the publication of SPE in 1968. According to Sommerstein (1977:114), SPE became a classic at once. He, however, notes that:
Subsequent work (which) has tended to take SPE model as a point of departure, begin by pointing out weaknesses that seemed to call for refinement or revision (concluding) that this or that weakness was fundamental and required a completely different model of description.
To be fair to Sommerstein, it is noted that he is against this kind of criticism saying that none of the subsequent models ―superseded SPE in practical use‖ (:114). This is
21
debatable. It may well be that SPE was the ground-breaking work, but it is this study‘s contention that works like Hooper (1975, 1976, 1979) and Goldsmith (1976, 1979, 1990) are quite important in practical use and go a long way in addressing the weaknesses evident in the first eight chapters of SPE.
NGP has been very helpful in dealing with two fundamental problems caused by the GP model of SPE, namely:
1. The problem of rule formalism and 2. The problem of abstractness of phonological representation.
Underlying these problems, as Massamba (1996:156) points out;
Are questions of naturalness, the relationships between phonology and the lexicon, linear representation of phonological items, constraining phonological rules and explanatory versus descriptive adequacy. NGP was advanced by scholars like Vennemann (1972, 1974), Hooper (1975, 1976, 1979), Hudson (1975), Hyman (1975) and Rudes (1976) among others. However, it is Hooper (1976, 1979) who systematized and streamlined the theory to give it the impetus it has today. This theory maintains that speakers construct only generalizations that are surface-true and transparent. These surface true and transparent generalizations are more natural and they do provide a sound basis for the formulation of universal substantive principles of phonology and morphology18. As a matter of fact, Hooper suggests that the major strengths of the NGP lie in its claim to bring together phonology and morphology of a given language19 . Of more interest to this study on NGP has been the use of rule
22
types; where rules that are surface-true generalizations are divided into types on formal basis. NGP classifies these rule types into four groups.
The first group involves rules or processes whose statements contain only phonetic information, i.e. the phonetically based features and phonetically motivated boundaries. 20 These are said to be phonetically motivated rules / processes and are usually referred to as P-rules, such rules are productive and cannot be suppressed. They do actually influence foreign language acquisition 21 as they apply even in loan-word adaptation. All the P-rules can be explained phonetically in reference to synchronic phonology. These rules occur quite naturally, hence the contention that they cannot be suppressed and that they are also exceptionless. They are used to describe sound changes that are universal; those that transcend language boundaries like the weakening and strengthening of sound segments as well as syllable structures.
The second group includes those rules that make larger structural changes on the whole. These are known as the morphological-rules, otherwise referred to as the MP-rules. The MP-rules are non-universal. These rules are language specific and as result they have exceptions. They are morphologically, syntactically or lexically motivated and are, therefore, of three types. The first type consists of the morphophonemic rules and these are the rules that capture the alternations of a morph in different morphemic environments. The second are the morphological spell-out and word formation rules, these show how morphs are strung together to form words, they also show the changes they undergo in order to acquire their surface forms. The third type consists of the syllabification rules. These rules assign syllable boundaries to phonological strings.
23
The third group of rules is known as the via-rules. These are rules that denote special relationship between lexical items. Via-rules enable people to relate lexical items that would otherwise be differentiated on the surface structure. According to Katamba (1989), via-rules are not productive and the few items linked by a via-rule have to be individually marked. Furthermore, he adds, a particular via-rule is not assumed to be part of every native speaker‘s linguistic competence.2 2
The last group includes the sandhi-rules. These rules occur in between the P-rules and the MP-rules. They take word boundary into consideration and at times word boundary may coincide with either syllable boundary or a pause and may, therefore, appear as a P-rule.
It is noted that as a theory, NGP does not allow for a lot of abstractness in accounting for the speaker‘s competence, in addition, the theory recognizes the various levels of language, i.e. phonology, morphology, syntax and semantics. This is one other reason why it differs from GP since Chomsky et al do not recognize the level of morphology and thus, to them, every process can only be described as a phonological process. NGP separates the two levels, hence the P-rules and the MP-rules. It is as a result of this separation that, as Matthews (1997: 238) points out, NGP is:
distinguished by a strict division between phonology and morphology to constrain either phonological rules or the forms on which they operate by conditions which exclude abstract representations. These include a condition by which rules are offered only intrinsically.
24
This separation helps in identifying the various phonological changes and to account for them using various rules. It also helps in explaining the various sound changes that cannot be accounted for phonetically. Katamba (1989:140), for example, states that:
The phonological component need only deal with transparent, phonetically motivated, regular, productive processes …. All other regularities should be handled by the morphological component. Where psychologically valid synchronic relationship is doubtful, though some semantic and phonological link exists, lexical via-rules located in the lexicon should be used to indicate the link. The above statement by Katamba just about sums up the major tenets of NGP.
1.9.2
The Autosegmental Phonological Theory
The second theory that this study has applied is the APT. This theory complements NGP in dealing with the weaknesses evident in the GP model. This study has applied the APT model as expounded by Odden (1996). The theory had earlier been advanced by scholars such as Clements & Keyser (1983) and Hayes (1986). However, it is noted that it is Goldsmith (1976) who, in his doctoral dissertation titled Autosegmental Phonology (later published in 1979), set the foundation for the development of this theory. Later, in 1990, Goldsmith systematized the theory in Autosegmental and Metrical Phonology. According to Katamba (1993:156),
The choice of the name autosegmental phonology for this model is intended to
25
reflect the fact that phonological representation consists of segments like stress, tone, vowels and consonants that appear in autonomous tiers. (Italics added) Thus, as Durand (1990) puts it, traditionally, APT has been used particularly to illustrate aspects of prosody (like tone) in language. However, Durand (1990: 252) adds that tone is by no means the only property that comes under study using this theory. Aspects such as nasalization and vowel harmony can be best studied using APT, he states. These two aspects are subjects of research in this study and APT has been applied, giving new insights into them in Standard Kiswahili.
It is, however, noted that some linguists view nasalization and vowel harmony as suprasegmental processes. One such scholar is Hyman (1975). Hyman (1975:233ff) refers to the two as ―other suprasegmentals‖, thus implying that they are peripheral in relation to what he regards as mainstream or core suprasegmentals. This approach was mainly used by linguists belonging to the Firthian school of thought in the 1950‘s through to the 70‘s but as Hyman (1975:236) rightly observes, the approach later received ―less and less support‖. This research does not subscribe to that approach and as will be evident the two are segmental and not prosodic processes. In fact, while one may argue a case for nasalization as a prosodic feature, vowel harmony is evidently and obviously a segmental one.
One of the major weaknesses of GP (as proposed in SPE) that APT addresses concerns the linear theory of phonology that it proposes. The theory holds that a phonological representation is an M x N matrix of feature values were M has universally determined
26
rows that correspond to the distinctive features and N columns corresponding to the number of segments and boundaries in a string.23 Here, the SPE assumes that speech - at least at the phonological level - is linearly arranged and that it consists of sequences of phonemes separated by clear boundaries. These 2-dimensional feature matrices with rows and columns; rows representing distinctive features and columns representing a sequence of phonemes; failed to describe the overlaps found in some languages, especially African.
As a result, these descriptions started to appear not to be as close as previously thought on the segments that they describe. Goldsmith (1976) suggested that speech segments may proceed in different ways in language-specific cases. Consequently, in APT, the potential independence of different phonological parameters is crucial.
At another level, before the advent of APT, it was generally agreed that segmental and suprasegmental levels were independent entities with both linear and sequential attributes. By the 1970‘s there was more focus on the relationship between segmental and suprasegmental representations. The earlier assumption (as stated in the SPE) became questionable and it was realised that different languages may at times proceed in different ways insofar as these segments are concerned.
In general, in APT, all phonological features are arranged on separate autosegmental tiers (Katamba: 1993) and phonological representations are 3-dimensional. There are numerous tiers linked to a central organizational core of timing units. This has reduced abstractness found in earlier theories. According to Odden (1996:7),
In contrast to the linear theory, a non-linear theory postulated a richer system of
27
representation where the number of occurrences of a particular feature in a string is not necessarily the same as the number of segments. The fact that features can be represented on a separate level of structure or tier, from other features makes APT less abstract and can, therefore, describe the overlaps.
On rule formation, there is a difference between the linear and non-linear models. The non-linear rules are written to spread a feature from one node to another (subject to the condition that lines of association may not cross) or to de-link one node from another.24 Association lines are some kind of mental mechanism that takes place to relate the different tiers. However, as Massamba (1996) states, for the association lines to perform their role successfully, a condition (which Goldsmith (1979) refers to as well-formed-ness condition) is invoked. It is under this condition that the crossing of association lines is prohibited.
There are, therefore, basically three major tenets of this theory that have been very useful to this study, viz:
1. That the phonological and phonetic representation consists of a multi-linear sequence of segments in a form of tiers and occurring parallel to each other. 25 2. That the analysis of phonological phenomena utilizes less feature changing rules; it is carried more in terms of rules that delete and re-organize the various segments. 3. That the rules are not as abstract as earlier stated by GP.
28
As mentioned above, this theory has been used to study nasalization and vowel harmony. In so doing, this study has deviated from the traditional view that considered APT to be purely a theory of suprasegmental phonology, specifically dealing with tonal aspects, Durand (1990). It has also been used, to a lesser degree, in the study of loanwords.
1.9.3
Distinctive Features
This study has used mutatis mutandis the distinctive features that were proposed by Chomsky & Halle (1968:289ff)26. However, before embarking on a brief discussion of the features, an observation is in order. It has been noted that other scholars, both before and after Chomsky & Halle (1968), have laid down - in different forms - distinctive features. However, this study has deliberately used the SPE features because of some very compelling reasons. One, the SPE features are very comprehensively stated and do thus capture almost all aspects of a language‘s phonology. Infact, the two authors state that their features are universal and may thus apply to all languages (:294). Two, these features are ably and exhaustively used to analyse the phonology of Standard Kiswahili to the satisfaction of this researcher. Three, the SPE features seem to encompass the entire features written before and after it. It is, for example, obvious that the SPE features are greatly based on the work of Jakobson, Fant and Halle (1952) as well as that of Jakobson & Halle (1956). Four, since the theories used in this study are direct descendants of GP, then, these GP features are the best suited to describe sound changes under the chosen theories.
29
The main reason why this study has avoided the use of the features stated by Jakobson & Halle (1956) (although they too were comprehensive) arise from the realization that they laid more emphasis on the acoustic parameters while this study deals with segments and hence the preference to SPE‘s articulatory approach. Ladefoged (1982) too, in what are at times referred to as ―traditional features‖ [Clark & Yallop (1995:432)] seems to lean more on the acoustic features although he does use the articualtory parameters as well. Anderson and Ewen (1987) also came up with their own analysis where they preferred the use of the term ―gesture‖ to feature. These gestures do not seem comprehensive enough to serve the purposes of this research.
Now, back to the SPE features. Although, broadly, this study has used these features, deviations and minor modifications are evident. Two reasons motivated this; firstly, to incorporate some of the modifications that have emerged since 1968 and two, to enable a thorough analysis of Standard Kiswahili.
SPE groups the features into five major categories:
1. Major class features 2. Cavity features 3. Manner of articulation features 4. Source features 5. Prosodic features
It is the first four categories that have been used in this study. The major class features are defined by degrees of stricture. This feature is used to distinguish widely varied
30
phonemes. The major classes have the features sonorant & non-sonorant which distinguishes sonorants from obstruents, consonantal & non-consonantal which distinguishes consonants from vowels and semi-vowels and the last feature is syllabic & non-syllabic (the term vocalic was earlier preferred) to distinguish syllabic nuclei (like vowels) from syllabic margin sounds.
Cavity features involve the place of articulation vis-à-vis the oral cavity. These features are many and differ from one language to the other. Those that have been used in this study include the coronal, anterior, labial and distributed features. All these appear in binary form. Others are the features high & non-high, low & non-low, back & non-back; all of which relate to the tongue body.
The third category include manner of articulation features. These relate to the flow (or obstruction) of air during the articulation of various phonemes. The major components applied in relation to Standard Kiswahili include: continuant & non-continuant, lateral & non-lateral, nasal & non-nasal (oral) and lastly lax (lenis) & tense (fortis) parameters.
Source features include voiced & voiceless feature, itself a laryngeal feature; strident & mellow as well as the advanced tongue root [ATR] and retracted tongue root [RTR] feature. Some of the prosodic features include tone, stress, pitch, intonation, duration and length. This study does not apply any of the prosodic features. These segmental features have been very helpful in analysis of various sound changes.
31
1.10
Literature Review
Although there are a few remarkable works on Kiswahili language that were written early such as Ashton‘s (1944) Swahili Grammar, it is this study‘s view that Polome‘s (1967) Swahili Language Handbook was the first work to attempt a comprehensive and systematic analysis of Kiswahili language. In his analysis, Polome uses the Mrima dialect that is spoken in Zanzibar. Although this study is on Standard Kiswahili, it is worth mentioning that Standard Kiswahili had Kiunguja, a dialect that is close to Kimrima, as its base for standardisation. As Chiraghdin & Mnyampala (1977) point out, there is quite some similarity between the two dialects. However, because of the model that was used in his study, Polome does not show the inter-relationship between phonology and morphology. This study looks at Polome‘s work as a good study that sets a base for the linguistic understanding of Kiswahili language.
Meinhof (1968) in the work Introduction to the Phonology of the Bantu Languages writes on the phonology of Kiswahili language. As he puts it, the work chiefly deals with Zanzibari dialects although it occasionally makes reference to Mombasa dialects. Some of the phonological data used concerning sound shifting, assimilation, dissimilation and coalescence have been of importance to this study. In the same work, Meinhof has expounded greatly on such phonological processes as sound shifting, assimilation and dissimilation27 . These are some of the aspects of Standard Kiswahili morphophonology that form part of this study. Another aspect of Meinhof‘s work that has been of great interest to the current study is the analysis of proto-Bantu. Given that Kiswahili is a Bantu language, this analysis has been very helpful. It is, however, noted that since
32
Meinhof‘s was a study on a wide variety of languages, the exposition given concerning Kiswahili is quite skeletal while the current study is an in-depth analysis.
Maundu (1980) focuses on the main consonantal sound changes in Kikamba. Maundu uses synchronic phonological forms to determine what the reconstructed proto-forms are suspected to have been. Four Kamba dialects are compared. Although Maundu‘s is a study in Kikamba language both Kikamba and Kiswahili are Bantu languages and there have been some interesting points of convergence and divergence. Natural generative phonology is used as the main theory while the distinctive features developed by Chomsky & Halle (1968) are also used. These two aspects are similar to the framework that has been used in the current study. However, unlike in Maundu, transformational generative framework has not been used in this study. This study does not compare Kiswahili dialects. Lastly, this study has not attempted a reconstruction of Kiswahili proto-sounds. It is an analysis of synchronic phonological forms only.
One of the most comprehensive studies in this area that has been encountered is The Morphophonology of Kenyan Swahili Dialects by Bakari (1982).28 In this work, Bakari analyses the morphophonology of seven Kiswahili dialects that are spoken at the Kenyan Coast. These dialects include Chi-Chifundi, Kivumba, Kijomvu, Kimvita, Kiamu, Kitikuu and Kisiyu. Essentially, the study attempts a scientific classification of these dialects. The present study does not indulge in dialect classification as it only deals with one dialect - Standard Kiswahili. Further, it is noted that while the dialects studied by Bakari are indigenous, Standard Kiswahili is a dialect that arose out of deliberate efforts to have a standard dialect to be used in education, religion, trade and official circles. On a
33
theoretical standpoint, the study uses the natural generative phonology as popularised by Joan Hooper (1975, 1976) and Theo Venneman (1972, 1972b, 1974). This is one of the theories that the present study has applied.
Mutahi (1983) attempts a classification of southern Mt. Kenya dialects of Kikuyu and Kiembu on the basis of historical sound changes. The writer shows morpholexical differences between the dialects and the historical sound changes that may have caused them. The work uses a natural generative approach and the theory of sound change. Using this theory, the work explores historical sound systems and the rules that link the proto and the synchronic stages. An application of synchronic phonological rules as well as the rules that fix dialect boundaries is made. Although Mutahi‘s work is not on Kiswahili, it deals with Bantu languages that are phonologically very close to the subject of the current study. Like Bakari‘s (1982), Mutahi‘s work is on classification of different dialects using both the diachronic and synchronic approaches. The current study is not on dialects and it uses only the synchronic approach. However, the synchronic phonological rules evo1ved by Mutahi (1983) have been of some interest to this study.
Massamba (1986, 1987) analyses the various phonological changes that occur in Kiswahili language in two different circumstances. In the first paper titled, ―The Influence of Local Languages on Kiswahili: The Case of Mara Region in Tanzania‖ 29, the researcher studies the phonological changes that occur in Standard Kiswahili as used in the Mara region. In the second paper, ―The Effects of Modernisation on the Phonological System of Kiswahili Language‖ 30, he looks at how modernisation has affected the sound systems of Kiswahili. Some of the effects are on sound changes. These two papers have
34
been of importance to the current study especially in the chapter on borrowed vocabulary. More important has been the fact that Massamba (1986, 1987) deals with Standard Kiswahili. It is however noted that there are fundamental differences in Massamba‘s approach and that of the present study. For one, this study explores the morphophonology of Standard Kiswahili without reference to specific instances like modernisation or the effects of specific languages on Kiswahili as Massamba does in those two papers. Maringah‘s (1987) M.A. thesis is basically on sound changes in verbal extensions in both Kimbeere and Kiswahili. It has a chapter on Kimbeere and Kiswahili phonology and morphophonology which is comparative in nature. The work indicates various phonological processes using Hooper‘s natural generative phonology. Although the work is relevant to the present study, it is quite brief and Maringah states that:
Due to time limit, we will only discuss a few of the phonological processes found in each of the two languages. We will select the major sound changes.31 The present study has discussed all phonological processes that are evident to the researcher, not just a few.
In his PhD dissertation, Maganga (1990) discusses sound changes evident in Standard Kiswahili. However, this discussion forms only a part of a large dissertation that also discusses the sound changes evident in Kitumbatu, Kimakunduchi and Kipemba dialects. It is however noted that the present study is a comprehensive discussion of Standard Kiswahili‘s morphophonology. Further, this study tests different hypotheses from Maganga‘s. The whole approach is actually different. However, it is worth noting that
35
Maganga‘s work is similar to Bakari‘s (1982, 1985) and Mutahi‘s (1983). These three works put emphasis on issues concerning dialect classification. This is not the case in the present study.
Njuguna (1992) has another comparative study titled Mofofonolojia ya Kiswahi Sanifu na Kikuyu Sanifu: Mathalani Kikuyu cha Kabete: Ulinganishi. The study compares the morphophonology of Standard Kiswahili to that of Standard Kikuyu 32. This work is not very broad as it is an M.A. study. Moreover, the writer does not state all the rules involved, for example, the rule concerning palatalisation has not been discussed. The issue of Standard Kiswahili phonology is taken for granted as no introduction on it is given. On morphology, only the nominal classes are mentioned and no explanation is offered. These are some of the gaps that the current study fills.
Mberia (1993) explores Kitharaka segmental morphophonology using the natural generative phonology model. The work examines the processes that link the underlying and surface realisations of consonants and vowels in Kitharaka language. The rules that underlie and govern the processes are also discussed. The distinctive features used are based on the system proposed by Chomsky & Halle (1968). The work also studies the behaviour of segments in vocabulary borrowed from Kiswahili and English into Kitharaka. This work, though not on Standard Kiswahili, is quite relevant to the current study. The approach used is similar to the one the current study uses. However, it is noted that while the present study is on a single dialect of a broad language, Mberia‘s study is on a language. The writer even notes that the language under study - Kitharaka - ―has no
36
orthography‖ (Pg. 39). This is in great contrast to Kiswahili which is - due to deliberate efforts - standardised.
Odden (1996) gives a description of phonology of Kimatuumbi, a Bantu language spoken in Tanzania. The work also attempts an exposition of the morphology of that language. The writer acknowledges the fact that the morphophonemic of Kimatuumbi is highly complex. Of great importance to this study is the part on theoretical framework where both non-linear and lexical phonological theories are used. The chapter on segmental and syllabic phonology where the various phonological processes of Kimatuumbi are examined is also of importance to this study. It is, however noted that the study examines a different language from that of the present study. Further, only a few of the many phonological processes are put under study.
Another work that is of importance to this study is Mtalaa wa Isimu: Fonetiki, Fonolojia na Mofolojia ya Kiswahili by Mgullu (1999). This work is important to the current study as it confines itself to Standard Kiswahili. It uses the SPE distinctive features model, the same one used in this study. The work gives a detailed account of Kiswahili phonology but morphophonology is given a sub-title ―The phonological processes in Kiswahili‖ 33a where ―Kiswahili phonology‖ 33b is the main title. The work does not relate phonology to morphology but this is understandable considering the theoretical framework used. In the introduction, the author states that ―the theoretical framework used in this book is generative grammar.‖33c
Herein lies the major difference between the current study and Mgullu (1999) since different theories are used. It is further noted that since Mgullu‘s is an introductory book,
37
the author does not posit morphophonological rules to show the various processes, something that is among the mainstay of the current study.
Sarufi Maumbo ya Kiswahili Sanifu by Kihore et al (2001) is another work that has made an attempt to study Standard Kiswahili phonology; however, this is done in just a single chapter. The analysis is not detailed and has been put in a very simplified manner because, as stated in the introduction, the book is meant for students and teachers in secondary schools and teachers training colleges. The phonological processes are described in a very detached manner and there doesn‘t seem to be any attempt to relate phonology and morphology34. As a matter of fact, the work is limited in theory, approach and detail on the basis of how aspects core to the current work are handled. However since the work is in and on Standard Kiswahili, and gives examples using the same, it has been of some relevance to the current study. Mohammed (2001) has a sub-topic ―Morphophonemic Changes‖ under the topic ―Swahili Phonology.‖ The treatment given to this sub-topic in the book titled Modern Swahili Grammar is quite sketchy. The whole exposition on phonology is covered in about five pages. It is this study‘s contention that the sub-section is narrow in theoretical orientation and detail. Some of the claims made seem outrightly wrong, for example the use of ―Charahani” (sewing machine) to exemplify palatalisation in the Ki-Vi class; or the suggestion that the choice between /tu/ and /tw/ is a stylistic feature35. However, since the work is on Standard Kiswahili, it has been of some relevance to the current study - it has made the study have a critical look at the processes. The chapter on morphology has been quite helpful. Admittedly, the book is on Kiswahili grammar and maybe this explains the
38
manner in which phonology is treated. This chapter seems detached from the rest of the book.
Habwe & Karanja (2004) in their book on Kiswahili grammar have a chapter on Kiswahili phonology. They describe Kiswahili phonemes and a few phonological processes. The authors have even made an attempt at rule formulation. On the whole, the book gives good insights into Kiswahili phonology. It deals with Standard Kiswahili just as this research does. However, as the title Misingi ya Sarufi ya Kiswahili denotes, the book is largely on grammar and as a result the bit on phonology is not detailed. Many processes have been left out. Rules have not been posited both generally and formally as is the case with this study. The distinctive features, a guide to phonological analysis, have not been set out comprehensively and systematically in the work. In fact, as expounded in chapter two of this work, even the approach of analysing Standard Kiswahili phonemes taken by Habwe & Karanja (2004) is different from that of this study.
Another recent work is Fonolojia ya Kiswahili Sanifu by Massamba et al (2004). The book discusses general aspect of Kiswahili phonology and also has two chapters on phonetics. However, although the work has good insights on the phonological study of Standard Kiswahili, it is shallow and too elementary. The part on phonological changes for example, has been tackled in less than ten pages. No rules have been posited. Furthermore, the theoretical orientation of the work is not clear; the work is thus clearly different from the current study in terms of approach, depth and analysis. It is similar in that it deals with Standard Kiswahili.
39
Njogu et al (2006) also have a chapter on Kiswahili phonology in their book titled Sarufi ya Kiswahili: Uchanganuzi na Matumizi. Once again, the book is on grammar and the treatment given to phonology lacks the technical and specialized approach espoused in the current study. However, because the work analyses Standard Kiswahili, it is both relevant and similar to this study. It is noted that very few phonological processes have been analyzed in the work and, even then, in a very basic way. Although the sound changes have been indicated through use of symbols; a proper rule formulation has not been done, the current research deals generally with Standard Kiswahili phonology and specifically with sound changes and thus, rule formulation is a major component of the study. Njogu et al (2006) analyse the sound changes in less than ten pages. This research analyses the same comprehensively.
1.10.1 Other Relevant Works
Apart from the above works that relate directly to this research, some other works have been very helpful and deserve special mention. These include Guthrie (1970/71) Comparative Bantu Vols 2, 3 & 4 as well as Hinnebusch (1972) Prefixes, Sound Change and the Sub-grouping in Coastal Kenyan Bantu Languages. These works have been important in providing background information especially on matters concerning classification of Kiswahili language and issues on Proto-Bantu.
At the theoretical level, many works have been of great relevance in the exposition of various theoretical and general perspectives. These include: Chomsky & Halle (1968) on
40
the issue of distinctive features and generative phonology as a whole; Hooper (1975, 1976, 1979) on natural generative phonology as well as Goldsmith (1976, 1979, 1990) on autosegmental phonology. Others are Hyman (1972), A Phonological Study of Feʔ feʔBamileke, Sommerstein (1977), Katamba (1989, 1993), Durand (1990), Kenstowicz (1994), Clark & Yallop (1995) as well as Odden (1996) and Massamba (1996).
1.11
Methodology
This is mainly a qualitative study. It has employed two methods in collecting the necessary corpus of data. The major method used is library research. Since the standardisation of Kiswahili in the 1930‘s a lot of material has been published. It is noted that a number of Standard Kiswahili researchers inter alios Mgullu (1999), Kihore et al (2001), Habwe & Karanja (2004) and Njogu et al (2006) have written on Kiswahili phonology. However, it is the opinion of this study that these works lack the analytic adequacy espoused in this study by use of the chosen theoretical underpinnings. None of these studies suggests the use of a comprehensive theory such as NGP. It is this that has enabled this study to lay a phoneme inventory and matrices. This study has applied purposive sampling to the published works to form its corpus. However, before an item from a certain source is accepted as standard, counter-checking from other sources is done. Standard Kiswahili textbooks have been used but, mainly, the corpus is got from published dictionaries which include:
1. Tuki (1990)
:
Kamusi Sanifu ya Isimu na Lugha.
41
2. Bosha (1993)
:
Taathira za Kiarabu Katika Kiswahili Pamoja na Kamusi Thuluthiya.
3. Tuki (2000)
:
English-Swahili Dictionary (2nd Edition).
4. Tuki (2001)
:
Kamusi ya Kiswahili – Kingereza.
5. Massamba, (2004)
:
Kamusi ya Isimu na Falsafa ya Lugha
6. OUP & Tuki (1981, 2004)
:
Kamusi ya Kiswahili Sanifu
Other available dictionaries as well as glossaries of various books have also been very helpful in building the necessary corpus. In carrying out purposive sampling, the various phonological processes were identified through reading materials on sound changes and the works quoted above (though not exclusively) on Standard Kiswahili phonology. After that, the researcher picked purposively fifty words for each process. Each was was tested before the eventual twenty (for each process) were finally sampled out. These are the words that have been used as data either overtly or covertly. These words were chosen depending on their suitability to analyse the various processes. This sampling took about one and a half years.
Use of library research was seen as the best data collection method since Standard Kiswahili, just as many other standard dialects, is hardly ever used in every day speech. In fact, at the begining of this research, this study found out that even in classrooms and offices, Standard Kiswahili is never used consistently. As a result, the only reliable source of data has been the written. Spoken Standard Kiswahili was only used totest and confirm the data.
42
The use of the term corpus is deliberate. According to both Matthews (1997) and Encarta Africana [Ms Encyclopedia (2005)], a corpus is a collection of written (and at times spoken) examples of language use meant for linguistic analysis. In the case of this study, written forms (mainly from the above cited works) have been used. In so doing, as aforementioned, this study applies what Kombo & Tromp (2006) refer to as ―purposive sampling‖. Of course purposive sampling is a non-probability sampling method. The method is applied because the underlying as well as the proto forms are not easily available since these consitute past phenomena.
The second method that has been used is intuition or introspection. Having been a Kiswahili scholar for quite some time, and Kiswahili being his second language, this researcher is a near native speaker of the language. He has, therefore, used the knowledge he has acquired to make some judgments. It is this ability as a near native speaker that enables him to tell, for example, that whilst phonemes /β/ and /v/ may appear close, the former is not a Kiswahili phoneme. Furthermore, since this researcher has regular access to native speakers of the language as well as near-native speakers of Standard Kiswahili, the approach used is invariably what Crystal (1991: 86) describes as ―corpus-based‖ as opposed to ―corpus-restricted‖. However, this method has been employed under deliberately strict terms that have involved counter-checking using other methods. These other methods include getting suggestions and corpus from Kiswahili scholars. This was usually done through reading (by the scholars) of sentences with words picked from the purposive sampling. By so doing, individual biases have been avoided. Using the above methods, enough and correct corpus has been collected and analysed.
43
1.12
Concluding Remarks
This is basically an introductory chapter which deals with the technical aspects that generally guide this research. The chapter has thus set the necessary base of the current study by setting out the research problem, the hypotheses and objectives of the study. In it, available literature is reviwed and the theoretical framework guiding the research articulated. The research methodology is described and scope and limitation set. Also discussed are general issues concerning Kiswahili such as its classification and standardisation.
NOTES 1. These views are similar to those expressed by Bakari (1982). 2. See Chimerah (2000:105,132) and Matthews (1997:365) who similary view Kiswahili as a lingua franca. 3. According to other sources such as Chiraghdin & Mnyampala (1977: ix), the strip measures about 1,500KM 4. Chimerah (2000: Viii) writes; This standardized dialect has ever since become the mainstream Kiswahili, eclipsing even the more prestigious and arguably more sophisticated dialects of the northern Swahililand, such as Kimvita (the dialect of Mombasa in Kenya) and Kiamu (the dialect of Lamu and neighboring islands in northern Kenyan coast). (Emphasis added) 5. There were times when the issue of Bantuism of Kiswahili was a burning one. Owing to massive work done by researchers, it is safe to now just assume that this is an issue of general knowledge to anyone with any pretence to Kiswahili
44
scholarship. However, we find it appropriate to mention some ground-breaking works which include Guthrie (1947-67, 1970/71), Greenberg (1966), Whiteley (1969), Chiraghdin & Mnyampala (1977), Mbaabu (1978), Chittick (1974/75/79) Nurse (1985) and Nurse & Spear (1985) among a host of many others. 6. Massamba (1994), in what he refers to as ―Massamba‘s hypothesis on the origin of Kiswahili,‖ seems to have a different view. He starts by stating that Kiswahili as a language did not originate from one specific place.
According to him,
Kiswahili arose as a result of interaction of various coastal Bantu languages especially in trading activities. Later, he argues, these various languages formed one language which was then known as Kiswahili. These other languages then became Kiswahili dialects. Massamba (1994:39) states: Kiswahili is a sum total of the dialects that emerged at the Coast as a result of the interaction of different languages. There isn‘t one specific language known as Kiswahili, there are numerous languages. This study does not subscribe to ―Massamba‘s Hypothesis‖. It instead is of the view that Kiswahili arose as a language (in the manner stated by the likes of Guthrie (1947), Greenberg (1966), Nurse & Spear (1985) among others), and then split into the different dialects. 7. E.g. Nurse & Spear (1985) and Polome (1967) 8. In section 1:2 9. As a language grows, new concepts and thoughts enter its realm and this makes it imperative to have new lexicon. 10. For a comprehensive exposition on the differences that arose on the choice of a dialect to be used as the basis of standardization, see Chiraghadin & Mnyampala (1977:54-56) and Whiteley (1969). 11. See section 1:8, Scope and Limitations 12. See Section 1:10, Literature Review 13. With the writing of this study, Bakari‘s hopes are now a reality. As indicated elsewhere in this research, Maganga (1990) also undertook a study such as the one envisaged
by
Bakari.
45
14. For example see Habwe (1999:4) 15. Chimerah (2000:75) also holds this view. 16. See sections below which expound on the major weaknesses found in the SPE. 17. See Sommerstein (1977: 115ff) 18. See details in Massamba (1996:160ff) 19. See Hooper (1976: ix) 20. For details on P-rules and Mp-rules, see Massamba (1996:160-161). 21. These rules have been applied in this study in the analysis of the phonology of borrowed vocabulary. 22. Katamba (1989:141) actually posits a very good example from the English language. The example is worth quoting. …you should see why this should be so by considering relationships between some English words, e.g. stink and stench, drink and drench, break and breach. The words in each pair are semantically and phonologically related. But probably most English speakers are unaware of this fact. Those who are aware of the connection may have a via-rule linking each pair. Evidence of such is ample in Kiswahili. Pairs such as: /taka/ need (verb)
[uta i]
need (noun)
/pika/ cook
[upi i]
cookery
/andika/ write
[uandi i] writing
for example are abound in the language. Accordingly, such pairs may only be linked by a via-rule. The link may not be obvious to all Kiswahili users. 23. See Katamba (1993: 154ff)) and Odden (1996:7) 24. See Odden (1996: 10). 25. Massamba, D.P.B. (1996:174) 26. The distinctive features in SPE have been used and written on by numerous scholars e.g Hyman (1975), Katamba (1989, 1993), Durand (1990), Clark & Yallop (1995), Mgullu (1999) and Akmajian et al (2001). Matthews (1997) refers to SPE as giving a
46
―classic account of generative phonology‖ and as ―the source for one widely used taxonomy of phonetic features‖. 27. Meinhof (1968:12-16) 28. Bakari‘s PhD dissertation (1982) was later published in 1985. 29. Paper in Kiswahili Vol. 53/1 pp 67-83, a journal of the Institute of Kiswahili Research, Dar-es-Salaam. 30. Paper in Kiswahili Vol. 54/1 PP 142-151, a journal of the Institute of Kiswahili Research, Dar-es-Salaam. 31. Emphasis ours. 32. This study is of the view that the issue of Standard Kikuyu is quite debatable. Although the writer attempts to explain that Kabete‘s is the standard dialect since people form areas like Ndia, Embu and Gichugu call every other Kikuyu dialect Gi-kabete, this argument is neither linguistic, academic nor pragmatic. 33. (a, b & c ): Our translation. 34. Mwangi P. I. (2002:5-9) in a paper titled ―Mofofonolojia ya Kiswahili: Nyufa Pana Katika Uchapishaji Wake‖, which was presented at an International Conference on Kiswahili in Nairobi, Kenya cites a number of errors and omission in Kihore‘s (2001) work. He specifically shows that the section on morphophonology lacks a systematic approach based on clear theoretical underpinnings. 35. This was a stand taken earlier by Polome (1967) as the author states.
47
CHAPTER II A Survey of Some Aspects of Standard Kiswahili Phonology and Morphology
2.1
Introductory Remarks
This is an introductory chapter which aims at laying a background for the research. The chapter surveys the major aspects of Kiswahili phonology leading to the laying down of vowel, consonant and semi-vowel matrices. An application of the distinctive features is done in each case. The syllable structure is very important in analyzing the various sound changes that occur and this has therefore been highlighted.
On Kiswahili morphology, the nominal types, structures and classes are explored. Different structures do occasion different sound changes.
Since Kiswahili language
observes concordial agreement, the nouns of the language are quite dominant. They have a big impact insofar as sound changes are concerned. Lastly, the verbal types as well as the structure of the verbs are explored and exposed. These two word categories will be referred to time and again, hence the need to have a pretty good understanding of them.
2.2
Some Aspects of Standard Kiswahili Segmental Phonology
This study uses a morphophonological theory that recognizes both the phonological and morphological levels of a language. Consequently, it is important to understand these two levels well vis `a vis Kiswahili in this study. As a matter of fact, to understand the morphophonology of a given language well, it is important, actually imperative, to have a
48
good grasp of that language‘s phonology. It is in view of the above that this section on Kiswahili phonology has been deemed necessary. Kiswahili phonemes fall into three categories; these are: i)
Vowels
ii)
Consonants
iii)
Semi-vowels
This study will, at this point, review these phonemes.
2.2.1
Standard Kiswahili Vowel System
Kiswahili language has five vowels. These vowels are /i,u, , ,a/. Using the three features that are universally applied to the description of vowels, these vowels may be represented diagrammatically as follows: (2)
/i/
/u/
/ /
/ / /a/
There seems to be a general agreement on the features of four of the vowels among Kiswahili linguists. However, the feature [BACK] for vowel /a/ appears to be debatable. For example, Mohammed (2001) suggests that it is a low front vowel. Kihore et al (2001) are of the opinion that /a/ is a low centre (near back) vowel. This is the same view held by Massamba (2004). On his part Mgullu (1999) states that /a/ is a low mid vowel, the same view held by Njogu et al (2006). Habwe & Karanja (2004) opine that in Standard Kiswahili, /a/ is a low back vowel. This study, is of the view that /a/ is actually
49
a low centre vowel in Standard Kiswahili. The different opinions may be a result of varied influences from different Kiswahili dialects. It is further noted that all the back vowels have the feature [+ROUNDED] while the front vowels have the opposite feature, [-ROUNDED]. However, /a/ seems to defy the binary mode of the distinctive feature as it is neither fully rounded nor fully spread.
This state of affairs gives weight to the
argument that it is, actually, a centre vowel. However, it is important to note that in order to correspond to the Chomskian binary system, /a/ is usually viewed as a back vowel insofar as the feature [ROUNDED] is concerned. Consequently, /a/ acquires the value [+] for that feature. As a result, the features may be stated thus:
/i/ is a front high unrounded vowel /u/ is a back high rounded vowel / / is a front mid-low unrounded vowel / / is a back mid-low rounded vowel /a/ is a centre (near back) low rounded vowel
Once it is indicated whether a vowel is [+BACK] or [-BACK], it may not be necessary to give the value for the feature [ROUNDED] since as already stated, all back vowels have the value [+] for the feature while all the front vowels have the value [-] for the same feature. Mohammed (2001:2) refers to the five pure phonemes as Kiswahili‘s ―cardinal vowels‖ and goes further to state that: Apart from the five cardinal vowels, Swahili also has secondary vowels…
50
In his work, the secondary vowels are stated as: /I/, high unrounded front vowel, lower than /i/ /U/, high rounded back vowel, lower than /u/ / /, open mid high front vowel / /, half open mid-rounded vowel. This exposition by Mohamed is erroneous since ―cardinal vowels‖ refers to a set of standard reference points or fixed positions and not specific sounds. These are Standard Kiswahili‘s ―pure vowels‖ (or monophthongs) not ―cardinal vowels‖.
This study will, however, deal only with the pure vowels. These are the vowels that are clearly evident in Standard Kiswahili although it is worth mentioning that, as Jones (1957) observed, a phoneme is not just a single sound but a group of sounds with similar distinctive features which relate to the sounds. Consequently, when five vowels are made reference to in this study, it is these five vowels and the other sounds related to them by virtue of having similar distinctive features.
It is interesting to note that while most Bantu languages have a seven-vowel system, Kiswahili has five. It is the thesis of this study that wherever the proto-Bantu vowels /e/ and /o/ appeared, they were diachronically height assimilated so as as to be realized as / / and / / respectively. The Standard Kiswahili phonemes form what Matthews (1997:384) refers to as a ―triangular vowel system‖ which appears thus:
51
(3)
/i/
/u/
/ /
/ / /a/
It is, therefore, clear that the higher a front vowel is, the more the front it is and the lower a front vowel is the less the front; the more the back it is. Consequently, though both /i/ and / / carry the feature [-BACK], /i/ is more frontal than / /. Similarly, /u/ is more back that / /, which is lower than it.
On the feature [ROUNDED], it has already been noted that all front vowels are [-ROUNDED]. However, it is further noted that the higher a back vowel is, the more rounded it is, consequently /u/ is more rounded than / /. For the front vowels, the higher a vowel is, the more spread it is. /i/ is thus more spread that / / although both are unrounded. /a/ appears to be in what Mberia (1993) refers to as rounded-unrounded border zone. In summary, Kiswahili vowel matrix displays the features in (4) below: (4)
HIGH
/i/ +
/u/ +
/ / -
/ / -
/a/ -
LOW
-
-
-
-
+
BACK
-
+
-
+
-
FRONT
+
-
+
-
-
ROUND
-
+
-
+
-
TENSE
+
+
-
-
+
52
This is the model used by Chomsky & Halle (1968) in SPE though minor adjustments have been made in order to have a factual analysis of Kiswahili.
2.2.2
Standard Kiswahili Consonant System
Kiswahili language has about 25 consonants and two semi-vowels.1 A remark is in order. The use of the word ―about‖ is deliberate.
This is because the issue of Kiswahili
phonemes lacks consensus among scholars. Some, like Habwe & Karanja (2004) suggest that Kiswahili has 23 consonants (excluding the 2 semi-vowels) while others like Massamba et al (2004) have listed a total of 27 consonants. Njogu et al (2006) postulate a table with 27 consonants and 2 semi-vowels, thus bringing the total number of Kiswahili phonemes to 34.
There are many other scholars with different views. Mgullu (1999), for example, states that the number is 24. There seems to be a general agreement on 23 of the phonemes but the contentious ones include /x/ (which Habwe & Karanja state is not a Standard Kiswahili phoneme), / / - which is not mentioned by many scholars but whose use in Kiswahili is obvious and /ph/, /th/, /kh/ and / h/, whose case Massamba et al argue for strongly, but which Mohammed (2001) suggests are merely allophones.
Other
contentious consonants include, /mb/ /nd/ /g/, / /, /mv/ and /nz/. This study has listed the 23 non-controversial consonants and also / / and /x/. / / is used widely in Kiswahili phonology and although /x/ appear to have Arabic roots, it is still acceptable as a Standard Kiswahili phoneme. The phoneme is actually used in Standard Kiswahili dictionaries inter alia, TUKI & OUP (2004) and TUKI (2001) where words such as /xa:/
53
(an interjection expressing anger) and /xalifa/ (caliph / calif) appear. The lexical items /ax ra/ (paradise) and /axijari/ (better, more beautiful) are also used in Standard Kiswahili writings. This, ipso facto, assumes the existence of sound /x/ in Standard Kiswahili. The decision on / / has been arrived at by use of articulatory parameters. However, its position is still quite disputatious especially using the functional approach which basically views any phoneme purely as a phonological reality. This, in effect, means that for any sound to be consindered a phoneme, it must exhibit a minimal pair relationship when contrasted to another phoneme. This does not seem to be the case with / / in Standard Kiswahili. If this were to be taken into consideration, then the viable alternative would be to have two phonemes; / v/ and / f/. This proposal appears to be phonologically viable.
A different view on / / is that it is an allophone. This is because /m/ and / / seem to form a suspicious pair. Using this approach, it is possible to easily realise that the two do not form a minimal pair but that they occur in complementary distribution. While [m] may appear in both initial and medial positions of a word (but never as a word final given that Kiswahili discourages closed syllables); [ ] only occurs at the initial position. The occurence of [ ] is dictated by the environment where it occurs only before the labiodental fricatives. [m] never occurs in such environments. This argument also seems plausible. The four aspirated phonemes suggested by Massamba et al (2004) have not been considered as Standard Kiswahili phonemes due to reasons to be discussed later in this section while dealing with the issue of allophony in Kiswahili.
54
As indicated in section 1:6:3, this study will apply, mutatis mutandis, the distinctive features proposed by Chomsky & Halle (1968) in defining the 25 consonants and the two semi-vowels.
On the feature cavity features, two main attributes stand out; coronal and anterior. Anterior sounds are those sounds that are articulated at the front part of the oral cavity, i.e before the hard palate meaning, no further than the alveolar-ridge. Phonemes with the feature [+ANTERIOR] include three bilabial sounds2, /b/, /m/ and /p/ as well as three labio-dentals, i.e. /f/, /v/ and the nasal / /. Others are two dental sounds: / / and // as well as seven alveolar sounds; /s/, /z/, /t/, /d/, /n/, /l/ and /r/. Consequently, a total of 15 phonemes share this feature.
The other 10 consonants have the opposite feature [-ANTERIOR], meaning that they are articulated at points after the hard palate. These phonemes include the palato-alveolar // and three palatal sounds; / /, / / and / /. The others are five velar phonemes; /x/, //, /k/, /g/ and // as well as one glottal sound, /h/.
The phoneme /h/ calls for at least two comments. One, although it doesn‘t seem to be articulated at the front part of the oral cavity, /h/ is not a back phoneme either. This is because it does not fit in the definition given by SPE (that in articulating the back sounds the tongue root should be pulled back). The tongue root seems to remain in a neutral position. From the aforegoing, it may seem as if /h/ has the value [+/-] vis-à-vis the
55
feature [ANTERIOR]. But since the theory rejects such an occurrence and calls for a binary approach, this study, for the purposes of convenience, opts to give /h/ the value [-], however, some proponents of SPE do give it the opposite value [+].
The second comment involves the status of /h/. SPE treats this phoneme (and also the glottal stop //) as a glide. One reason why it was suggested that the SPE features would be applied mutatis mutandis surrounds this phoneme. This is because this study wishes to take a different stand. This research lays heavy emphasis on articulatory parameters; and these parameters dictate that /h/ is an obstruent since there is evident and highly audible friction/ obstruction of the air-flow at the glottal area as the vocal chords articulate the phoneme.
The other features described by SPE are the major class features. These features classify sounds in relation to degrees of stricture. It is these features that define consonantal sounds versus the non-consonants; sonorants versus non-sonarants (obstruents). More important to this section is the fact that this feature also groups consonants into their major categories. Drawing from the way the various sounds are articulated, Standard Kiswahili exhibits six major consonant categories.
The first
category includes the plosives, and these are six: /p/, /b/, /t/, /d/, /k/ and /g/. The second group of consonants, fricatives, has /h/, /f/, /v/, / /, //, /s/, /z/, //, // and /x/. It is worth noting that this category has the largest number of Standard Kiswahili consonants. The language has two affricates: / / and / /. The resonants in Kiswahili includes 2 liquids (the lateral /l/ and the trill /r/) and five nasals which are /n/, / /, //, /m/ and //. These
56
five nasals confirm the fact that the language uses two cavities: the nasal cavity and the oral cavity. While most of the other consonants are articulated largely through the use of the oral cavity, the nasals make greater use of the nasal cavity. On the feature [VOICED] (+ or -), Kiswahili uses both the open and closed states of the glottis, consequently, the language has both voiced and voiceless sounds. The voiced phonemes include: /b/, /d/, /g/, /v/, //, /z/, //, / /, /l/, /r/, /m/, / /, /n/, / / and //. All the other Kiswahili consonants are voiceless. 3
Other important parameters include manner features. These are the features that distinguish the nasals from the non-nasals as well as the laterals and the non-laterals. These two sound types have been mentioned above. These features also denote the continuants and the non-continuants. The continuants in Standard Kiswahili include: /l/, /r/, /j/ and /w/ all of which are sonorants as well as the obstruents /v/, /f/, //, //, /z/, /s/ / /, //, /x/ and /h/.
All the other Kiswahili consonants carry the feature [-
CONTINUANT]4.
On the surface, it may appear as if the issue of Kiswahili consonants is straightforward, however, underneath lie major problems. One such problem revolves around ―double consonants‖. This problem has been appreciated by a number of Kiswahili linguists. Ashton (1982), for example, is of the view that ―mb‖, ―mv‖, ―nd‖, ―nz‖, ―n ‖ and ―ng‖ are Kiswahili phonemes; individual consonants to be particular. 5 This is the same view held by Njogu et al (2006) who view ―mb‖, ―nd‖, ―n ‖ and ―ng‖ as distinct phonemes.
57
This appears to be so because at times these pairs of sounds are articulated as one. Mohammed (2001) points out this problem but does not offer any suggestion or view over the same.
Polome (1967) in his pioneering work does not include them as
phonemes. Kihore et al (2001) as well acknowledge the problem.
Mgullu (1999) suggests that this should be treated as an area requiring intense research by phonologists. As is evident, the problem arises when a nasal is followed by another consonant in a harmonized environment. The two seem to combine to form one sound, like when articulating the words in (5): (5)
/mbuzi/
goat
/ndani/
inside
/ vua/
rain
/mb ga/
vegetable
At other times, as in (5a), the nasals seem to be distinct sounds even in the same environments: (5a)
/mg m /
riot
/mbab /
warlord
/mg n wa/
sickling
As Mgullu (1999) states, the problem lies in deciding when the nasal has combined with the other consonant to create one phoneme and when it has not. It is obvious that it is never alleged that the two sounds always combine when they are in neighbouring environments. The two combine in specific environments.
58
This study adopts the
suggestion put forth by Mgullu (1999) that articulation be used as the basis of identification and not orthography. This study adopts this view because the features used in SPE, the model applied here, pay great emphasis on articulation as opposed to any other form of classification. The suggestion, in this case, is that each word be studied on its individual merit and in specific contexts. If the two sounds are articulated as one phoneme, then it should be taken that the two have combined, and, for linguistic purposes, they should be viewed as one phoneme. If the two are pronounced as two distinct sounds, then, they should be treated as such. This is the approach that will be favoured in this study when confronted by the problem.
Another phonological problem in Kiswahili linguistics concerns the presence or absence of allophones in the language. This problem may be viewed vis-à-vis that of the presence or absence of aspiration in Standard Kiswahili. While some Kiswahili linguists like Mdee (1986), Mohammed (2001), Habwe & Karanja (2004), and Massamba et al (2004) suggest that there exist allophones in Kiswahili, others like Mgullu (1999) are of the view that they do not exist. It is, however, important to add that even amongst those who argue for the existence of allophones, the parameters used to describe them are very different.
Drawing from the views of research organizations and linguists such as Lyons (1968), Hartman (1972), Richards et al (1985), Crystal (1985, 1991), Fromkin & Rodman (1988), TUKI (1990), Clark & Yallop (1995), Yule (1996), Mgullu (1999), Akmajian et al (2001) and Seth & Dhamija(2004), among others, this study adopts the views of
59
Ladefoged (1962) and Verma & Krishnaswamy (1989) that there are three basic characteristics which sounds must meet for them to qualify as allophones. These characteristics are; 1.
Allophones of the same phoneme do not cause change in meaning when used.
2.
Allophones of the same phoneme never occur in the same environment; they exhibit complementary distribution.
3.
Allophones of the same phoneme are very similar phonetically.
This study suggests that the above three benchmarks form the basis of identifying allophones. Mdee (1986) states that [x] and [h] are both allophones of /h/. He states that these are allophones because when they are used in words such as, /x ri/ and /h ri/ (better), there is no change in meaning. But it is this researcher‘s view that the two are not allophones but are actually two different phonemes. Two reasons guide this view; first, these two sounds are not similar phonetically as is evident in (6) below: (6)
/h/ + glottal - high -Back + spread
/x/ + velar + High + back -spread
(SPE Model)
Secondly, the use of the two sounds in the example given above deviate the rule of complementary distribution; that is they occur in the same environments (word initial, before the mid-high front vowel). It is this study‘s opinion that what actually occurs is a free variation of two distinct phonemes. This also applies to other examples given like in the case of [] and [g] being allophones of /g/. The two sounds occur in the same
60
environment and cannot, therefore, be allophones. Further, /g/ is a stop while // is a fricative.
Habwe & Karanja (2004) state that [a] and [ã] are allophones of /a/. The gist of their argument is that when /a/ is preceded by a nasal, then it becomes nasalised (see section on nasalisation in chapter 4) as in [mãmã] (mother) as opposed to it‘s articulation in [sasa] (now). This study is in agreement with this suggestion since, even though the two sounds occur in the same position, the environment is different and distinct. The use of either [a] or [ã] does not occasion any change in meaning. Secondly, the two are very similar phonetically and thirdly, the two occur in different environments. /a/ is realized as [ã] in the environment of being preceded (or followed) by a nasal while [a] occurs in the environment of following a phoneme that has the value [-] for the feature [NASAL].
Nevertheless, the two authors also fall in the snare of meaning considerations alone when they suggest that [p] and [f] are allophones of /p/ (:56) since the realization of /p/ as /f/ in the items /mlapi/ (glutton) and /mlafi/ (glutton) does not cause a change in meaning. The two are distinct phonemes. While /p/ is a bilabial plosive; and thus a stop, /f/ is a la biodental fricative.
The two differ in many other features like [STRIDENT] &
[CONTINUANT] where /p/ has the value [-] and /f/ has the value [+] for both. /f/ and /p/, in the example above, occur at the same position and under the same environment; consequently, again, this is a case of free variation and not complementary distribution (allophony).
61
However, it is Mohammed (2001) who engages in a complex argument. First, he suggests that aspiration as a feature is present in Kiswahili (7). He further states that aspiration causes allophony in Kiswahili (:10, 11).
Mohammed states that [p] and [ph] are
allophones of /p/ while [k] and [kh] are both realizations of /k/. This study wishes to tackle this issue at three levels, viz. aspiration, environment and meaning.
Aspiration: The issue of aspiration in Kiswahili - even in ―other‖ dialects - is quite contentious; for example, Polome (1967:39) observes that: The problem of the voiceless aspirates is one of the most complex of Swahili phonemics. Whereas their phonemic status seemed fairly well established a few generations ago, there is considerable inconsistency as regards aspiration in the idiolects (sic) nowadays in Zanzibar and the neighbouring coastal areas. (Emphasis added)
In his work, Polome does not suggest that he is dealing with Standard Kiswahili. He uses a cross-section of Kiswahili dialects but still concludes that aspiration does not seem to be a feature in modern Kiswahili. He gives a number of probable reasons ranging from its low functional yield at the semantic level and the absence of notation of the aspiration both in the older Arabic writing system of Kiswahili and the current Romanization to the growing influence of numerous non-native Kiswahili speakers of the language.
Bakari (1982), however, categorically states that aspiration is a feature that is universal or common to all the Kiswahili dialects, ―including the standard dialect‖. Bakari further adds that most Swahili scholars have noted the phenomenon. And indeed they have.
62
Katamba (1989) even posits a rule that denotes aspiration. Massamba et al (2004) on their part state that even though aspiration in Kiswahili has been ignored by many Kiswahili linguists, it is ―an important feature in the language‖ (42). They state that it is the non-coastal Kiswahili speakers who are unable to identify the feature and therefore state that it is non-existent. It is, however, noteworthy that for Massamba et al, aspiration in Kiswahili does not cause allophones but phonemes.
The view of this research is that aspiration may be an important feature in other Kiswahili dialects like Kimvita, Kiamu and Kiunguja; and is recognisable mostly by native Kiswahili speakers and scholars. However, in practical use, aspiration does not seem to be an important feature in Standard Kiswahili. It may even be absent, given that the so called non-coastal (non-native) speakers of Standard Kiswahili are quite many. The nonexistence of aspiration would then denote the absence of ―aspirated allophones‖ in Standard Kiswahili. Indeed, this study is of the opinion that such allophones do not exist in the dialect. Even if it were assumed that the feature exists; what would be justified is the existence of more phonemes [Massamba et al (2004)] and not allophones. Environment: Mohammed (2001:11) states that: The allophone is a subclass of sounds which is in complementary distribution with another …. In other words, the various variants of a phoneme are called allophones. This study has no problem with that definition, but it is Mohammed‘s application of it in Standard Kiswahili that poses a problem. examples:
63
Mohammed (2001) gives the following
[phaka]
cat
[ph]
[mpapai]
pawpaw tree
[p]
[khapa]
zero
[kh]
[mkat ]
bread
[k]
In giving the above examples, the author implies that the realization of either [p], [k] or [ph], [k h] is dictated by the position of occurence (this is important since allophones are also referred to as positional variants). In fact, Mohammed (2001:11) goes further to state that: The aspirated [ph] occurs in the initial position whereas in the latter, the unaspirated [p] occurs in the medial position. This means that the two variants of the phoneme /p/ are allophones. They occur in different environments and are thus said to be in complementary distribution.
Whereas Mohammed is very particular about the position of the variants here, the whole issue becomes hazy since Mohamed, elsewhere, correctly states that even [p] and [k] do often occur in the initial positions of words in Standard Kiswahili.
He gives the
following examples: [paa]
scale
[kaa]
sit
The fact that in these and other words /p/ and /k/ do appear in the initial position indicate that the two pairs; /p/ & /p h/ and /k/ & /k h/ are not positional variants and that they do not denote complementary distribution. The only conclusion one can draw from this is that the two pairs do not constitute allophones in Kiswahili language6 .
64
Meaning: It was earlier stated that the use of allophones of the same phonemes should not occasion change in meaning.
However, in his exposition, (Mohamed (2001:7)
indicates that aspiration does occasion change in meaning in Standard Kiswahili. This is evident in the following examples which he gives:
[phaa]
deer
[paa]
scale
[khaa]
crab
[kaa]
sit
Given the above information, this study arrives at the conclusion that what we have above are minimal pairs and as a result what Mohamed refers to as allophones would appear to be phonemes. Massamba et al (2004) take the latter approach and enumerate /ph/, /th/, /k h/ and / h/ as Kiswahili phonemes. This would not be peculiar to Kiswahili. In Hindu, /k h/ and /k/ form a minimal pair; they are different phonemes.7
From the aforesaid, it is obvious that the examples given by Mohammed are ambiguous, self contradictory and, seemingly, erroneous.
Aspiration per se, it must be emphasized,
is not the only necessary feature for the identification of allophones.
On his part, Mgullu (1999) has insisted severally, (for example on pages 60, 64 and 71), that in his view, Kiswahili language has got no allophones. This conclusion is borne out of his stand on aspiration. The second reason, in this study‘s view, is his (narrow?)
65
approach as regards the issue of allophones not appearing in the same environment. It appears he confuses the terms position and environment and thus leaned heavily on the position at the expense of environment. The example given by Habwe & Karanja (2004) concerning /a/ and /ã/ depicts the wide scope of the terms environment and neighbourhood of sounds. This issue is discussed in detail in section 4:6. It is, however, worth noting that Mgullu‘s contribution in defining and explaining the concept of allophones theoretically cannot be gainsaid. It is enormous.
These aspects of aspiration, allophony and ―double consonants‖ may seem peripheral. They are, however, an important part of Kiswahili phonology and their understanding is very crucial when analyzing the various sound changes. Giving a clear line to be followed by this study has thus been deemed to be of paramount importance, hence the above discussion.
Having given the above insights concerning Kiswahili consonants, this study now finds it opportune to lay down a summary of the consonants in a matrix, thus:
66
(7)
Categories of Kiswahili Consonants
PLACE OF
Categories According to Manner of Articulation OBSTRUENTS STOPS PLOSIVES AFFRICATES
ARTICULATION
Bilabial Labiodental Dental Palatoalveolar Alveolar Palatal Velar
SONORANTS FRICATIVES NASALS
/p/ /b/*
/m/ / /
/f/ /v/ / / // // /s/ /z/
/t/ /d/
/n/ / /
/ / // /k/ /g/
Glottal TOTALS
6
LIQUIDS LATERAL TRILL
/l/
/r/
/x/ // // /h/ 10 5 1 1 *In cases where pairs appear, the sound to the left is voiceless while the second is voiced.
2
While referring to the SPE model, Katamba (1989) summarises the distinctive features into matrices which contain arrays of features necessary for the representation of various sounds. This study will in (8) and (9) below, mutatis mutandis, reproduce those matrices so that only those phonemes that are applicable to Standard Kiswahili are represented.
(8)
Const. Cont. Nas. Lat. Lab. Ant. Cor.
Distinctive Feature Matrix for Sonorants
+ + + + -
m
n
+ + + + -
+ + + +
+ + -
+ + +
l
r
j
w
+ + + + +
+ + + +
+ +
+ + -
As is evident in (8) above, Standard Kiswahili sonorants are five nasals, two liquids and two semi vowels.
67
(25)
(9)
Distinctive Feature Matrix for Obstruents
+voice
b
d
g
v
z
-voice
p + + -
t + + -
k + + -
f + + + + -
+ + + -
s + + + + + -
x + + + -
Cont. Strid. Distr. Ant. Lab. Cor. High Low Back Constr. Spread
+ + -
+ + + + + -
h + + +
+ + + -
The features in (8) and (9) are the ones that will be applied in this study in discussing the various sounds and the changes that they undergo.
2.2.3
Standard Kiswahili Semi Vowels
Standard Kiswahili has two semi-vowels, viz. /w/ and /j/. As in other languages, these semi-vowels function as consonants but their articulation has some semblance to that of vowels.
Some of their features have been enumerated in the matrices containing
summary features of the consonants. It is noted that the SPE model treats /h/ as a semi vowel but this study, as stated in the previous sub-section, treats it as a consonant. Although as indicated in the matrices, /w/ and /j/ have some consonantal features, they lack the phonetic characteristics normally associated with consonants such as friction and closure. Instead, as Mohammed (2001) observes, their qualities are phonetically similar to those of vowels. In the SPE model above, /w/ is indicated as a labial. This study, as indicated earler, is of the view that Standard Kiswahili /w/ is a voiced labio-velar
68
phoneme. This is a unique sound since, unlike in other sounds, the two points of articulation are far apart. In most cases where two speech organs co-operate to articulate a sound, the two organs usually occur in neighbouring environments.
Maybe it is
because of this that scholars like Ashton (1944), Mfaume (1984), Kihore et al (2001:25), Mohammed (2001) and Njogu et al (2006) view it only as a labial phoneme. However, after much investigation, this study adopts the view proposed by Mgullu (1999) that /w/ be viewed as a labio-velar sound. Apart from the rounding of lips, (itself very much a vowel feature) 8, there is some mild friction at the velar region when articulating /w/. This mild friction, and that which is present at the labial area, denote consonantal features. /j/ is a voiced palatal sound. The minor friction evident at the hard palate is caused by the movement of the tongue upwards towards the hard palate.
It is noted that most works classify the semi-vowels with the consonants. However, for convenience; and as a result of their active role in various sound changes, it has been found expedient to offer the above brief exposition. Rules such as those governing glide formation and harmonization of /n/ and /j/ have a direct bearing on semi vowels.
2.2.4
Standard Kiswahili Syllable Structure
The last aspect of Kiswahili phonology this study wishes to comment on concerns the syllable structure. In any study on morphophonology, the syllable is an important component since morphophonology partly deals with pronunciation. Language users pronounce sounds in groups known as syllables. Each language has its own syllable structure. Syllables are constructed by combining phonemes together in neighbouring
69
environments and forming a pattern that is articulated as one. Any study on a language‘s morphophonology cannot ignore its syllable structure because sound changes usually occur when different phonemes are in neighbouring environments. Quite often, these neighbourhoods end up forming syllables.
Proponents of NGP like Hooper (1972) assert that the understanding of the syllable structure of a language is important in the analysis of phonological processes. Katamba (1989) states that the syllable is the basic unit in terms of which phonotactic rules are best stated. These are the rules that control the language users in terms of the possible sound combinations of that given language. It is the syllable that dictates which sounds combine (or do not combine) in a given language. It is in view of the above reasons that an over view of Kiswahili syllable structure has been deemed necessary.
On theoretical perspectives, it was stated in the section on theory that in NGP, syllabification forms a great part of the theory. In fact, syllabification rules are a core component of the MP-rules. Consequently, this section has the backing of theoretical orientations from the chosen main theory.
This study has identified at least nine syllable structures in Kiswahili. Where the issue of the controversial ―double consonant phonemes‖ comes up, it is treated as stated in subsection 2:4:2. The first structure is what is usually referred to as the preferred syllable structure; that of a consonant – vowel cluster (CV). As in many Bantu (and indeed
70
world) languages, this is the structure that forms the bulk of syllables in Standard Kiswahili. There are words like: (10)
/talaka/
divorce
(CV $ CV $ CV)
/mara i/
perfume
(CV $ CV $ CV)
/hukumu/
judge/judgment (CV $CV $ CV)
/h tuba/
speech
(CV $ CV $ CV)
The structure in all the lexical items in (10) is CV$CV$CV.
The second structure is that of CCV. In this structure, four types of consonants are used. The initial position is always taken by any of the five nasals, /m,
, ,
and n/, followed
either by an affricate (/ / or / /), a fricative like /v/ or by plosives /p,b,t,d,k or g/. Any of the five vowels takes the final position of the syllable. Examples include:
(11)
/ vuvi/
fisherman (CCV$CV)
/ a:/
hunger
/nd g /
bird/aeroplane (CCV$CV)
(CCV)
/gamia/
camel
(CCV$CV$V)
/mb ga/
vegetable/cabbage (CCV$CV)
71
The third structure consists of C½VV. Various consonants are used in the initial position and are then followed by a semi-vowel. A vowel takes the final position of the syllable. There are quite a number of words with such syllables like: (12)
/mw zi/
moon/month
(C½VV$CV)
/ afja/
sneeze
(CV$C½VV)
/bw ka/
bark (of a dog) (C½VV$CV)
/upja/
a new/afresh
/pwagu/
a cheat/ a con person (C½VV$CV)
(V$C½VV)
In each of the above forms, one of the two semi-vowels in Kiswahili, /w/ and /j/, is used in the medial position.
Another structure is CC½VV.
It is, however, noted that very few lexical items in
Standard Kiswahili exhibit this structure. In almost all cases, a nasal consonant stands at the initial position of the syllable. It is always followed by another consonant, usually a plosive, then a semi-vowel (which is usually the labio-velar) and the final position is always taken by a vowel. Examples of these syllables are contained in words such as:
(13)
/mbw ha/
fox/jackal
/mbw mbw /
arrogance/ flamboyance (CC½VV $ CC½VV)
/gw
coins, money
(CC½VV $ CVV)
desert
(CV$CC½VV)
/ agwa/
/
(CC½VV $ CV)
72
/gw na/
crocodile
(CC½VV$CV)
/gwamba/
hoe
(CC½VV$CCV)
It is worthy mentioning that though this study proposes some position on pre-nasalized phonemes in Standard Kiswahili in Section 2.2.2, that position is not applied in the analysis in (13) above. This is because such phonemes are not part of the inventory laid in (7), (8) and (9) which guides this study.
Standard Kiswahili also has the vowel syllable (V). Such syllables have a single vowel and have a high prevalence. They occur in any position of Kiswahili words. Examples are:
(14)
/inama/
bend
(V$CV$CV)
/arusi/
wedding
(V$CV$CV)
/amua/
decide
(V$CV$V)
/ua/
kill/fence/flower (V$V)
/dua/
prayer
(CV$V)
/ a/
marry
(V$V)
/u vu/
evil
(V$V$CV)
Also evident in Standard Kiswahili is the consonant syllable (C). Research has shown that only nasals are used in this structure; and even then, not all nasals are used – only the bilabial and alveolar ones. It is also noted that these nasals do not form syllables in all instances but in specific environments and in specific words. At other instances, they 73
have to be together with other phonemes in order to form syllables. They are syllables only at the beginning or middle of a word; never at the end.9 Examples of these syllables include:
(15)
/mt t /
child
/m awi/
witch
(C$CV$CV) (C$CV$CV)
/atamtuma/ he/she will send him/her (V$CV$C$CV$CV) /
a/
/ /mg
r /
tip
(C$CV)
a small vessel
(C$CV$CV)
wa/ a sick person
(C$CV$CC½VV)
Examples in (15) above attest to the fact that Kiswahili language does have single consonant syllables. 10
Another syllable structure is that of a consonant and double vowel (CVV). In essence, the double vowels denote the existence of long vowels in Kiswahili. Although length is not regarded as a basic phonetic property in SPE (Katamba 1989:52), it is nonetheless an integral phonetic property in Kiswahili. It is, actually, the only distinctive feature in some minimal pairs such as in: (16)
/ka:ka:/ palate
and
/kaka/ brother
/da:wa/ a court case and
/dawa/ medication
/mi:k / taboos
/mik / cookingsticks
and
74
It is, however, noted that this structure is not very common in Standard Kiswahili. Usually, when double vowels appear in the language, as will be illustrated in Chapter IV, one of two things happen i.e, one of the vowels is either deleted or hardened. Other words with CVV structure include: (16a) /da: ari/
to endanger
(CVV$CV$CV)
/ma:bara/
laboratory
(CVV$CV$CV)
/ma:imi /
celebrations
(CVV$CV$CV$CV)
/ma:fa/
misfortunes
(CVV$CV)
/sa:/
time/watch
(CVV)
/ga:ga:/
roll
(CVV$CVV)
Standard Kiswahili also has a closed syllable structure. Syllables are said to be closed when they have a consonant taking the final position of the syllable. This structure is not very common in Kiswahili and most other Bantu languages. Research has proved that in Standard Kiswahili, words with closed syllables are usually the borrowed ones; mostly those of Arabic and English origin. Such words include:
(17)
/halma auri/
board
(CVC$CV$CV$V$CV)
Arabic
/hamsini/
fifty
(CVC$CV$CV)
Arabic
/imla/
dictation/dictatorship
(VC$CV)
Arabic
/akraba/
relatives
(VC$CV$CV)
Arabic
/dafra /
collision
(CVC$CV$V)
Arabic
/alfab ti/
alphabet
(VC$CV$CV$CV)
English
/daktari/
doctor
(CVC$CV$CV)
English
75
All the items in (17) have foreign roots and entered Kiswahili as borrowed words. The phonology of these and other borrowed words will be discussed in Chapter V. It is, however, noteworthy that Massamba et al (2004) do not seem to hold the view that closed vowels do exist in Kiswahili. They instead offer a structure that denotes the open syllable structure as in:
(17a) /labda/
perhaps
(CV$CCV) - Massamba et al (CVC$CV) - This research
Given the definition given by this research and the method used in analysing the syllable structures, Massamba‘s description in (17a) does not appear acceptable to this study.
The last syllable structure as stated by Massamba et al also occurs in borrowed words, especially those borrowed from English. Massamba et al (2004: 85/86) captures this structure in words such as: (18)
/skrubu/
screw
(CCCV$CV)
English
/sprigi/
spring
(CCCV$CCV)
English
This structure of a triple consonant cluster followed by a vowel (CCCV) does evidently appear alien to Kiswahili. Infact, one is tempted to isolate the phoneme /s/ as a separate syllable so as to form the structure (C$CCV$CV). To this research, the latter analysis seems more plausible and realistic than the one proposed by Massamba et al (2004).
76
2:3
Some Aspects of Standard Kiswahili Morphology
As stated, the morphophonological theory used in this study takes cognizance of both the phonological and morphological levels of a language. This section brings to the fore key areas of Kiswahili morphology that have a direct bearing to the language‘s morphophonology.
The areas discussed fall into two major categories: nominal
morphology and verbal morphology. These two domains contain a lot of information to which reference will be made repeatedly in discussion of the various sound changes.
As already stated, Kiswahili, as many other Bantu languages, displays concordial agreement, consequently, all other word categories have their forms greatly dependant on the nominal classes. To exemplify this point, the use of an adjective will do. The adjective –zuri (good/nice) is virtually incomplete when its class is unmarked. The marking is done by the use of a prefix which is wholly dependant on the class to which the noun being qualified by the adjective belongs. Consequently, there are the following forms:
(19)
Singular
Plural
Example of Nouns
{m+zuri}
{wa+zuri}
/m+t t /, /wa+t t /
child, children
{m+zuri}
{mi+zuri}
/m+ti/, /mi+ti/
tree, trees
{n+zuri}
{ma+zuri}
/ iwe/, /ma+we/
stone, stones
{ki+zuri}
{vi+zuri}
/ki+ti/, /vi+ti/
seat, seats
{m+zuri}
{n+zuri}
{n+zuri}
{ma+zuri}
/u+limi/, /n+dimi/ / +d b /, /ma+d b /
77
tongue, tongues tin, tins
{m+zuri}
{ma+zuri}
{m+zuri}
{n+zuri}
/u+amuzi/, /ma+amuzi/ decision, decisions /u+kuta/, / +kuta/
wall, walls
By using the class prefix, it is possible to construct even long grammatically correct sentences. Understanding of Kiswahili nouns leads to easier appreciation of such word categories as adjectives and pronouns. Although the dominance of concordial agreement in Kiswahili is appreciated, this study by no means implies that this state of affairs is exclusive. It is worth noting that the use of some lexical items in Kiswahili constructions denote a deviation from this norm. The use of the adjective /tajiri/ (which in some environments is also realized as a noun) will suffice as an example; the correct use of the word in a sentence is: (20)
Mtu tajiri sana (a very rich person) and not
(20a) Mtu mtajiri sana.* This, despite the fact that it has been noted that {m-} is the prefix morph that carries the singular morpheme in class 1 nouns, as in (21) below: (21)
Mtoto mtukutu (a naughty child)
This study‘s basic research has revealed that most of the lexical items that have a zero morph as the class prefix are those loaned from other languages.
However, the
percentage of such words in Kiswahili is not particularly high. Thus, even as the verbs are discussed, it is important to note the primacy of the nouns. The verb forms depend heavily on the nouns which function as subjects to the various verbs. Yet, verbs are key words that must be present in every Standard Kiswahili sentence.
78
2.3.1 Standard Kiswahili Nominal Types Kiswahili language exhibits at least eight types of nouns. The major consideration in evolving this typology is largely functional and thus semantic. The first category consists of the proper nouns.
These are nouns that denote specific things be they people,
countries, lakes, seas, rivers, oceans, mountains or places. One unique characteristc of these nouns is that they are always in the singular form and always start with a capital letter orthographically. Such nouns include:
(22)
/a a/, /nair bi/, /rufi i/ and /u rumani/ i.e. Asha, Nairobi, Rufiji and Germany respectively.
The other type includes the abstract nouns. These are the ―things‖ that exist in human thoughts or feelings only. They have no definite shape and cannot be identified through use of the five senses; thus, they cannot be touched, smelt, seen, tasted nor heard. Examples include words like: [wivu] jealousy, [ur mb ] beauty, [busara] wisdom and [u iga] stupidity.
The collective nouns denote a group or a collection of things. One individual noun represents a group of similar things. Examples of collective nouns are [ [kamati] committee, [bug ] assembly and [g g ] gang.
79
i] army,
Another category is that of common nouns. These are general nouns that do not refer to very specific things. They refer to things that are usually found in abundance. Examples of common nouns include [ vua] rain, [dawati] desk, [kit wa] head and [mt t ] child.
Concrete nouns are the exact opposite of abstract nouns. These are nouns that denote real forms. They are nouns that may be realized by any of the five human senses. Some like [asali] (honey) can be tasted, others like [wimb ] (song) can be heard while [uvund ] (odour) can be smelt and [ ua] (sun) can be seen. [ ama] (meat), can be touched, seen, tasted, smelt but can never be heard. Some of the things described as concrete cannot be heard or seen by use of naked eyes or ears.
However, with the aid of various
technological devices, this may be possible. Such nouns include [virusi] (viruses) and [bakt ria] (bacteria).
Nouns may also be classified as either countable or uncountable. Countable nouns include: [mt t ] (child), [samaki] (fish), [vik mb ] (cups/trophies) and [miti] (trees). Uncountable nouns may be exemplified by words such as [m I] (smoke), [maziwa] (milk), [uga] (flour) and [mat ] (saliva).
A further classification of Kiswahili nouns denote the animate and inanimate nouns. All the nouns in class 1/2, yu/a-wa are animate and include nouns such as: [mtu] (a human being), [samaki] (fish), [
mb ] (cow/cattle) and [vip fu] (blind people). All nouns in
other classes, irrespective of their prefixes, are inanimate.
80
Another classification involves the gerunds, the verbal nouns that appear in class 20 (ku). These are said to be nouns only in specific contexts, otherwise in form, they do appear as verbs and are at times actually used as verbs in different contexts. These are nouns like: [ku
za] (play), [ku
ka ] (laugh) and [kus ma] (study).
2.3.2 Standard Kiswahili Nominal Structures As in other word types in Standard Kiswahili – and indeed in many Bantu languages – native Kiswahili nouns do not exhibit the existence of infixes. However, this structure seems to be changing as indicated in section 2.3.4.2 and exemplified in (80). Consequently, all the morphs that are affixed to native Standard Kiswahili nominal roots are either prefixes or suffixes. Apart from the nouns arising from affixation, there are also nouns that occur as word-roots and which consist of a single/free morph. Such nouns include: (23)
[mama]
mother
[sugura] hare [kalamu] pen [kuku]
hen
[k
Kenya
a]
A second morphological structure consists of nouns that have single prefixes affixed to the nominal roots. The affixed morphs carry different morphemes, for example, some denote the singular or plural morpheme as in:
81
(24)
Noun
Singular Morph
Noun
Plural Morph
/ki+ ik / spoon
{ki-}
/vi+ ik /
spoons
{vi-}
/ i+w / stone
{ i}
/ma+w /
stones
{ma-}
/m+ti/ tree
{m-}
/mi+ti/
/m+t t / child
{m-}
/wa+t t / children
trees
{mi-} {wa-}
Each of the nouns in the above gloss has two morphs; the root and the prefix. Other nominal prefixes denote the augmentative and dimunitive morphemes.
T he
augmentative morpheme in Standard Kiswahili is seemingly represented by the allomorphs {Ji-} and { }. An exemplification is as follows:
(25)
Noun
Augmentative
/ki+su/ knife
/ i+su/
a big knife
/m+tu/ person
/ i +tu/
a mammoth person
/ +meza/ table
/ i+m za/
a big table
/m+t t / child
/ +t t /
a big child
/m+buzi/ goat
/ +buzi/
a big goat
In this gloss, the nouns comprise of two morphs; one that carries the augmentative morpheme and the other that carries the lexical morpheme. There are, however, some nouns that exhibit an interesting phenomenon. Such a noun is [ umba] house: (26)
/ +umba/ house → / i+umba/ → [ +umba] a big house
82
It so happens that the vowel /i/ is deleted and therefore the surface realization of the augmentative form is / /. This issue will be revisited comprehensively in section 4:3.
Another structure, though rarely used, involves nouns that denote unusually big objects. In such cases, three morphs are realized since a second prefix (/ i/) is affixed as in the following examples:
(27)
Noun
Augmentative Form
Unusually Big
/ +m za/ table
/ i+m za/
/ i+ i+m za/
/ki+tabu/ book
/ i+tabu/
/ i+/ i+tabu/
/m+ti/
/ i+ti/
/ i+ i+ti/
tree
It is also noted that there is an inclination to use double nominal prefixes in the realization of the ‗unusually big‘ category. Words such as /m+ i+m za/ and /m+/ i+tabu/ clearly imply a higher degree of augmentation in contradistinction to / i+m za/ and / i+tabu/.
On the other hand, nouns that denote the dimunitive form usually have three morphs; {ki} which carries the dimunitive morpheme, {- i-} which is also realized as {- -} and the nominal root, thus: (28)
Noun
Dimunitive Form
/ki+tabu/
book
/ki+ i+tabu/
/ki+atu/
shoe
/ki+ i+atu/
83
/m+tu/
person
/ki+ i+tu/
/m+t + t / child
/ki+ i + t t /; /ki+ + t t /
/ i+n /
/ki+ + in /
tooth
Habwe & Karanja (2004) observe that the diminutive form can also be used to infer a pejorative connotation.
The other nominal structure is realized as a result of nominalization. Standard Kiswahili forms many nouns through nominalization of verbs, adjectives and even other nouns. Many abstract nouns are formed through nominalization of concrete nouns [Mgullu (1999)]. The structure of nominalisied nouns include the nominalization prefixes {mu -} and {wa-}; {ki-} and {vi-}; {m-}; {u-}; {ma-} and { }, the nominal root and the nominalization suffixes.
However, although many nominalised nouns possess the
nominalization suffix, some, especially the abstract ones, do not; as is the case in the gloss in (29) below: (29)
/u+g n wa/
ailment
/u+ tani/
devilish behaviour
/u+ta iri/
wealth
/u+z /
old age
In the examples in (30), the nouns have the prefix, root and the nominalization suffix. (30)
/m+
k+a+ i/ he who laughs
/ki+lim+ /
farming
/ +vami+zi/ invader
84
/mu+und+ /
structure
/+tamk+ /
pronuncement/proclamation
/u+ agu+zi/
election
/u+r mb+ /
beauty
These few examples of nominalization give a glimpse to the structure of nominalised nouns.
Such types of nouns are very prone to sound change.
Dealt with in this
synchronic and surface-realised form, the structure of Kiswahili nouns appear fairly simple; however when phonological process occur, or are put into consideration, the structure may appear a bit blurred. It is these processes that form the core of this study.
2.3.3
Standard Kiswahili Nominal Classes
One aspect of the nominal morphology that is of much interest to this study is the understanding of Kiswahili nominal classes. Three major approaches have been used to classify Kiswahili nouns. Each of the classification systems has its own strengths and weaknesses. These approaches are: i)
The morphological/ structural classification
ii)
The syntactic classification
iii)
The semantic classification.
The first two methods are usually supplemented by use of numerals. This study proposes a fourth approach; the eclectic classification of nominals. Each of these approaches is expounded below:
85
2.3.3.1 The Morphological Classification For a long time, this has been the preferred mode of classification by Kiswahili linguists. It is the approach with the longest history since the days of Wilhelm Bleek (1869). It is he who wrote the ground-breaking work: Comparative Grammar of South African Languages then. In that work, Bleek used the initial noun prefixes to classify Kiswahili nouns. Since then, this method has been used by many other Kiswahili Scholars. Bleek numbered the classes he identified and the singular and plural forms were placed in different classes. This is the same approach used by Polome (1967) who identified 16 classes; Mbaabu (1978) also identified 16 classes11 . Kihore et al (2001), Mohammed (2001) and Habwe & Karanja (2004)12 have also in one way or the other used the same approach and identified 16 classes. The 16 classes are usually identified as: (31) 1
M-
2
Wa-
3
M-
4
Mi-
5
Ji-
6
Ma-
7
Ki-
8
Vi-
9
N-
10
N-
86
11/14 U15
Ku-
16
Pa-
17
Ku-
18
Mu-
Classes 12 {ka-} and 13 {tu-} are never used in Standard Kiswahili although they are functional in many other Bantu 13 languages. Equivalents in Kiswahili would be: (32)
CLASS 12. /ka+m+tu/ a tiny person*
(32a) CLASS 13. /tu+m+tu/ tiny people* The two classes denote the dimunitive form. Class 13 has nouns that are plurals of class 12 nouns. Habwe & Karanja (2004) observe that the two forms, though not standard, are widely used in non-formal Kiswahili discourse.
Mbaabu (1992)
suggests that these classes once existed in Kiswahili but were deleted. He however notes that due to an increase in the number of non-native speakers of Kiswahili, usage is slowly creeping back due to the influence of other (mostly Bantu) languages.
A slightly different approach, but which follow the same principle was pioneered by Steer (1870) and Krapf (1882). Linguists in this school also used initial prefixes but instead of designating the singular and plural forms as different classes, they classified them as pairs. While Steer (1870) identified 9 nominal classes, Krapf (1882) identified 8, Burt (1917) 10, Broomfield (1931) 8, Loogman (1965) 8, Brain (1966) 6, Mgullu (1999) 9, Mohammed (2001) 8 and Kihore et al (2001) 12 this classification, nouns are grouped as follows. 15
87
14
. In
(33) 1. Mu-Wa 2. M-Mi 3. Ji-Ma 4. Ki-Vi 5. U-N 6. U-Ma 7. U- 8. -Ma 9. - It is noted that in the morphological classification, the nominal root and suffix are not considered. In addition, it is observed that the dimunitive and augmentative forms are never used to form the basis of classification.
This mode of classification has some obvious weaknesses. One of them is that not all nouns have prefixes [as in the case of /kuku/ (chicken), /kalamu/ (pen/pens) for example] and as a result, in several cases, the zero morph is used. This can cau se ambiguity. It is also evident that quite a number of classes have similar prefixes (as in classes 5, 6 and 7 above) and this, especially to a person who is not well versed with the language, may be a major problem.
However, the major weakness lies in the fact that some of the prefixes, looked at face value, are misleading as they do not agree with their concordial markers in some words.
88
For example, the words /ki+ ana/ (youth) and /vi+ ana/ (youths) may seem to fall in the Ki-Vi class because of their prefixes, but this is a wrong signal. The concordial markers show that the two lexical items belong to the Mu-Wa class, thus: (34)
Kijana aliyepotea amepatikana.
The lost youth has been found.
Vijana waliopotea wamepatikana. The lost youths have been found. Of course it would be a gross grammatical error to construct the sentence in the following form: (35)
Kijana kilichopotea kimepatikana.*
This is so despite the fact that the sentence appear to have concordial agreement given the initial nominal prefix. Such nouns (with seemingly misplaced prefixes) are many in Kiswahili. They include: (36)
/ki+ziwi/
a deaf person
/samaki/
fish
/ki+faru/
rhinocerous
/mi+ugu/
gods
/mi+tum /
prophets, disciples, apostles
/maiti/
corpse
All the nouns in (36) belong to the Mu-Wa class. /ti ag / (a piece of meat), belong to I-, Zi- class; not Ki-, Vi-. It is these shortcomings that prompted Kiswahili linguists to explore other ways of nominal classification.
89
2.3.3.2
The Syntactic Classification
In response to the inconsistences evident in the morphological classification, Kiswahili linguists such as Polome (1967), Nkwera (1978), Kapinga (1983) and Mbaabu (1992) advanced a syntactic system of classification. This system has since been discussed by the linguists who go across the board like Mgullu (1999), Waihiga (1999), Mohammed (2001), Kihore et al (2001) and Habwe & Karanja (2004).
According to Mohammed (2001), this approach analyses nouns by looking at the varieties of the grammatical markers technically known as subject prefixes. Polome (1967) states that these markers include pronoun prefixes and verbal prefixes. One may also add adjectival prefixes. All these prefixes are used in reference to the noun, which noun, governs all the prefixes used in a particular construction. Since these forms are only used for grammatical purposes, and always in sentences or their constituents, this method is seen to have syntactic leanings, hence, the use of the term syntactic classification. As in the earlier classification, here too, there is a difference in numbering. Some linguists like Mgullu (1999) and Nkwera (1978) use the pairing method while others like Kihore et al (2001) give different numerals to the singular and plural classes. In total Kihore et al (2001) have 16 classes, Mohammed (2001) has 18 and if Mgullu‘s (1999) 12 were to be split, they would form 21 classes. However, in all these cases, the principle is the same. These classes may be presented as follows:
90
(37)
Syntactic Classification of Kiswahili Nouns
Number Class 1 u/a wa 2 u i 3 li ya 4 ki vi 5 i zi 6 u 7 u ya 8 u zi 9 ya 10 i 11 ku 12 pa mu ku
Noun mtoto watoto mchungwa michungwa jiwe mawe kijiko vijiko nyumba nyumba uvivu ugonjwa magonjwa uta nyuta maji sukari kucheza mahali mahali mahali
Verb With Prefix a+nalia wa+nalia u+mekatwa i+mekatwa li+meanguka ya+meanguka ki+menunuliwa vi+menunuliwa i+mejengwa zi+mejengwa u+linikasirisha u+naoua ya+nayoua u+nakosa zi+nakosa ya+meletwa i+menunuliwa ku+naudhi pa+napovutia m+namovutia ku+nakovutia
Gloss The child is crying The children are crying The orange tree has been cut The orange trees have bee cut The stone has fallen (down) The stones have fallen (down) A spoon has been bought Spoons have been bought A house has been built Houses have been built (His) laziness annoyed me An ailment that kills Ailments that kill A bow is missing Bows are missing The water has been brought The sugar has been bought (His) game annoys me An attractive place (definite) An attractive place (within) An attractive place (indefinite)
The major weakness of this approach seems to emanate from the fact that the classification is done using subject prefixes of the pronouns, adjectives and verbs and not the nouns which are being classified. Furthermore, the approach is not able to account for defiant forms which deviate from concordial agreement such as: Wake zake
his wives
Waume zao
their husbands
However, that aside, as Mohammed (2001) observes, in comparison, this mode of classification appears more constructive than that used in traditional grammars. He further observes that this method which lays its emphasis on function rather than form
91
has simplified the process of grouping nouns.
Many shortcomings have been
overcome through it, for example: 1
All human and animal nouns can be placed in the same group instead of them being dispersed into different groups as was the case with traditional grammars.
2
Nouns have been properly grouped. In most traditional grammars as earlier demonstrated, nouns like {ma+ovu} (evil deeds), and {ma+gonjwa} (ailments) were placed under U-class even though their adjectives, pronouns or verbs did not bear the prefix U- in in the plural form. 16
3
The problem of class markers being different from other concordial markers is also overcome by use of this method.
It is thus obvious that this approach offers a better and logical exposition of Kiswahili nominal classes.
2.3.3.3 The Semantic Classfication The semantic classification also came into being as a response to the shortcomings evident in the traditional morphological classification.
What the proponents of this
approach did was to analyse the meanings of the various nouns in a given nominal class and then posited those meanings under the concerned prefixes. Consequently, for example, they would state that Mu-Wa class consists of nouns which denote living beings
92
like people and animals. In this classification, all the nouns depicting living beings, irrespective of their defining prefix, fell under one class. Such words include {samaki} (fish), {ki+pofu} (a blind person), {tajiri} (a wealthy person), {nyuki} (bee) among many others. One of the earlier proponents of this approach was Ashton (1944:10) who states that: Each class is associated with one or more underlying ideas.Thus, nouns with M-Wa as their distinguishing prefixes for singular and plural respectively express the names of human beings …Words in other classes are associated with more than one underlying meaning.
Others with similar views include Loogman (1965) and Mohammed (2001). However, some like Mgullu (1999) dismiss this approach saying that it cannot comprehensively tackle the issue of Kiswahili nominal classes. He observes that it is almost impossible to place all the nouns of a given language into specific semantic fields and groups. In fact, this same problem had been appreciated earlier by even the staunchest proponents of the approach. Ashton (1944:10), for example, cautions that:
It must not, however, be imagined that nouns are marshaled into their classes strictly in accordance with these ideas. Consequently, in all classes, nouns will be found which do not conform to the general tendency of the class concerned.
Similar fears have been raised by Kihore et al (2001) who states that due to the changing nature of language, (lexical items included), it is almost impossible to use this approach
93
without the danger of causing ambiguity. For this reason, Kihore et al (2001) suggest that highly technical expertise would be needed were this approach to be used. Mberia (1993) is faced by this problem head-on when he attempts a classification of Kitharaka nouns. He is unable to classify some three nouns which he labels ―problematic‖ and he eventually leaves the issue open for further investigation (:72). This study, however, is of the view that it is possible to lay general semantic fields for each of the nominal classes although this might not be exhaustive. Mohammed (2001), for example, is able to do it to a fairly good degree.
2.3.3.4
An Eclectic Classification
On matters concerning nominal classification, this is a new terminology which has been found necessary by this study; the idea, though, is not quite original. This is a kind of classification that combines more than one of the above approaches. Mohamed (2001), for example, uses both the syntactic and semantic approaches. 17 Kihore et al (2001) have also tried a combination of approaches.
This study is of the view that none of the afore-explained approaches has the ability to tackle the question of Kiswahili nominal classes without raising some pertinent questions.They do however complement each other. Consequently, the current study proposes a sort of classification where all the approaches are combined. The different approaches are viewed as complementary where each one of them fills in the gaps left by the others. This is the approach being referred to as eclectic classification. Here below, this study attempts an eclectic approach to the classification of Kiswahili nouns. For
94
convenience, and ease of reference, the approach using numerals will be applied together with the other three. It is these numerals that have been referred to in indicating the various classes in latter chapters. It was indicated in section 2:5:3:1 that Bleek (1869) pioneered the work of classifying proto-Bantu nominals. His approach was used by many other Bantuists like Jacottett (1896), Meinhof et al (1932) and Guthrie (1967; 70/71). It is not all the classes identified in proto-Bantu that appear in each given language. They vary from one language to the other; and at times, from one researcher to another.
This study, in a somewhat new approach, postulates 21 nominal classes in Kiswahili. They range from class 1 to 23. These classes are partially based on the proto-Bantu classification.
Classes 12/13 are omitted since they are not evident in Standard
Kiswahili. It is, however, emphasized that the numbering in the current study is slightly different from that of earlier works. This has been found necessary in order to cater adequately for the needs of the various approaches used and hence avoid ambiguity. The twenty one classes identified are:
95
(38)
An Eclectic Classification of Kiswahili Nouns
CLASS NUMERALS
CLASS PREFIXES
EXAMPLES OF NOUNS
GLOSS
SYNTACTIC/ MORPHOLOGICAL
1-2
yu/a-wa
m-wa
3-4
u-i
m-mi
5-6
li-ya
ji-ma
7-8
ki-vi
ki-vi
9-10
i-zi
-
11
u
u
11-14
u-ya
u-ma
11-15
u-zi
u-n
11-16
u-zi
u-
17-14
li-ya
18
ya
-ma
19
i
20
Ku
ku
21-22-23
Pa-ku-mu
pa-ku-mu
Mtoto Watoto Mti Miti Jiwe Mawe Kiti Viti Nyumba Nyumba Ulevi* Uvivu Ugonjwa Magojwa Ufa Nyufa Ukuta Kuta Debe Madebe Maji* Maziwa Sukari* Chumvi Kucheza* kuimba Mahali -hapa* -huku -humu
A child Children A tree Trees A stone Stones A seat Seats A house Houses Drunkenness Laziness Ailment Ailments Crack/cleft Cracks/clefts Wall Walls Tin Tins water milk sugar salt his/her game his/her singing place-definite -indefinite - within
*Subsequent words do not denote plural but are words from same nominal class (safe for 21-22-23).
So far, the above nominal classes have numerals, syntactic and morphological prefixes. The next level is to lay down some semantic field to each of these classes. As mentioned in the last sub-section, this is not an easy task given the changing nature of language. Below is a postulation of the eclectic classification of Kiswahili nominal classes: 96
1-2, yu/a-wa; M-Wa classes These classes consist of nouns that refer to all living things. 18 These include both human beings and animals.Also included in this class are things that are taken ipso facto as living like /malaika/ (angel/angels), /tani/ (the devil) and /ibilisi/ (the devil). Even when the nominal prefixes do not match the ones indicated above, the syntactic prefixes may offer the guide.
When that fails, the semantic field ―living things‖ then becomes
important. These nouns include: Class 1 In the singular forms, there are items such as: (39)
/mu+t t /→ [mt t ]
child
/mu+alimu/→[mwalimu]
teacher
/mu+ ama/→/m ama]
animal
/mu+linzi/→[mlinzi]
guard
/mu+tu/→[mtu]
person
It is evident that in the deep structure of most of these nouns, the nominal prefix is {mu -} but it undergoes various morphophonological changes19 and consequently is realized as any of the allomorphs {mu-}, {m-} or {mw-}. A further observation is that the use of the syntactic and semantic approaches allows the inclusion of nouns with different prefixes such as: (39a) /ki+p fu/→[kip fu]
a blind person
/ki+ziwi/→[kiziwi]
a deaf person
/+tajiri/→[tajiri]
a wealthy person
97
/+samaki/→[samaki]
fish [one]
/+ mb /→[
a cow
mb ]
Class II The nouns in this class are the plural forms of class 1 nouns and they include words like: (40)
/wa+t t /→[wa t t ]
children
/wa+alimu/→ [walimu]
teachers
/wa+ ama/→[wa ama]
animals
/wa+linzi] →[walinzi]
guards
/wa+tu/→ [watu]
people
Others are: (40a) /vi+p fu/→[vi p fu]
blind people
/vi+ziwi/ →[viziwi]
deaf people
/ma+tajiri] →[matajiri]
rich people
/+samaki/→ [samaki]
fish (many)
/+ mb /→ [ mb ]
cows
3-4, u-i; M-Mi Classes Nouns in these two classes are drawn from many different semantic fields. Most of them involve inanimate things but some refer to body parts of animate things, especially the parts that are slender and long. Others refer to natural phenomena. However, the bulk of nouns in these classes denote trees and other plants as well as their products.
98
Class 3 This class, also, has three allormorphs as the prefixes denoting the singular morpheme and they are: {mu-}, {mw-} and {m-}. (41)
/mu+ ugwa/→ [m ugwa]
an orange tree
/mu+kuki/→ [mkuki]
a spear
/mu+guu/→[mguu]
a leg
/mu+ili/→[mwili]
the body
/mu+ zi] → [mw zi]
month/moon
/mu+lima] → [mlima]
mountain
/mu+ik /→ [mwik ]
cooking stick
Class 4 Nouns in this class denote the plural forms of class 3 nouns, thus, (42)
/mi+ ugwa/→[mi ugwa]
orange trees
/mi+kuki/→ [mikuki]
spears
/mi+guu/→ [miguu]
legs
/mi+ili/→[miili]
bodies
/mi+ zi/→ [mj zi]
months/moons
/mi+lima/→ [ milima]
mountains
/mi+ik /→ [miik ]
cooking sticks
5-6, li-ya; Ji-Ma classes The nominal prefix {ji-} does not in any way give a glimpse to the nouns found in class 5. Research has shown that a good number of them have {} and also the allormorph {j}
99
(as is the case with / in /, tooth). At the semantic level, nouns in these classes fall under diverse semantic fields, for example; Class 5 (43)
(a) There are those nouns that denote body parts like: / i+
/→ [ i
]
eye
/ i+in /→[ in ]
tooth
/+siki / → [siki ]
ear
/+b ga/→ [b ga]
shoulder
/+g ti/→ [g ti]
knee
Linguists have found out that these are parts that normally occur in pairs. (b) Other nouns denote the augmentative forms of nouns that appear in other classes like: / i+tu/ → [ itu]
(mtu) a hefty person, a giant
/ i+bwa/→ [ ibwa]
(mbwa) a bulky dog
/ i+ti/→ [ iti]
(mti)
a gigantic tree
(c) Other nouns refer to parts of trees and plants together with their fruits and off-shoots like: /+ ugwa/
→[ ugwa]
an orange
/+ua/
→ [ua]
a flower
/+ ani/
→ [ ani]
a leaf
/+ ina/
→[ ina]
a stem
100
(d) There are other assorted nouns traversing different semantic fields like: /+tatiz /→[tatiz ]
problem
/+pat /→[pat ]
earning
/+ ati/→ [ ati]
shirt
/ +amb /→ [ amb ] something Class 6 Nouns in class six are basically the plural forms of class five nouns. 20 (44)
(a)
(b)
(c)
/ma+
/→ [ma
]
eyes
/ma+in /→ [m n ]
teeth
/ma+siki /→[masiki ]
ears
/ma+b ga/→ [mab ga]
shoulders
/ma+g ti/→ [mag ti]
knees
/ma+ i+tu/ → [ma itu]
hefty people/ giants 21
/ma+ i+bwa/→[ma ibwa]
bulky dogs
/ma+ i+ti/→[ma iti]
gigantic trees
/ma+ ugwa/→ [ma ugwa] oranges /ma+ua/ → [maua]
flowers
/ma+ ani/→[ma ani]
leaves
/ma+ ina/→ [ma ina]
stems
101
(d)
/ma+tatiz /→[matatiz ]
problems
/ma+pat /→[mapat ]
earnings
/ma+ ati/→ [ma ati]
shirts
/ma+mb /→ [mamb ]
something
As is evident, these two classes have nouns from very many different semantic spheres. 7-8, ki-vi; Ki-Vi Classes In class 7, /k/ is also realized as / / in specific environments. /k/ is realized when the root of the nouns begins with a consonant. When the noun root begins with a vowel, then, /k/ is realized as / /. However, the rule does not seem to be exclusive as there are exceptions as in the case of /ki+atu/→ [kjatu] (shoe) and [vjatu] (shoes).
22
The nouns found in these
classes are drawn from diverse shades of meaning, including: Class 7 (45)
(a) Nouns that refer to objects made of trees and plants such as: /ki+ti/→ [kiti]
a chair
/ki+tanda/→ [kitanda]
a bed
/ki+ ana/→ [ki ana]
a haircomb
/ki+kapu/ → [kikapu]
a basket
(b) All nouns that refer to the various languages like, /ki+swahili/→ [kiswahili]
Swahili
/ki+faransa/→ [ kifaransa]
French
/ki+kikuju/→ [kikikuju]
Kikuyu
/ki+taita/→ [kitaita]
Taita
/ki+ ina/→[ki ina]
Chinese 102
(c) Some other nouns in this class refer to various body parts which include; /ki+un /→ [kjun ]
waist
/ki+d vu/→ [kid vu]
shin
/ki+sigin /→[kisigin ]
heal
/ki+ wa/→ [ki wa]
head
(d) This class also has nouns that denote the demunitive forms of words from different classes like: /k+ iti/→ [ki iti]
a stick from [mti] tree
/ki+limi/→ [kilimi]
uvular from [ulimi] tongue
/ki+t t /→[kit t ]
a baby from [mt t ] baby
/ki+ ibwa/→ [ki ibwa]
puppy from [mbwa] dog
Research has shown that when the root of the noun has just one syllable, the argumentative / i/ is retained
as in [ki iti] (stick) and [ki itu] (a dwarf).
However, when the root has more than one syllable, the morph { i} is deleted as is the case in [kit t ] (a baby) and [kilimi] (uvular). (e) There are other nouns that do not fall clearly under any clearly defined semantic field such as: /ki+fug /→ [kifug ]
button/imprisonment
/ki+li /→ [kili ]
a cry
/ki+ mb /→[
a vessel
mb ]
/ki+ ti/→ [
ti]
certificate
/→[
]
pit latrine/toilet
/ki+
103
/ki+akula/→[ akula]
food
Some of these and other such nouns are formed as a result of nominalization of verbs while others are loan words from different languages. Class 8 Nouns that form this class denote the plural forms of class 7 nouns, thus: (46)
(a)
/ vi+ti/→ [viti]
chairs/seats
/vi+tanda/→ [vitanda]
beds
/vi+ ana/→ [vi ana]
haircombs
/vi+kapu/→ [vikapu]
baskets
(b) On languages, although it is theoretically possible to have a plural noun like [viswahili] - maybe to imply the different varieties of Kiswahili (both in positive and pejorative senses) – these plural nouns are hardly ever used. Languages are, by and large, viewed as proper nouns hence the unfamiliarity when it comes to the use of plural forms.
(c)
/vi+un /→ [vjun ]
waists
/vi+d vu/→ [vid vu]
shins
/vi+sigin /→[ visigin ]
heels
/vi+ wa/ →[vi wa]
(d)
heads
/vi+ iti/→ [vi iti]
sticks
/vi+limi/→ [vilimi]
uvulae
104
/vi+ mb /→ [vj mb ]
vessels
/vi+ ti/→ [vj ti]
certificates
/→ [vy
/vi+
]
latrines/toilets
/vi+akula/→ [vjakula]
foods
9-10, i-zi; - Classes These two classes have nouns from many diverse semantic fields. Traditionally, these were referred to as N-N class. However, it is noted that the nouns which posses the prefix N- are comparatively few. Many of the nouns have {} prefix. It is also noted that unlike in many other classes, in these two the singular noun is phonologically similar to the corresponding plural noun. The distinction between the nouns of the two classes is however evident when one applies the syntactic approach since the verbal, pronoun and adjectival subject prefixes are subject to concordial agreement. Semantic fields evident in these classes include: Class 9 (47)
(a) Nouns that denote products of trees and plants like: / + ndizi/→ [ndizi]
a banana
/ +nazi/→ [nazi]
coconut
/+karafuu/ → [karafuu] clove (b) Other nouns include those which refer to countries and towns like, /k
a/
Kenya
105
/ufaransa/
France
/uganda/
Uganda
/nairobi/
Nairobi
/aru a/
Arusha
(c) Some abstract nouns are also evident in this class and they traverse over different semantic fields. Examples of abstract nouns are: /+fitina/→ [fitina]
discord
/+ uki/→ [ uki]
dislike
/+hasara/→ [hasara]
loss
(d) There are also other assorted nouns from different semantic fields. Most of them are loan words from other languages and they include: /+r di / →
[r di ]
a radio
/+baisk li] → [baisk li]
a bicycle
/+r li/ →
rail
[r li]
Apart from the above semantic fields, some scholars have classified other nouns under this class; especially fluids and liquids like [ ai] (tea) and [supu] (soup). However, for reasons to be stated while discussing class 19 nouns, this study has deemed it fit to place them elsewhere. Class 10 This class bears the plural forms of class 9 nouns. As stated earlier, phonologically, these nouns are similar to those found in their singular opposites in class 9. However, it is noted that the {} in the deep structure will in this instance represent the plural
106
morpheme while it represented the singular morpheme in class 9. The nouns denoting countries and towns may not have any plural forms as they are proper nouns, although, in theory, this is possible.
11A, U-; u/ Class This class has been viewed differently by different Kiswahili morphologists.
Their
conclusions depend on the approaches they take. 29 This study takes a radical departure from the earlier versions. It splits class 11 into different sub-categories marked by letters. This is deemed necessary because of fundamental semantic differences evident among different nouns of the class. It is also done out of the realization that by using an eclectic approach (especially combining the syntactic and semantic approaches), a more viable conclusion may be reached. In the first categorization (11A), it is noted, the class has only one form. However, in various other subsequent categories (such as 11B, 11C) where the class appears, the singular-plural pairing is observed. Class 11A This first category of class11 nouns include abstract nouns, most of which have the prefix U-. A further observation is that most of these nouns are derived from verbs. Examples are: (48)
(a)
/u+l vi/→[u l vi]
drunkenness
/u+vivu/→[uvivu]
laziness
/u+r mb / → [ur mb ]
beauty
/u+kw li/ → [ukw li]
honesty/truth
107
/u+z mb /→ [uz mb ]
idleness
This class also include nouns that denote things that are ground or that are powdery. These are, naturally, uncountable. Examples include: (b)
/ug r / → [ug r ]
snuff
/uga/ → [uga]
flour
/ud g /→ [ud g ]
soil, clay
Other uncountable nouns in this class include some liquids such as: (c)
/u i/ → [u i]
porridge
/umand /→ [umand ]
dew
The phoneme U- in these latter five examples is not a prefix but part of the nominal root. 11B/14
U-ya; U-Ma class
These two classes do not have many nouns. As a matter of fact, the nouns in these classes have often been placed under class 11A. However, since they exhibit similar variables under the syntactic and morphological approaches, this study finds it prudent to put them under these two classes. Class 11B The nouns in this class take U- both as the class prefix and subject prefix. These nouns fall under different semantic fields, however, most of them are abstract as in the examples below: (49)
/u+g wa/→ [ug n wa]
ailment
/u+asi/→ [uasi]
rebellion
/u+g mvi/→ [ug
vi]
108
argument/quarrel
/u+amuzi/ → [uamuzi]/[wamuzi] decision /u+andi i/→[uandi i]
writing
Class 14 This class has nouns that denote the plural forms of class 11B nouns. These nouns have Ma-as the class prefix and ya- as the subject prefix, thus, (50)
/ma+g wa/→[mag n wa]
ailments
/ma+asi/ → [ma:si]
rebellious
/ma+g
11C/15
vi/→ [mag
vi]
arguments/quarrels
/ma+amuzi/ →[ma:muzi]
decisions
/ma+andi i/→[ma:ndi I]
writings
U-Zi; U-N classes
These classes have nouns that belong to diverse semantic shades. It is however noted that the bulk of them refer to things that are long and narrow/slender. The subject prefixes in these classes are U- for the singular form and Zi- for the plural. The singular noun prefix is U- which is also realized as W-. The plural noun prefix is N- which has the allormophs { -}, {m-} and {n-}. Class 11C Examples of these nouns are: (51)
/u+us / → [us ]
face
/u+uta/→[uta]
bow
/u+d vu/→ [ud vu]
beard [single]
/u+ mb /→[w mb ]
razor
109
/u+ba /→[uba ]
board
/u+bavu/→[ubavu]
rib
Class 15 This class has nouns that denote the plural forms of class 11c nouns, thus: (52)
/ni+us / →
[ us ]
faces
/ni+uta/ →
[ uta]
bows
/n+d vu/ →
[nd vu]
beard
/ni + mb / → [
11D/16
mb ]
brades
/n+ba / →
[mba ]
boards
/n+bavu/ →
[mbavu]24
ribs
U-Zi; U- classes
Nouns in these classes refer to things that fall under diverse semantic fields. They refer to things of different shapes and sizes but the bulk refer to things that are long and narrow. In their singular forms, they are very similar to class 11c, but since they exhibit a particular, similar pattern in their plural forms, it has been found necessary to distinguish them as distinct. Class 11D Class 11D nouns take u- both as the subject and noun prefix. Examples include: (53)
/u+k p /→ [uk p ]
eyelash
/u+kurasa/→ [ukurasa]
page
/u+ku a/→ [uku a]
fingernail
/u+kuta/→ [ukuta]
wall
110
/u+k
/ →[uk :]
clan
/u+t p ]→ [ut p ]
ribbon
Class 16 Nouns in this class denote the plural of class 11D nouns. These nouns have zi- as the subject prefix and {} as the noun prefix. Consequently, the plural forms of the above nouns are: (54)
/+k p /→[ k p ]
eyelashes
/+kurasa/→[kurasa]
pages
/+ku a/ → [ku a]
fingernails
/+kuta/ → [kuta]
walls
/ → [k : ]
clans
/+k
/+t p ] → [t p ]
ribbons
17/14 la-ya; -ma classes Most of the nouns in these classes denote apparatus or equipment made by human beings for their own use. Class 17 Nouns in this class have la- as the subject prefix and {}as the noun prefix. Examples include: (55)
/+ mb / → [ mb ]
hoe
/+ka a/→[ka a]
box, chest
/+d b /→ [d b ]
tin can
111
Class 14 This class has nouns that denote the plural of class 17 nouns. These nouns have ya - as the subject prefix and ma- as the nouns prefix, thus: (56)
/ma+ mb /→[ ma mb ]
hoes
/ma+ ka a/→ [maka a]
boxes/chests
/ma+ d b /→[mad b ]
tin cans
18 ya-; {} Class Nouns in this class are referred to by Mberia (1993:55) as pluralia tantum. According to him, these nouns are always perceived to be in plural form. Further, they are both concrete and uncountable. They have ya- as the subject prefix and {} as the noun prefix. The syllable ma- found in class 18 Kiswahili nouns is not a prefix but part of the nominal root. Examples include: (57)
/mat/
saliva
/ma i/
water
/mafuta/
oil
/maziwa/
milk
As Mgullu (1999) rightly observes, these nouns have neither the singular nor the plural prefix. The nouns consist of only one morph. Further observations are that these nouns appear to fit in class 6 above. However, due to semantic considerations, it has been found fit to place them in a different class. These nouns have a specific, definable semantic field and are not plural forms of class 5 nouns as with other class 6 nouns. Most of these
112
nouns denote liquid forms, forms that have a large liquid percentage, those that could easily turn into liquids or those formed from liquids. 19. I-; {} class Like in class 18, nouns in this class do not have singular and plural prefixes. They are single morphs. They have i- as the subject prefix and {} as the nouns prefix. They always denote the plural form. They fall under different semantic categories e.g.: (58)
(a) Those that denote liquids or near liquids like: /s da/
soda (beverage)
/ ai/
tea
/samli/
ghee
/supu/
soup
(b) Things that appear in granule forms like: / u vi/
salt
/sukari/
sugar
(c) Nouns from assorted shades of meaning, and, especially, borrowed ones like: /miwani/
spectacles
/miadi/
appointment
Some Kiswahili morphologists place these nouns under class 4, but for the same reasons as those given in the discussion of class18 nouns; this study has given these nouns a different class. 20. ku-; ku- class: Nouns that fall under this class are verb infinitives. They do actually correspond to the English form of the infinitive. In Kiswahili, some verbs do function as nouns when used
113
in certain contexts. Mberia (1993) observes that such verb infinitives are qualified by adjectives and, moreover, like ordinary nouns, they do have a concordial agreement pattern. Just as in Kitharaka which Mberia discusses, this is also the case in Kiswahili. Examples in Kiswahili include: (59)
/ku+imb+a/→ [kuimba]
his/her singing
/ku+
z+a/→ [ku
za]
his/her playing
/ku+
k+a/→[ ku
ka]
his/her laughing
/ku+lim+a/→[ kulima]
his/her farming
/ku+fund+a/→[ kufunza]
his/her teaching
Although these lexical items look like verbs on the surface, they at times function as nouns when put in specific syntactic context, for example; (60) /ku+fund+a ku+ak ku+na+p nd+ z+a sana/→[kufunza kwak kunap nd za sana] teaching his/hers is admirable very*;
which translates to:
His/her teaching is very admirable.
According to Mohammed (2001), these nouns are also referred to as verbal nouns or gerunds. He further states that since these nouns express the act of doing, of becoming or the state of being, then, they have no singular for plural forms. A more interesting observation by Mohammed, and which seem to have escaped many a Kiswahili morphologists, is that nouns in this class can also be used in the negative form. This is done by inserting the grammatical partical {-to-} between the infinitive ku- and the verb stem,25 thus:
114
(61)
/ku+t +imb+a/ → [kut imba]
failure to sing
/ku+ t +
z+a/ → [ kut
za]
/ku+ t +
k+a/ → [ kut
ka] failure to laugh
failure to play
/ku+ t +lim+a/ → [ kut lima]
failure to dig
/ku+ t +fund+a/ → [kut funza]
failure to teach
In a sentence, therefore, this negation may be exemplified thus: (62)
/ kut funza kwak hakukumfurahi a mwalimu mkuu/ His/her failure to teach did not go down well with the Headteacher.
A further observation is that monosyllabic verbs do not drop the infinitive ku- as is the case with the above examples. Consequently, two infinitive markers are exhibited in such verbs, thus: (63)
Verb Root
Positive
Negative
/-l-/ eat
/kula/ to eat
/kut kula/ failure to eat
/-f-/
/kufa/
/kut kufa/ failure to die
die
/- w-/ drink
21/22/23 Mahali;
to die
/ku wa/ to drink
pa-
ku-
mu-
/kut ku wa/ failure to drink
classes
These three classes indicate place or location. Each of these classes consists, uniquely, of one noun. Although they all denote location, their functions differ.
Class 21 Pa- imply a specific or a definite place, position or location, thus:
115
(64)
/pahali(mahali) hapa panap nd za/ This place (definite) is pleasing.
Pa – appears as the subject prefix and also as the noun prefix though in this latter case it is interchangeable with MaClass 22 Ku- denotes an indefinite or a generalized place. This maybe exemplified thus, (65)
/mahali huku kunap nd za/ This place (indefinite) is pleasing.
Ku- is the subject prefix while Ma- is the noun prefix. Class 23 Mu- denotes area, ―alongness‖; as Mohammed (2001) puts it. For example: (66)
/mu+ahali humu mna kiza/ This place (within) is dark
Mu- is the subject prefix. The nominal prefix is also Mu- but this is at time realized as Mw- or Ma-, thus, /mw+ahali/ and /ma+hali/. The former is realised as a result of glide formation caused by strengthening of /u/ while the later is as a result of deletion of the same vowel /u/.
Remarks In conclusion to this part, it is noted that the use of an eclectic approach offers a comprehensive and broad outlook as regards Kiswahili nouns and their classes. The approach goes a long way in minimizing ambiguity since the various forms of classification complement each other and where one is not clear, the others shed more
116
light. This classification will be very helpful in the latter sections of this study as it will offer an easy and fast reference.
2.3.4
Standard Kiswahili Verbal Morphology
Verbs are key words in every Kiswahili sentence. While all other word types (including nouns) may be deleted or substituted, this is not possible insofar as the verbs are concerned. If verbs, and especially the main verbs, are deleted in a sentence, the sentence automatically loses its meaning. All Kiswahili sentences must, therefore, consist of at least one verb. In fact, there are many one-word sentences in Kiswahili. This one word is essentially always a verb to which various affixes are agglutinated to represent the different morphemes that may be necessary. Examples of such ―one-word sentences‖ are: (67)
{ni+ta+ku+pik+i+a} /nitakupikia/ I will cook for you or I will cook on your behalf. {si+ku+
z+a}
/ siku
za/
I did not play.
Like many Bantu languages, Kiswahili is an agglutinating language. As a result, it has many affixes and this is especially evident in the verbs. Consequently, Kiswahili verbs have a relatively complex structure. It is in the process of agglutination that most sound changes occur in Kiswahili words, hence, it has been found appropriate to have an overview of the verbal structure. Sound changes occur in all forms of Kiswahili verbs, be they in their infinitive forms as in: (68)
(a)
{ku+imb+a} /kuimba/
to sing; and
{ku+to+imb+a} /kut imba/ failure to sing
117
{ku+chek+a} /ku
ka/
{ku+to+chek+a} /kut
to laugh; and ka/ failure to laugh
Or in their imperative forms as in: (b)
Toka! {tok+a} /t ka/
get out!
Tokeni! {tok+e+ni} / t k ni/
get out (all/both of you)!
Cheza! {chez+a] /
play!
za/
Chezeni! {chez+e+ni} /
and
and
z ni
play (all/both of you)!
Usicheze! {u+si+chez+e} /usi
z / do not play! (singular)
Or even in their subjunctive forms such as:26 (c)
{u+end+e} /u nd /
you go (singular)
{mu+end+e} / mw nd /
go (both/all of you)
{u+pik+e} /upik /
you cook (singular)
{mu+pik+e] /mpik /
cook (both/all of you)
{u+si+pik+e} /usipik /
do not cook (singular)27
In this section, two aspects of the Kiswahili verb will be considered; they are: 1. The verbal types and 2. The verbal structure Understanding of these two aspects, and especially the latter, will be very useful when analyzing sound change in the next chapters.
118
2.3.4.1 Standard Kiswahili Verbal Types Traditionally, at least according to Kihore et al (2001), three verbal types have been identified in Standard Kiswahili. These verbal types are; the main verbs, the auxiliary verbs and the copular verbs.28 However, this approach has been criticized by some Kiswahili linguists like Kihore et al (2001) on the basis that it is largely functional and structural and therefore it has obvious syntactic leanings. These linguists suggest an approach that lays emphasis on form for a ―better‖ morphological study of verbs. They, for example, suggest that morphologists should just investigate forms such as: (69)
{ambi+a} /ambia/
tell
{on+a}
/ na/
see
{rudi}
/rudi/
come/go back
{pend+a} /p nda/
like/love
and not those that denote syntactic structures like: (70)
{a+li+ni+ambi+a} /aliniambia/
he/she told me
{tu+ta+on+a+na} /tuta nana/
we will see each other (later)
{ha+wa+ta+rudi} /hawatarudi/
they will not come back
{a+na+ni+pend+a]
/ananip nda/
he/she likes/loves me
To them, such verbs are heavy with grammatical information and are, therefore, a forte for the grammarians and syntacticians and not one for morphologists. While this study appreciates the fact that information such as that in (70) above is largely grammatical, it nonetheless finds the proposed approach too restrictive and out of consonance with modern linguistic approaches. Modern trends put a lot of emphasis on the interface between the various levels of language. From the outset (see section on Theoretical
119
Framework), this study has endevoured to use an approach that takes all the levels of a language as complementary and not as isolated entities divorced from each other. A further remark is that morphologists, basically, deal with words - be they lexical, orthographical, phonological, grammatical or otherwise – and the said items in (70) above are, for all intents and purposes, words. They are, therefore, as much a business of morphologists as they are of syntacticians. Consequently, this study takes the approach referred to as traditional by Kihore et al (2001).
2.3.4.1.1 Main Verbs These are the verbs that carry the lexical meaning in a sentence. They are also referred to as full verbs. According to Mohammed (2001:80), the main verb informs the listener or reader several things, thus: Firstly, we learn … (about) the actions taking place … secondly, a full verb tells us the voice, whether the sentence is an active or passive voice. Finally, we also learn …about the mood of the sentence; whether the sentence is imperative, subjunctive or indicative. (Emphasis added)
These verbs relay the action being taken or that which is done to the noun.
It is
especially this type of Kiswahili verb that displays a complex structure; a structure that often allows for the affixation of numerous morphs to the verb root and consequently providing a rich ground for sound change. Examples of such verbs are: (71)
Root {-pik-}
Main Verb { wa+na+tu+pik+i+a] they are cooking for us, thus:
120
Wale wapishi watatu wanatupikia The three cooks are cooking for us/ on our behalf {-chez-}
{a+ta+ku+chz+e+a} he/she will play on your behalf thus: Usipowahi, Juma atakuchezea If you do not make it, Juma will play on your behalf.
{-ib-}
{u+si+li+ib+e} do not steal it Nakusihi usiliibe gari lile Please, do not steal that vehicle.
The main verbs form a large part of Kiswahili lexical corpus.
2.3.4.1.2 Auxiliary verbs Auxilliary verbs hardly ever exist alone but are almost always accompanied by main verbs. In Kiswahili, these verbs serve several purposes. For example, they do mark the tense like in: (72)
Waumini wanafaa kutenda mema kila wakati Believers should always commit good deeds.
The presence of the morph {-na-} in the auxiliary verb wanafaa in (72) denotes the present continuous tense. In the following example, morph {-li-} denotes past tense: (73)
Alikuwa amekufa tulipofika. He/she was dead by the time we arrived.
Alikuwa is the auxiliary verb in this sentence. Further, auxiliary verbs denote whether a sentence is in the positive or negation form; thus: (74)
Ali ({ha+a+ta+wez+a}) hataweza kushinda uchaguzi. Ali will not win the election.
The negation morph {-ha-} is evidently in the auxiliary verb hataweza.
121
Thirdly, as Mohammed (2001:81) observes, auxiliary verbs help in construction and production of acceptable sentences like: (75)
Abu amewahi kufungwa jela mara nyingi. Abu is a jailbird.
In the absence of the auxiliary verb {-wahi} the construction in (75) above would not be acceptable syntactically. Finally, a different phenomenon is apparent in the following examples:
(76)
Wageni wangali wanafika.
(76a) Wageni wanafika.
The guests are still arriving The guests are arriving
In the above exemplification, the use of the auxiliary verb {-ngali} in (76) denotes a change in meaning vis-à-vis (76a). While the second sentence implies one group of guests arriving simultaneously, the first one suggests that the guests have been trickling in for quite a while. It is, therefore, evident that the auxiliary verb at times has a major semantic role to play in various constructions. This is something that seems to have escaped notice of many Kiswahili linguists whose works have been used in this research.
It is, however, noteworthy that auxiliary verbs do not form a large part of Kiswahili lexical items. Other examples of auxiliary verbs include: (77) /kwi a/
complete/finish 29
/kuwa/
be
/bidi/
necessited
122
/pasa/
supposed to
/wahi/
has ever
/ nda/
go
/w za/
be able to
/taka/
want
/gali/
still
/ku a/
come
/pata/
get
/p nda/
like
2.3.4.1.3 Copular Verbs Like the auxiliary verbs, comparatively, Kiswahili does not have many copular verbs, Mohammed (2001) notes that they have little independent meaning and therefore their main function is to relate other elements of clause structure and moreso subject and complement. Waihiga (1999) comments that such verbs relate various things in their being, behaviour or even environment.
Most of the copular verbs in Kiswahili are bound morphs which, left alone, have little or no meaning at all. Examples of such verbs are: (78)
{ndi-} {ni}
{tu}
{wa} {u} {yu}
{-mo} {-ko} {-na} {-li} and {si-} To exemplify, using a few of them, there are sentences such as: (79)
(a)
Kabwele ndiye mchawi anayeogopwa sana.
123
Kabwele is the most feared witch. (b)
Mimi si mwizi. I am not a thief.
(c)
Wanafunzi wamo shuleni. The students are in school.
It is, however, noted that the copular verbs play a very important syntactic and semantic role in communication in Kiswahili.
2.3.4.2
Standard Kiswahili Verbal Structures
Although it has been noted that Kiswahili verbs can, and do carry many affixed morphs, these morphs may be categorized into only three main parts, which are:
(a)
the root
(b)
the prefixes
(c)
the suffixes
Kiswahili, in its native form, does not have infixes. However, due to borrowing, it is possible that in future there will be a paradigm shift. Some borrowed words, especially those of Arabic origin, display the existence of infixes. Such words are slowly entering Kiswahili especially through poetry. However since most have not entered the realm of Standard Kiswahili, this work notes, but does not analyse them. Such words include include the following nouns (it is noted that there still is no evidence of verbs acquiring infixes):
124
(80)
Singular
Gloss
Plural
Gloss
abdi
a slave
abidi
slaves
afwaji
a regiment (armed)
afuwaji
regiments
mali
wealth
amuali
lots of wealth
nuru
light
anwaru
lights
biladi
town
buldani
towns
mursali
prophet / messenger
mirisali
prophets / messengers
The various prefix and suffix slots may be occupied by any of the different morphs applicable, subject to the intended morpheme. However, not all morph slots are filled in any one verb; it is infact, impossible to have all possible slots in a single verb occupied. This is so especially given that some morphs are in direct opposition and thus cannot appear in the same verb together for example {na-} and {si-} where the former is stative and the latter indicates negation.
2.3.4.2.1 The Verb Root Verb roots are the morphs that carry the lexical meaning of the verbs. If the root of a verb is changed, then, essentially, the basic meaning of that particular verb is altered. It is the unit to which various affixes, with varied meaning and different grammatical information, are added.
Different verb roots display different phonological patterns. However, Polome (1967:83) notes that the basic pattern consists of an initial consonant, a vowel and a final consonant,
125
hence the CVC pattern.30 Once a suffix, usually a vowel, is affixed to the verb root, the preferred syllable structure of CVCV is realized. After such affixation, the resultant form is referred to as the stem or the verbal base. This pattern is evident in roots such as: (81)
{pig-} { pig+a} /piga/
beat
{let-} {let+a}
bring
/l ta/
{fug-} {fug+a} /fuga/
rear (animals)
However, there are many other patterns such as: (82)
{mwag-}
(C½VVC) {mwag+a}
/mwaga/
pour
{fuat-}
(CVVC) {fuata}
/fuata/
follow
{fung-}
(CVCC)
/fuga/
shut
{twang-}(C½VVCC) {twang+a}
/twaga/
thrash
{vu-} (CV)
{vu+a}
/vua/
remove(cloth)
{f-}
( C)
{f+a}
/fa/
die
{o-}
(V)
{o+a}
/ a/
marry
{it-}
(VC)
{it+a}
/ita/
call
{andik-}(VCCVC)
{andik+a}
/andika/
write
{umb-}(VCC)
{umb+a}
/umba/
create
{fung+a}
These patterns; and the list is by no means exhaustive,31 show how flexible the pattern and structure of the Kiswahili verb root is.
2.3.4.2.2 Verbal Prefixes Different Kiswahili scholars have come up with different numbers of prefix slots depending on how they view different morphs. Polome (1967), for example, recognizes 126
five slots, Habwe & Karanja (2004) six slots while Mgullu (1999) recognizes ten. However, it is noted that whatever the number of slots, all necessary morphemes are accounted for. The difference appears to be in emphasis. This study recognizes the following nine slots.
1. The first slot is occupied by the negation morph that is variently referred to as the preinitial. The prefix used in this position is the negative morph {ha-} which has three allomorphs, namely {ha}, {hu-} and {si-}, thus: (83)
(a)
{ha+a+chez+i}→ /ha
zi/
He/she is not playing
(b)
{ha+u+chez+i}→ /hu
zi/
You are not playing
(c)
{si+i+chez+i} → /si
zi/
I am not playing
The morph {ha-} is usually used in reference to the 2 nd and 3rd persons both in singular and plural. {hu-}, which occurs as a result of deletion of vowel /a/, is only used to refer to the 2 nd person singular while {si-} is used to negate in reference to 1 st person singular. However, it is important to note that {ha-} is actually used to negate in all Kiswahili classes, for example: (84) (a) {ha+ki+ku+nunu+li+wa} /hakikununuliwa/
It was not bought (eg. a chair)
(b) {ha+u+ku+pend+e+za}
/haukup nd za/
It did not please (eg. thread)
(c) {ha+u+ku+kat+w+a}
/ haukukatwa/
It was not cut (eg. a tree)
2. The second slot is usually occupied by the initial morph that is also called the concordial prefix. There are two types of prefixes involved here. Firstly, there are the
127
personal concords which represent the subject. In the first person is the morph {ni-} in singular and {tu-} in plural, thus: (85)
(a)
{ni+ta+chez+a} /nita
za/
I will play
{tu+ta+chez+a} /tuta
za/
We will play
In the 2 nd person singular {u-} is realized while {m-} appears in the plural form, hence: (b)
{u+na+chez+a} / una
za/
{m+na+chez+a] /mna
za/
you (singular) are playing you (plural) are playing
{a-} and {wa-} represent 3rd person singular and plural respectively, thus: (c)
{a+li+chez+a} / ali
za/
{wa+li+chez+a} /wali
He/she played
za/ They played
The second type of the initial morph is the class concord. This classification is used in the non-personal classes, for example: (86)
(a)
Mgongo unauma /mg g unauma/ (The) back aches
(b)
Jiwe limeinuliwa / iw lim inuliwa/ (The) stone has been lifted (up).
(c)
Unga umenunuliwa / uga um nunuliwa/
Flour has been bought.
Consequently, different concord morphs are realized in different classes. 32 These initial morphs form the start of a verb that is not in the negative form.
3. The slot of the post-initial morph is occupied by another negation prefix. The prefix is specific and may be used to negate in all nominal classes and in reference to all persons. This is the morph {-si-} as in: (87) {u+si+ni+chez+e+e}
/usini
z :/
Do not play for me; Do not play games on me.
128
{pa+si+chez+e+we}
/pasi
{i+si+chez+e+we}
/isi
z w / Do not play at that place. z w /
{ki+si+chez+e+we} / kisi
Do not play with it (eg. needle)
z w / Do not play with it (eg. shoe, chair)
This study would like to make some distinction. It is true that{-si-} may be used both as a pre-initial prefix and as a post-initial in reference to 1 st person singular. When it appears after the initial prefix, then, it is, essentially, a post initial. As a result, the examples given by Habwe & Karanja (2004:99) as representing post-initial morphs actually represent pre-initial prefixes. Their corpus is: (88)
{si+lal+i}
/silali/
I am not sleeping
{si+tak+i}
/sitaki/
I do not want
{si+p nd+i}
/sip ndi/ I do not like/love
A more accurate representation of post-initials using the same data - albeit with difference in meaning – would be:
(89)
{u+si+lal+ } /usilal /
Do not sleep
{a+si+tak+ }
He/she should not want
/asitak /
{u+si+p nd+ } /usip nd /
Do not love
As observed by Mgullu (1999:186), Kiswahili has only two main negation morphs and these are {si} and {ha}, {hu} is merely an allomorph of {ha}. Further, it is noted that the morph {to}, though not a main negation verb, is used to negate in nominalised verbs such as:
129
(90)
{ku+to+chez+a} /kut {ku+to+lal+a}
za/
(his/her) not playing
/kut lala/
(his/her) not sleeping
{ku+to+pend+a} /kut p nda/
(his/her) not loving
4. The other slot is occupied by the tense marker. According to Mathews (1997), such morphs form a category that indicate the time of an event. Crystal (1991) states that the morph indicates how grammar marks time. Consequently, without going into polemics concerning temporal and tense aspects, it is safe to state that this morph indicates the time when an action took (or did not take) place. As in all other languages, Kiswahili has three tense markers, thus: (91) Past tense has {li} as the positive marker while {ku} indicates negation. Present tense has {na} as the positive marker and {} for negation. Future tense is represented by the morph {ta}. When affixed to the verb roots, the resultant verbs appear thus: (92)
{ni+li+chez+a}
/nili
za/ I played
{u+na+chez+a}
/ una
za/ You are playing
{a+ta+chez+a}
/ali
za/
He/she played
The present tense may also be represented by morphs {} and {a} as in: (93)
{a++chez+a}
/a
{wa++chez+a}
/wa
za/
You/they are playing
/jua
za/
He/she is playing
{yu+a+chez+a}
za/
130
He/she is playing
5. The fifth morph slot, and which is closely related to the tense marker, is occupied by the aspect marker. The prefixes that occupy this slot indicate whether an action has ceased to happen, is continuing, is habitual or whether it happens repeatedly. Consequently, these are the continuous prefixes, perfective prefixes, habitual prefixes and single event prefixes. The perfective markers include {me} and {ja} for example: (94)
{ni+me+chez+a} /nim {ha+ja+chez+a} /ha a
za/ I have (already) played za/
He/she has not played
The continous prefix is usually {na} as in: (95)
{a+na+chez+a} /ana
za/ He/she is playing
{a+na+lim+a} /analima/
He/ she is digging
{hu} usually denote the habitual morpheme, thus, (96)
{hu+chez+a} {hu+lim+a}
/hu
za/
/hulima/
plays digs
Another aspect that is usually marked pertains to that which happens once eg. (97)
(alianguka) {a+ka+f+a} /akafa/ (He/she fell and) died
{ka} implies that the action took place once and was complete. Another example is: (98)
{a+ka+pik+a} /akapika/ (he/she) cooked
The last aspect to be exemplified here is realized when two or more actions follow each other continuously. The prefix morph usually used here is known as the narrative {ka} as in: (99)
Nikafika, nikakoga, nikapika, nikala kisha nikalala. I arrived, had a bathe, cooked, ate and then retired to bed.
131
6. The other prefix slot is occupied by the conditional prefixes. In Kiswahili, there are at least three of them and these are {nge}, {ngali} and {ngeli}. The three of them are allomorphs of the same morpheme. Mohammed (2001:168) points out that: The three variants (of the same morpheme) have the same function, namely, to indicate a hypothetical condition with no reference to a particular time. The three allomorphs may be used thus: (100) Ningekuwa Profesa, nisingekuwa nikisoma sasa. If I were a Professor, I would not be studying presently. The above statement is only relevant as a supposition and hence, the events that arise are quite imaginary. Another example is: (101) Ungalisoma kwa bidii zaidi, ungalikuwa Profesa Had you put more effort in your studies, you would (by now) be a Professor It is obvious that the second part of the statement is wholly dependent on the first. A final example is: (102) Ningelikuwa Profesa, nisingelisoma tena. If I were a Professor, I would not bother studying again. The dependency evident in the above examples give rise to the terminology conditional sentences. The second action is only tenable on condition that the first takes place. Further, these sentences refer to hypothetical or imaginary situations.
132
7. The seventh slot is occupied by the relative particle which is also referred to as coreferential. Polome (1967:110) inadvertently refers to this morph as an ―infix‖. It is, however, noted that since the morph appears before the root, then it is a prefix and not an infix. Usually, the co-referential morph appears immediately after that of time and aspect marker. It refers to the subject and this usually occurs before the verb. In Kiswahili the co-referentials change according to the various nominal classes thus enhancing concordial agreement. In class 1/2 for example, the relative particle morphs are {ye} for singular forms and {o} for plural forms, thus:
(103) (a) (b)
Mtoto {a+li+ye+anguk+a}
the child who fell
Watoto {wa+li+o+anguk+a} the children who fell
In class, 3&4 {o} appears for the singular forms and {yo} for the plural, as in: (104) (a) (b)
Mti {u+li+o+anguk+a}
the tree that fell
Miti {i+li+yo+anguk+a} the trees that fell
In (104) and most other cases, the relative particle is formed by adding the class concord to the referential ―O‖. 8. The relative particles are followed by the object markers. Usually, they occur just before the verb root. These morphs also change according to the nominal classes to which they belong and also according to personal pronouns in the case of those morphs that relate to human beings. The following morphs are used: (105a) 1st person singular {ni}, plural {tu} 2nd person singular {ku}, plural {wa}
133
3rd person singular {m}, plural {wa} thus: (105b) watu {wa+li+o+ni+pig+a} the people who beat me. Watu {wa+li+o+tu+pig+a} the people who beat us. Examples in class 5/6 include: (106) (a) (b)
Jino {li+li+lo+ng‘o+le+wa}
the tooth that was plucked out.
Meno {ya+li+yo+ng‘o+le+wa} the teeth that were plucked out.
In Class 7/8 examples are: (107) (a) (b)
Kiti {ni+li+cho+ki+nunu+a} the seat that I bought Viti { ni+li+vyo+vi+nunu+a} the seats that I bought
9. The last prefix that this study established is the reflexive {ji}. This is used to denote an action that one pertakes on oneself. It is used in both singular and plural forms, as well as in positive and negative statements, thus; (108) {ni+li+ji+chek+a}
/nili i
ka/
I laughed at myself
{tu+li+ji+chek+a}
/tuli i
ka/
We laughed at ourselves
{si+ku+ji+chek+a}
/siku i
ka/
{ha+tu+ku+ji+chek+a} /hatuku i
ka/
I did not laugh at myself We did not laugh at Ourselves
This study acknowledges that there may be more verb prefix slots in the language but these nine appear to be the main ones.
134
2.3.4.2.3 Verbal Suffixes Broadly, this study has identified three main verbal suffixes in standard Kiswahili and these are: 1. Verbal extension suffixes. These are many and they depend on the voice in use. Examples include: (109) {pik+i+a}
/pikia/
{kom+esh+a} /k m a/
cook for stop/end (a happening)
{pig+ish+a}
/pigi a/
cause to be beaten
{ogo+fy+a}
/ g fja/
frighten
{nunu+li+wa
/nunuliwa/
be bought
The verbal extensions are very flexible and change regularly depending on the aim of the speaker. It is, however, important to note that though some of the verbs with certain extensions, especially the applicative {i}, may appear morphologically correct, semantically they may be anomalous. This is because some actions are neither fathomable nor possible. Such would include:
(109a) {talik+i+a}
/talikia/
divorce on behalf of
{on+e+a}
/ n a/
see on behalf of
{chek+e+a}
/t k a/
laugh on behalf of
{furah+i+a}
/furahia/
enjoy on behalf of
{shib+i+a}
/ ibia/
be satisfied on behalf of (i.e. food)
135
2. The final morph. These morphs mark the end of a word. Verbs of Bantu origin carry the morph {a} as is evident in words such as: (110)
{end+a}
/ nda/ go
{pik+a}
/pika/
cook
{umb+a}
/umba/
create
There, however, appear to be exceptions like /tiri/ (run, flee) but these are too few. The same morph is also used in negative verbs that appear in past and future tenses, thus: (111) {si+ta+lal+a}
/sitalala/
{si+ta+chez+a}
/sita
I will not sleep
za/ I will not play
{hu+ku+som+a} /hukus ma/ You did not read/study {ha+ku+ib+a}
/hakuiba/ He/she did not steal
Morph {i} is used in negative verbs that are in the present tense as in: (112) {ha+lal+i}
/halali/
He/she is not sleeping
{hu+imb+i}
/huimbi/ You are not singing
{si+chek+i}
/si
ki/
I am not laughing
{e} is usually used in imperatives as well as in polite requests, thus; (113) {u+si+som+e}! /usis m / do not read! {a+si+lip+e}!
/asilip /
He/she should not pay!
{u+si+chez+e}
/usi
Do not play (please)
{m+pat+i+e}
/mpatj /
z /
(please), give it to him/her
It is, however, noted that most of the views pointed out here concerning the final morph are not always applicable vis-à-vis borrowed vocabulary. Borrowed verbs seem to carry different morphs from the ones discussed above. The main reason for the aforegoing is 136
that when most of these verbs are borrowed, they are already complete roots and therefore, fitting them into Kiswahili lexicon renders a different morphology. Examples of borrowed words that exemplify this deviation – by ignoring the {a} rule are: (114) /safiri/
travel
/sam h / forgive /sahau/
forget
/hitimu/
qualify/pass
/fariki/
die
/ahidi/
promise
3. The post-final morphs. These are morphs that are heavily dependant on the nominal classes of the subjects and do change accordingly. Examples include: (115) Mtu {a+pend+a+ye} /ap ndaj / Miti {i+kat+wa+yo} /ikatwaj / Jiko {li+chom+a+lo}
/li
Mshale{u+u+a+o}
he who likes/loves (trees) that are felled
mal / (a charcoal stove) that burns
/uua /
(an arrow) that kills
The morph {-ni} may also appear as a post-final morph. This is so when it is used in refernce to 2 nd person plural, 1st person plural and 1st person singular when the object reference is plural as in: (115a) {som+e+ni}!
/s m ni/
read!
{nend+e+ni}!
/n nd ni/
go away!
{pik+e+ni}
/pik ni/
(please), cook
{chek+e+ni}
/
(please), laugh
137
k ni/
{ni} is also used in verbs where the subject is in 1 st person singular and the object is in plural as in: (115b) {na+ku+a+lik+e+ni}
/nakualik ni/ i invite you (plural)
{na+ku+omb+e+ni}
/naku mb ni/ i beseech you (plural)
{na+ku+shukur+u+ni}
/nakuukuruni/ I thank you (plural)
{ni} is also a post-final morph when it appears in some verbs denoting 1 st person plural such as: (115c) {na+tu+l+e+ni}
2.4
/natul ni/
let us eat
{na+tu+som+e+ni}
/natus m ni/ let us read
{na+tu+ pik+e+ni}
/natupik ni/
let us cook
Concluding Remarks
This chapter has covered two major aspects of Standard Kiswahili. The first aspect concerns its segmental phonology. The vowels, consonants and semi-vowels of Standard Kiswahili have been analysed and their distinctive features laid down. Ultimately, matrices for all Standard Kiswahili phonemes have been posited. In addition, the structure of Standard Kiswahili syllables is discussed. The second aspect is the morphology of Standard Kiswahili. Two word types have been surveyed – nouns and verbs. This has been done at two levels for both: typology and structure. These two aspects (phonology and morphology) have a direct bearing to sound change according to NGP. 138
NOTES 1. This research will, for purposes of convenience, treat semi-vowels separately in the next sub-section. They will, however, be alluded to here time and again & their distinctive features will be included in the consonant matrices that appear later in this section. 2. Some scholars like Mohamed (2001) and Kihore et al (2001) are of the view that /w/ is also a bilabial but this study treats it as a labio-velar, a view also shared by Mgullu (1999). 3. These do not include the semi vowels both of which are voiced. 4. Of course SPE has given many other features under each class but these are the major ones as regards Standard Kiswahili and are the ones to be applied in this study. It may be remembered that all the features do not apply to all languages all the time. 5. This study has found it necessary to point out this problem early as any conclusions made here will be of great importance when discussing the homorganic nasal assimilation process in Standard Kiswahili. 6. A good example of aspirited allophones is found in English voiceless stops. /p/ for example has [ph] and [p] as its allophones. Whenever /p/ appears as an initial as in [phin], then it‘s aspirited but it is unaspirited in any other position as in [thop] and [spit]. Other phonemes with such allophones in English are /t/ and /k/; it seems Mohammed (2001) was drawing from these examples. 7. As Akmajian et al (2001) state, the issue of whether or not a certain sound occasions a change in meaning is strictly a language-specific one. For example, in Japanese, Korean (and even Kikuyu); the sounds /r/ and /l/ are not distinct. They do not occasion any change in meaning.
They are simply variants of the same
phoneme, usually /r/. 8. Although it is observed that rounding of lips is a vowel feature, it is worth noting that while articulating /w/ the rounding is coupled with some mild friction not evident when one is articulating the rounded vowels.
139
9. Some borrowed words may seem to posses a consonant syllable at the end (e.g. /maalum/ specific) but they have since adapted to Kiswahili phonology. There are rules guiding this adoption as will be illustrated in chapter 5. There are, however some few exceptions to the adoption rule. These include words such as /takriban/ (approximately), /hususan/ (specifically) and /maalan/ (thus). 10. The articulation, and even orthography, of some of the words with this structure has been a matter of debate. Mgullu (1999:74, 79-80) sees it as a contentious issue which leads to ambiguity. He gives examples of words like nzi (fly) which he says is pronounced as /+nzi/ and mbwa (dog) as /m+mbwa/. In his view, as Kiswahili words are pronounced according to their orthography, further standardization should be carried out to right what he sees as an anomally. It is this study‘s contention that this is an area in dire need of further research; however, for the purposes of this study, the current conventional orthography will be maintained. Other examples given are : mba ( a skin disease), mbu (mosquito), mvi (grey hair), nge ( a scorpion) and nje (outside) 11. In 1978, Mbaabu sems to be greatly influenced by the morphological approach but in 1992, he re-considers the whole issue of Kiswahili nominal classes and takes an entirely new approach – the syntactic one. 12. Although Habwe & Karanja (2004) identify 16 classes in Standard Kiswahili, they however observe that classes 12 &13 {ka} and {tu}/ {tw}) are used in other forms of Kiswahili. Infact, Massamba (1986) observes that {ka} and {tu} represent the older forms of the Bantu prefix system but that they are no longer operative in Standard Kiswahili. He, however, adds that of late, the use of {ka-} instead of {ki-} (for class 12) is becoming more and more popular thus indicating a movement towards Bantuisation. 13. Bleek (1869) identified 18 nominal classes for proto-Bantu while Meinhof (1899) identifided 23 classes. This does not mean that each Bantu language must have all these classes, some have fewer. It has been shown, for example, that Kiswahili lacks classes 12 and 13. 14. Of the 12 though, Kihore et al state that some of them are not used in Standard Kiswahili. 140
15. For example, see Mgullu (1999: 153-156) 16. For a broader exposition of these views, reference may be made to Mohammed (2001: 47-51) 17. This is in his exposition on pp 48-51. He expounds the morphological approach separately. 18. The exception to this rule occurs when these living things are referred to in their diminutive or augmentative forms, then; they fall under different classes as will be discussed later in classes 5/6 & 7/8. Another observation is that some inanimate nominals like /maiti/ (the lifeless body of a human being) fall in this category. This is usually caused by para-lingual reasons that are determined by the speech community; in this particular case, it is respect to the dead. 19. These and other sound changes will be discussed in detail in chapters 3 and 4. 20. Most of the earlier researches were of the opinion that some nouns which do not have singular forms like maji (water), mafuta (oil) etc fall under this category. However, due to reasons to be given while expounding class 18 nouns, this study is of the view that these and similar nouns form a different class. 21. The use of prefixes in this augmentative category seems to be problematic. This issue may be looked at in a number of ways. In the first approach, the {ji} that appears to denote singular class prefix seems to be part of the nominal plural root thus: / itu/
{ma+jitu} a gigantic person/an ogre, derived from
{m+tu} a
person / iti/
{ma+ iti} a huge tree, derived from {m+ti} a tree
In this case, it seems as if a complex morph is formed which totally transforms the whole prefix system, thus, {m}+{tu} → {}+{ itu} → {ma}+{ itu}→ majitu Jitu appears to be a complex morph. In the second approach, it may be argued that such words have three morphs, thus, {+ i+tu} and {ma+ i+tu}
141
This second approach appears to be more convincing since a morpheme can be assigned to each of the morphs. 22. This issue will be discussed in detail while expounding on palatalisation as a morphophonological process. It is further noted that in both cases the high vowel /i/ hardens into the palatal glide /j/. 23. See, for example, Ashton (1944), Loogman (1965), Polome (1967), Broomfield (1975), Brain (1966), Gibbe (1983), Mbaabu (1978, 1985), Mgullu (1999) Kihore et al (2001), Mohammed (2001) and Habwe & Karanja (2004) among others. 24. It is noted that most of the nouns in this class will feature prominently in the discussion on homorganic nasal assimilation. 25. Mohammed writes that the inserted -to- is an ―infix‖ but this is a view that this study disagrees with. As has been observed elsewhere in this chapter, the core of native Kiswahili words do not exhibit the existence of infixes in the language in the proper sense of the term. This phenomenon spreads across Bantu languages and is not unique to Kiswahili.
However, this study does agree with the other
observations made concerning negation. 26. Mohammed (2001:75) also refers to the subjunctives as ―polite imperatives‖. It is important to note that these examples are meant to show the various forms of verbs and do not in any way imply existence of sound change in the various verbs. 27. For an in-depth exposition on the different kinds of Kiswahili verbs see Mohammed (ibid: 71ff) 28. See for example Waihiga (1999:21ff), Mohammed (2001:80ff) and Kihore et al (201:131ff) 29. These are largely direct translations and when the verbs are put into context, there are chances of the meaning changing drastically. 30. This view is, however, highy debatable given the large number of existing verbal root patterns. See 31 below. 31. Infact, Mbaabu (1992: 137 ff) has outlined 53 verbal root patterns in Kiswahili verbs. 32. Refer to the table used to describe the syntactic classification of Kiswahili nouns.
142
CHAPTER III Standard Kiswahili Consonant Morphophonology 3.1
Introductory Remarks
In this chapter, this study engages in a comprehensive analysis of consonant changes evident in Standard Kiswahili. In addition to discussing them and giving examples, rules are formulated for each of the processes discussed for easier comprehension.
The
consonant matrix laid in chapter two as well as the eclectic noun classification are used in giving the necessary gloss.
The consonant processes that are discussed include consonant weakening (lenition) and consonant strengthening (fortition) which are shown to be related in terms of linguistic analysis. In addition, palatalization is also discussed as well as homorganic nasal assimilation where both the regressive and progressive assimilation processes are analysed. Consonant deletion is analysed under several parameters: historical, Ganda Law as well as trill and nasal deletions.
However, before starting an investigation of the various sound changes, this study will make some general comments as pertains to the reasons that lead to the realization of phonological processes. In discussing the causes, Sommerstein (1977:254) observes that: There are a great number of factors, which can contribute to precipitating a phonological change, and it is probably rare for a single change to result from a single
143
factor. (Emphasis added) Consequently, when reasons for the emergence of a particular change are stated, it should be noted that the list would by no means be exhaustive.
Sommerstein identifies several factors that are relevant to this study. One of them is abstractness. When the phonology of a language is excessively abstract, rule re-ordering becomes necessary so as to reduce the abstractness and to increase both phonetic and semantic transparency. Another reason is ease of articulation. As will be evident in this study, this is one of the major reasons that trigger sound change. Changes such as deletion and assimilation do ostensibly ease the manner in which various items are articulated.
Some changes (eg. deletion) for example, allow the attainment of the
preferred syllable structure in a given language. This is true of Standard Kiswahili as will be shown later in this chapter as well as in chapters 4 and 5. Assimilation allows for the harmonization of place of articulation, again easing articulation. A third reason given is structural imbalance. Concerning this reason, Sommerstein (1977:255) observes that: Languages … seek on the one hand to exploit their contrasts maximally, and on the other to ensure that contrasting phonemes are perceptually far enough apart to obviate the possibility of confusion. If there is a ‗hole in the pattern‘ a ‗case vide‘, an unused combination of feature values, another phoneme may shift to fill the hole; if, on the other hand, there is an excessive number of phonemes within a given area of phonetic space, one of them may shift out of the area. In either case, such a shift may set up a chain reaction by creating a new imbalance at another point.
144
Sommerstein has been quoted here because his views on structural imbalance capture graphically another major reason for sound change and one which is clearly evident in Standard Kiswahili. This is clearly so in the case of Ganda law and nasal deletion. One other factor discussed is language contact. Where a language comes into contact with another (other) language(s), there is evidence that one language may occasion some influence on the other, (and vice-versa), thereby triggering some sound changes. In Standard Kiswahili, such processes as substitution, insertion and deletion are usually associated with language contact and borrowing.
There are other factors that cause sound change, but the ones mentioned above appear to be the main ones. Quoting Kiparsky, Sommerstein (1977) states that the major causes maybe summarized into three: ease of production, ease of perception and ease of acquisition. Kiparsky, it is noted, further argues that were a potential change, motivated by any of the above factors, to have bad effects in respect to another, the change takes place anyway, but is then followed by further changes which correct the damage done by the first. This ‗corrective change’ may itself occasion further deleterious effects at another point and therefore, it appears, equilibrium may never be achievable.
3.2.0 Preliminary Remarks to Lenition and Fortition In exploring the processes of weakening (lenition) and strengthening (fortition) of Standard Kiswahili phonemes, this study will apply the strength hierarchy as suggested by Hooper (1976). In her model, Hooper uses the manner of articulation parameter. This commonly accepted phonological strength hierarchy appears thus:
145
(116) (a) Voiceless > voiced (b) Stop>affricate>fricative > approximant>zero (Where > indicates a step towards a weaker articulation). Katamba (1989), proposes a single hierarchy to cater for the various articulatory concerns which he posits as: voiceless stop >voiced stop >voiceless affricate> voiced affricate >voiceless continuant >voiced continuant >nasal approximant. For the purposes of this study, the hierarchy maybe stated as follows:
(117) voiceless stop >voiced stop >voiceless affricate> voiced affricate> voiceless fricatives >voiced fricatives >nasal >trill >lateral >glide> vowel.
According to Sommerstein (1977: 228), the two processes depend on environment, he observes: Normally, strengthening and weakening processes are environmentally controlled, and it seems to be possible to single out some environments as conducive to weakening and others as conduce to strengthening. This is definitely the case as will be clear in the exposition below.
3.2
Consonant Weakening (Lenition)
There are several instances when Standard Kiswahili consonants weaken. Most of the weakening involves plosives. For example, as Polome (1967) observes, Kiswahili plosives weaken when they occur before the vowel /i/ during nominalization of verbs of
146
Bantu origin. Initially, all the lexical items that have the suffix /i/ after nominalization carry the suffix /а/ in their verbal forms before nominalisation.
Examples of the
nominalised forms are: (118) /mu+tuang+i/ → [mtwanzi] one who pounds grains from /twaga/ pound /mu+tung+i/ → [mtunzi]
composer
from
/tuga/ compose
/mu+ εng+i/ → [m εnzi]
builder
from
/ ga/
build
where: (118a) /g/→ [z]
/i/
A further example but which is [-nasal] is: (119) /mu+l g+i/ → [ml zi]
witch
from
/l ga/
bewitch
follower
from
/fuata/
follow
Then there is a situation where:
(120) /t/ →[s]
/i/
as in: (120a) /mu+fuat+i/ → [mfuasi]
However, it appears that this rule is not exclusive1. There are times when the plosive does not change irrespective of the environment; a case in point is in the following nominalized form: (121) /mu+tafiti/ → [mtafiti] researcher Several reasons may be advanced to explain this deviation. The first and most appealing is that the lexical item lacks a Bantu origin. The notion of systematic research may have been non-existant amongst the users of earlier forms of Kiswahili and thus the term may be a late entrant into Kiswahili diachronically. But then, since the word appears to have
147
the normal form of Kiswahili/Bantu words, if this explanation were to be valid, then one cannot avoid the suggestion that there is a chance that the term is still evolving and might, in future, have the sound /t/ change to /s/. However, it is important to note that unlike most verbs with Bantu roots in Kiswahili, the verbal base of this item does not end with the sound /a/ but with /i/, thus /mtafiti/ and not /mtafita/*, hence the assertion that the word may, as a matter of fact, lack Bantu roots.
A second argument would be on the arrangement of the various sound segments in the word where the existence of a plosive followed by a fricative militates against the next consonant being another fricative. However, there is no rule known to this researcher that would help sustain this argument. The last reason is hinged more on sociolinguistics than morphophonology. In Kiswahili language, there exists the word /fisi/ (hyena), an animal that is highly disregarded, consequently, it maybe suggested, the term /mtafisi/* would have a negative connotation, hence the variation. Given other examples 2 in the language, this explanation may not be far-fetched.
Further examples of weakening include: (122) /mu+lap+i/ → [mlafi] glutton /mu+lip+i/ → [mlifi] he who pays/payer In this case,
/p/→[ƒ]
from
/la/
from
/lipa/ pay
/iba/
steal
— /i/
In example (123), /b/→[z] — /i/ (123) /mu+ib+i/ → [mwizi] thief
from
Further examples of sound weakening abound when /k/ →[ ] — /i/, as in3:
148
eat
(124) /mu+lalamik+i/ → [mlаlami i] complainant
from
/lalamika/ complain
from
/tumika/
serve
/mu+andik+i /→ [mwandi i] writer/author
from
/andika/
write
/mu+pik+i/ → [mpi i]
from
/pika/
cook
/mu+tumik+i/ →[mtumi i]
servant
cook (noun)
This study however notes the existence of forms such as: (125) /mu+ irik+i/ → [m iriki] /mu+ itak+i/ → [m itaki]
an associate; a participant accuser
The above two items seem to be similar to the earlier corpus. Two reasons may be advanced for their apparent deviation. The first one revolves around the existence of the phoneme / / in the verb root. If /k/ were to weaken to [ ] after nominalization, the pronunciation would not only be unfamiliar to Kiswahili, but would also be complicated. Avoiding the change, therefore, allows for ease of articulation. The second reason, and one that appears more plausible, is that the verbal base of these words do not have the final prefix as // but as /i/, thus, /stahiki/ (deserve), / iriki/ (participate) and / itaki/ (accuse). As noted earlier, it is the verb roots that end in prefix /a/ that seem to experience this sound change.
The change also involves forms such as: (126) /mu+lind+i/ → [mlinzi]
guard
from
/linda/
guard
/mu+pnd+i/ → [mpnzi] lover
from
/pnda/
love
149
/mu+pand+i /→ [mpanzi] where /d/ → [z]
sower
from /panda/
sow
— /i/
However, there again appears to be exceptions as is evident in: (127) /mu+ ind+i/→ [m indi] winner
from
/ inda/
win
It once more appears as if the existence of the fricative / / in the root inhibits the weakening of /d/ to yet another fricative - /z/. Thus justification is even more plausible when one consinders the verbal base whose final is /a/. The plosive /t/ weakens to fricative /s/ in the following example:
(128) /mu+pat+i/ → [mpasi] he/she who gets wealth easily from /pata/ get
An interesting phenomenon is observed in: (129) /mu+suk+i/ → [msusi] weaver /mu+sak+i/ → [msasi] hunter
from /suka/ weave from /saka/ hunt
In the above items, /k/ weakens to [s] not [ ] as was the case in the examples given earlier. It appears that where the root initial is [s], the same phoneme is reduplicated in the weakening of /k/. The realization of [ ] would complicate articulation.
The last example involves /b/ weakening to [v] as in: (130) /mu+t imb+i/→[mt i vi] /mu+
mbi] → [m
vi]
mischievous person from /t imba/ undermine quarrel some
150
from /g mba/ quarrel
Before positing the rule that governs the processes exemplified above, it is important, at this stage, to comment on the various exceptions that have come out. It is noted that although, generally, sound weakening and strengthening processes fall under the P-rules, in specific cases; as is here, these processes may be discussed under the MP-rules. It is the MP-rules that allow for exceptions since they are non-universal. The examples given fall under the morphophonemic rules as well as the morphological spell-out and word formation rules.4 That is why, under different environments, exceptions occur at times. In formulating the rule that govern weakening in the examples given, we start by postulating as follows:
(131)
k
/s
g t
z →
s
d
z
p
f
b
z/v
— /i/ nominalization
Since all the phonemes on the left are plosives and those on the right are fricatives, this rule may be simplified thus: (131a)
+const. -son.
+ syll →
[+cont]
—
- const
-cont.
+ high
- del rel
- back
151
It is, however, noted that weakening of plosives does not only occur in nominalization. The process is also realised in some verbs as is evident below: (132) /ruk+i+a/ → /ruk+j+a/ → [ru a]
throw
/t εk+i+a/→ /t εk+ j+a/→[t ε a]
make to laugh
/pit+і+a/→ /pit+j+a/→[pi a]
allow to pass
In the above examples, several processes take place. First, the vowel /i/ hardens into a glide, secondly, the glide is deleted at the same time as the plosive weakens into a fricative. It is obvious that the major motivation for this change is ease of articulation. The rule for this weakening may be posited thus:
(133)
k t
→
____ /a/
Which can be further formulated as; (133a) +const. -son.
+ syll → [+cont]
-cont.
+ syll
____ - const
- const
+ high
+ low
- back
Thus: (133b)
+const. - son
+ syll → [+cont]
- cont.
___
- const + low
152
The last process to be discussed in which consonant phoneme weakening is expressed involve the affricate / / which changes to form the fricative /z/. This happens during nominalization and when the final phoneme is vowel /i/. The other process involved during the change is deletion of vowel /a/ whose position precedes that of the concerned affricate. This situation is evident in the following examples:
(134) /m+t afu+a+ +i/ → [mt afuzi] /m+t agu+a+ i/ → [mt aguzi]
trouble maker fastidious person
from /t afua/ cause trouble from /t agua/
choose
/m+g mb +a+ i/ → [mg mb zi] defender/contestant from /g mb a/ contest
In all these cases, the rule is: (135)
/ / → [z]
— /i/
and this can be indicated as in (135a): (135a) +const - son.
→
+syll.
+syll.
- const
- const
+ ant.
+ low
- high
+ cor.
+ mid
- back
-cont.
+ cont
___
– cor.
The rule may finally appear thus:
153
(135b) +const
+syll
- son - cont.
+ cont →
_____
- cor.
- const - high
+ ant.
- back
+ cor.
But there is a second argument concerning the verbal bases of terms such as [mt afuzi], [mt aguzi] and [mg mb zi]. There is a school of thought that suggests that the proto verbal bases are /t afula/, /t agula/ and /g mb la/ respectively. The proponents of this view go further to suggest that there actually is some semantic difference between [mt afuzi] and /mt afua i/ for example. While this argument seems plausible, this study does not wish to engage into polemics concerning the issue. The only observation to be made here is that if the argument holds, then the change of /l/ to [z] is evidently fortition. However, whichever the case, the analysis made in this study concerning lenition remain valid.
In conclusion, it is noted that consonant weakening in Kiswahili involves mostly plosives, both voiced and voiceless. Both plosives and affricates fall under the main category of stops. So, it may be true to state that stops weaken to become fricatives in the given environments. However, formulation of a rule to describe the above statement would lead to problems since this study has not noted any weakening of the voiceless palatal affricate, i.e / /, hence the failure to posit a general rule. This change of stops into
154
fricatives, or what is usually referred to as spirantisation, as has been evident, is very inconsistent. This inconsistency is not only evident in Standard Kiswahili but also in other languages like Luganda (Katamba: 1989).
Consequently, there are different
approaches to the issue.
3.3 Consonant Strengthening (Fortition) The discussion in this section should be seen in the light of the exposition in section 3:2. In fact, it should be viewed as a continuation of the same since the concepts of weakening and strengthening cannot be viewed independent of each other. The terms maybe said to be quite relative and thus measured on the same ‗line‘. Kiswahili language has two liquids and they are /l/ and /r/. Mberia (1993:133) notes that: Both of them (/r/ & /l/) harden into [d] when they are preceded by /n/ … This study fully agrees with the above observation and goes further to give examples which include: (136) /n +limi/ → [ndimi]
tongues
/n+l a/ → [nd a]
marriage
/n+l t+ / → [nd t ]
dream
/n+r fu/ → [nd fu]
long
The two liquids are articulated at the alveolar region as is the plosive /d/ which replaces them. The rule that controls this process seems to prohibit two resonants from occurring in the neighbouring environments. Apart from the nasals, liquids are the only other resonants in Kiswahili; consequently, when they undergo the phonological change, this
155
notion becomes a reality. This process also allows for a second process; homorganic nasal assimilation5. The process may thus be shown as: (137) l r
→
d
/n /
Further, the rule may be posited as follows: (137a) +const + son + cont
→
.
+const.
- son
+ son
- cont
- cont
+ ant
+ ant
+ cor
+ cor
______
A second example of consonant strengthening is when semi-vowel /w/ occurs as plosive /b/. This happens when /w/ is preceded by /n/ which is then realized as /m/ as in the word: (138) /n+wili/ →
/n+bili/ → [mbili]
two
/n+wingu/ → /n+bingu/ → [mbigu] heavens Again, the fact that after the change the nasal and /b/ fall under neighbouring environments allows for the conclusion that it is the second process involved; the nasal assimilation, which triggers this process. The nasal assimilation rule is discussed in section 3.5. The articulation of the alveolar nasal and the semi vowel at neighbouring
156
environment would be problematic thus necessitating the change. It however appears that Kiswahili nasals are quite influential and more often than not affect the neighbouring sounds as is the case here. In rule formulation, it can be stated that: (139) /w/→[b]
/n/ ―
This rule may further be stated as:
(139a) + son
+ const
+const
- son
+ son
- ant
- cont
- cont
- cor
+ ant
+ ant
+ cont.
→
+ cor
It is worth noting, however, that this phenomenon is not unique to Kiswahili. It was discussed by linguists like Schane (1968) and Antilla (1972) both of who see it as a natural rule of the assimilative type. Mberia (1993) discusses it at length and observes that the change of a continuant into a stop after a nasal consonant is a phonetically motivated process. It does, therefore, fall under the P-rules. The aim of the harmonization of the place of articulation (in this case the labial and alveolar) is to simplify articulation. In fact, Mberia (1993:133) observes that: In continuant hardening the continuant consonant assimilates to the feature [-CONTINUANT] of the nasal consonant ….. the closure of the articulators for the production of the nasal consonant is maintained into and during the production of what otherwise would have been articulated as a continuant consonant, thereby causing
157
the later to be a stop … (this is) progressive assimilation. Taking cognizance of the above, then, the general rule would be: (140) +const. +cont
+const → [-cont]
+ son
_____
- cont + ant + cor But there is an interesting phenomenon in Kiswahili where the same semi-vowel /w/ hardens not to a stop but to a voiced labiodental fricative. Examples include: (141) /l wa/ →
/m+l wi/
/m+l wi/ → [ml vi]
be drunk, a drunk a drunk
/l wi+a/ →[l via] → [l vja] cause to get drunk And: (141a) /r wu/ → /m+r wu/
tall
/m+r wu/ → [mr vu]
a tall person
/r wuka/ → [r vuka]
get tall
It appears that in such cases, the semi-vowel does not neighbour a nasal. Secondly, the hardening occurs when the semi-vowel precedes the high vowels /i/ and /u/. Where /a/ comes after it, it is realized just as /w/, thus: u (142) /w/ →[v] — i
This rule can be indicated as follows:
158
(142a) - const
+const
+ son
- son
+syll
+ ant
- const
+ cont
→
- ant
+ high
- cor
However, many words in the language do not exhibit this kind of change. The most prevalent form of phoneme strengthening in Standard Kiswahili however, involves vowels being realized as semi-vowels. Although it is noted that this is also a form of hardening, this process has been discussed in the next chapter, section 4:4, under the subtitle ―glide formation‖.
3.4 Palatalization In discussing this process, this study uses the definition stated by Lass (1984) that palatalization basically signifies the change of a consonant from being a non-palatal to a palatal one.
In Standard Kiswahili, this process is prevalent when the non-palatal
consonant is followed by a double vowel cluster and the first of the two vowels happens to be a high front vowel; mostly /i/. It is notable that when producing the sound /i/, the blade of the tongue is raised so high that it is usually very close to the palate. Consequently, if the consonant preceding the vowel is palatalized, then there is every reason to get some mild form of assimilation where the consonant is assimilated to the vowel.
159
In Standard Kiswahili, the consonant mostly affected is /k/ although a case maybe argued for /n/. This process affects diverse categories of words including nouns, adjectives and verbs. A further observation is that in Standard Kiswahili, palatalization is most pronounced in class 7-8 nouns as well as in the adjectives and verbs that relate to them. Examples include:
(143) /ki+andarua/ → [kjandarua] → [ andarua] mosquito net /ki+akula/ → [kjakula]→ [ akula]
food
/ki+ama/ → [kjama]→ [ ama]
party/association
/ki+
/ → [kj ]→[
]
/ki+ nd / → [kj nd ]→[ /ki+
/ → [kj
]→[
rank nd ]
]
fibre basket restroom
/ki+uma/ → [kjuma]→ [ uma]
iron/metal
/ki+u /→[kju ]→[ u ]
college
/ki+umba/→[kjumba]→[ umba]
a room
Since this study is using a synchronic approach, the interest will only be in the final/telescoped change. Telescoping occurs when some intermediate stages in a sequence of changes are left out of a discussion. Concerning telescoping Katamba (1989:113) observes that: Telescoping occurs when some intermediate stages in a series of natural historical changes get eclipsed or completely lost,
160
leaving behind a phonetically bizarre set of synchronic alternations.
However, it is important to note that there are several intermediate diachronic changes that occur in the case of palatalisation. For example, in the gloss above; it is evident that the vowel /i/ first hardens to become a glide (/j/)6. Secondly, the glide is deleted at the same time as palatalization occurs. These two processes seem to occur simultaneously since it is not possible to handle them differently interval-wise.
While discussing palatalization in Chi-Jomvu7, Bakari (1982:127) states that the original phonetically plausible rule before the front vowels can be stated as, (144) /k/ → [ky] — [i] and that this has telescoped to: (144a) k > Consequently, the rule may be posited thus: c v (144b) [+back]→[-back] — [-back] [except dimunitives]
However, in Standard Kiswahili, as will be discussed later, this rule does not apply as it is indicated above exclusively. There are many forms that are similar to the above and which are not in their dimunitive forms but which do not follow the rule. From the gloss above, it is possible to lay a rule that explains the process as: (145) /k/→[ ]
— /i/, [syll]
thus,
161
(145a) +const - cont - ant - cor - voice
+ cor
+ syll + const + high - back
→
+ syll - const , (not /i/)
In adjectives, palatalisation usually occurs in those adjectives qualifying class 7 nouns and whose roots begin with a vowel. The main cause for this change is because the class prefix is {ki-}, and, as noted, when /k/ is followed by /i/ and another vowel, usually, palatalization occurs, examples include: (146) /ki+ kundu/ → [
kundu]
/ki+ mbamba/→[
red
mbamba] slendar
/ki+a / → [ a ]
theirs
/ki+angu/ → [ agu]
mine
/ki+ak / → ak ]
yours
/ki+
w /→[
w ]
/hi+ki+ / → [hit ]
itself that
In verbs, palatalization occurs when dealing with class 7 nouns. It is usually evident in the morph referred to as the relative particle or the co- referential. Palatalization occurs when class 7 concord {ki-} is affixed to the relativiser {‗o‘}. For example, when talking of a book the change is as follows:
162
(147) (kitabu) {a} {li} {ki} {o} {ki} {nunu} {a} (The book) that he/she bought 1 2 3 4 5 6 7 /aliki kinunua/→ [ali
kinunua]
Here the 3rd and 4th morphs combine and at the same time palatalization occurs where /k/→[ ]. Once again, the two rules that dictate the combining and palatalization are intrinsically ordered and thus take place simultaneously in the speaker‘s mind. A second example is: (148) (kiti) {a} {pend} {a} {ki} {o} (the seat) that he/she likes. 1
2
3
4
5
(148a) /ap ndaki / → [ap nda
]
Morphs 4 and 5 combine and palatalize to form / (149) {ki} +{o} →{cho}, [
/, thus;
]
In these examples, telescoping is done. Other processes also take place in the above gloss. One of them is the hardening of vowel /i/ into glide /j/. The effect of this glide formation deserves some comment since it has a direct bearing to the palatalization process. Now, it has been noted that as far as vowels are concerned, /i/ is more palatal than any other vowel, so, naturally, it hardens into a palatal glide which is more front than back. This palatal glide then influences and assimilates the back (velar) voiceless plosive and thus giving it the feature [+anterior], and at the same time weakening it. These two processes take place simultaneously. The resultant sound, which may be seen as a compromise, is the voiceless palatal affricate.
163
A further observation concerning the palatalization rule of /k/ → [ ] is that there are many lexical items which at the face of them should exhibit the change but they do not. Since these items are many, and involve all vowels, this is an area in need of further research. Examples are:
(150) /ki+ali/
→ [kjali]
/ki+ambatit ] → /kjambatit ]
flame appendix
/ki+ l zi
→ [kj l zi]
adverb
/ki+ l l z
→ [kj l l z ]
illustration
/ki+ ta/
→ [kj ta]
nest
/ki+
→ [kj
mirror/glass
/
]
/ki+un /
→ [kjun ]
waist / loin
/ki+uga]/
→ [kjuga]
suburb
However, this is not purely unexplainable. The change of /i/ into a palatal glide is, to us, a form of palatalization given this research‘s working definition of the term. /j/ is a palatal glide. A second explanation arises from Socio-linguistics. Semantically, were most of these items to acquire [ ] in place of [k]; there would be overlaps as in / / (toilet) and /ki/ (mirror); / uga/ (graze) and /kiuga/ (suburb) as well as / ta/ (draw – like water) and /kita/ (nest). Even if this were not the case, residues of any sound change is a necessary condition in Natural Generative Phonology. This theory dictates that any sound segments that are said to be underlying should surface in some lexical 164
items of the language.
Any underlying form that fails to surface occasionally is
considered to be too abstract by NGP, and thus unacceptable. Therefore, such words may be considered as residues, the only problem is that they are too many for such a consideration.
A further comment is that although in these nominal examples /k/ is not palatalized, when it occurs in adjectives and verbs where the basics of the palatalization rules are apparent, then, /k/ →[ ], for example, (151) /ki+ ta ki+ ma/→[kj ta /ki+
ki+ak / →[kj
ma] a nice nest ak ]
his/her mirror
/ki+un a+li+ki+ +ki+p nd+a/→[kjun ali
kip nda] the waist that he/she liked
/ki+uga ki+a / →[kjuga a ] their suburb
In plural forms of class 7-8 nouns, both /i/ and /j/ are realized as [j], taking/i/ closer to being a palatal sound.
Another comment to be made in regard to palatalization is that the process of /k/ → [ ] is a form of consonant weakening where a plosive is realized as an affricate. On the other hand, the gliding of /i/ to /j/ is a form of phoneme strengthening where a vowel changes into a glide.
165
Lastly, on palatalisation, this study argues a case for the palatalization of the alveolar nasal. According to Meinhof (1968) and Guthrie (1970/71), most nouns in classes 9 and 10 have {n-} as the class prefix. As noted in the case of /ki-/, when /i/ is followed by a vowel, where the alveolar nasal precedes the vowel /i/, palatalization occurs, put simply,
(152) /n/→[ ] — /i/, [syll]
When the rule is formalized it appears thus: (153) + const + son - cont + ant + cor
_____ →
+ syll - const + high - back
[- ant]
+ syll - const , (not /i/)
This formal rule maybe summarised as: (153a) C
V
+nas +alv
+high -back ,
→
+pal
—
syll (not /i/)
As observed, /i/ maybe termed as a ‗palatal‘ vowel since it is a high front vowel. /i/ therefore influences the nasal /n/ to gain the feature [+palatal] in the following forms:
(154) /ni+umba/ → [njumba] → [ umba] /ni+ugu/ → [njugu] → 166
[ ugu]
house pot
/ni+ama/ →
[njama] →
[ ama]
meat
/ni+ag / →
[njag ] →
[ ag ]
thighs
/ni+
r / → [nj
r ]→ [
/ni+ ga/ → [nj ga] →
[
r ] ga]
/ni+und/ → [njund] → [ und]
black ant hip hammer
One necessary condition for this type of palatalization is that the high, front vowel must be followed by a different vowel, that is, it should not be itself. As is usually the case, exceptions do abound as is evident in: (155) /ni+aba/→ [njaba] on behalf This process, which may be refered to as nasal palatalization, was earlier noted by Polome (1967:70) where he argues that /n/ → [ ] before a noun or adjective stem with an initial vowel. He gives examples such as: [ ama] meat [ ta] star [ uki] bee [ up ] white
It is thus evident that, as a process, palatalization is fairly well evident in Standard Kiswahili especially as pertains classes 7-8; 9-10 nouns and adjectives.
3.5
Homorganic Nasal Assimilation
Abercrombie (1967) states that homorganic nasal assimilation is a natural process whose occurrence is dictated by the anatomy of the articulatory tract. This process at times blurs
167
the underlying segment which gives rise to the assimilation of the nasal to the following obstruent, especially in cases where a segment has been deleted.
Katamba (1989)
observes that such deletion often leads to loss of ―naturalness‖, and refers to this phenomenon as ―telescoping‖.
Research shows that homorganic nasal assimilation occurs in reference to voiced obstruents, and where a resonant8 occurs, then, through additional processes, it is realized as an obstruent.
As with most other phonological processes, homorganic nasal
assimilation aims at, among other factors, simplifying articulation of various phonological items and doing away with complexities. NGP theory argues that these processes eliminate the less natural segments and replaces them with more natural ones [see Katamba: 1989: 114-115)].
Homorganic nasal assimilation is one of the most
natural processes, [see Bakari (1982:125) and Mberia (1993:124)], consequently, it is evidently phonetically motivated and thus broadly exceptionless.
In Kiswahili, this
process is quite widespread, Bakari (1982:115) observes, To a significant degree, the assimilation of the nasal to the following obstruent is uniform in virtually all the Swahili dialects that we have investigated. Although Standard Kiswahili is not one of those dialects investigated by Bakari, this study does confirm that the feature is no less widespread in the standard dialect as will be exemplified below.
However, unlike Bakari, Katamba (1989) is of the view that the homorganic nasal assimilation rule in Kiswahili is ―morphologised‖. He observes that it is not automatically
168
triggered by phonetic information and thus, it requires morphological information. Although there is a problem with his data (where he places some class 1/ 2 nouns in class 9/10), Katamba correctly observes that any class 1 or class 3 nasal prefix is syllabic but a class 9/10 prefix is syllabic only if the root to which it is attached is monosyllabic. To him, the distinction between monosyllabic and longer roots is clearly phonological but information concerning noun class membership is morphological.
In Standard Kiswahili, this process mostly (but not exclusively) involves the archiphoneme /N/ and falls in the category of regressive assimilation. /N/ is realized either as [n], [], [ ] or [m] under different environments. This is most pronounced in class 9-10 nouns but is also evident in nouns in other classes including 1-2. In fact, Katamba (1989) notes that in Kiswahili a word-medial nasal or a nasal prefix marking classes 9/10 must be homorganic with the following consonant but that a nasal consonant marking classes 1 and 3 does not necessarily have to be homorganic (as is the case with /mtu/ person and /mti/ tree.). Further, he remarks that while the assimilation of a nasal to the place of articulation of the following consonant is itself a natural process, its implementation in Kiswahili, as observed above, is sensitive to nonphonetic factors. The changes are:
a) /N/ is realized as [n] when it precedes an alveolar obstruent 9, thus, (156) /N/→[n]
— [+alv]
examples include: (157) /n+zig / → [nzig ]
169
locust
/n+zumari/ →[nzumari]
clarinet
/n+s / →[ns ]
kidney
/n+swi] → [nswi]
fish
/n+ta/→ [nta]
wax
/n+ti/ → [nti]
ear ornament
/n+dama/→ [ndama]
calf
/n+d ge/ → [nd g ]
bird/aeroplane
/n+dw l /→ [ndw l ]
illness
/n+l a/→/n+d a/→[nd a]
marriage
/n+r fu/→/n+d fu/→[nd fu] long In most of these forms, the deep structure is very much like the surface structure. The reason for this is that since /t/, /d/, /s/, /z/, /r/ and /l/ are all alveolar consonants, just as /n/ is, as a result, there is no phonological condition dictating the change of /n/ at the surface level. The articulation of these sounds vis-à-vis /n/ is quite natural.
b) /N/ is realized as [] when it is followed by a velar obstruent, for example: (158) /n+ga / → [ga ]
shield
/n+galawa/ → [galawa]
canoe
/n+g li/ → [g li]
noun class
/n+g g / → [g g ]
catfish
/n+g ma/ → [g ma]
drum/dance
170
/n+guvu/ → [guvu]
strength/force
/n+gw na/ → [gw na]
crocodile
/n+gw n ] → [gw
]
(coin) money
The reasoning here is that articulation of /n/, an alveolar nasal in the environment of a velar obstruent is complex, therefore, the alveolar is harmonized to the velar leading to //, a velar nasal, consequently, (159) /N/ → [] — [+velar]
c)
/N/ is realized as / / in the environment of preceding a palatal obstruent as in: (160) /n+ a/ → [ /n+ i/ →[ /n+
a]
tip
i]
country
r /→ [ t r ]
canoe
/n+ aa/ → [ aa]
hunger/famine
/n+ ia/→[ ia]
way, path
/n+ ama /→ [ ama] conspiracy /n+ zi/ → [
zi]
/n+ uga/ → [ uga]
dream ankle bells
/n+ umu/ → [ umu] shoe cleats
171
Like / / and / /, / / is a palatal consonant and thus is the nasal that appears more naturally in place of /n/ which is an alveolar in a palatal neighbourhood. The phonological process may be shown thus: (161) /N/ →[ ] — [+pal] It is, however, important to note that in the cases where the nasal is followed by a vowel with the features [+HIGH, -BACK] and the vowel is in turn followed by another vowel (which is not itself), the process of glide formation precedes that of homorganic nasal assimilation. This, as indicated in section 3:4, is evident in lexical items such as: (162) /ni+umba/ → [njumba] → [ umba] house /ni+ugu/→[njugu] → [ ugu]
pot
/ni+ama/→ [njama] → [ ama]
meat
/ni+ag / → [njag ] → [ ag ]
thighs
It can, therefore, be argued that in such cases the palatalisation rule is actually a feeding rule to the homorganic nasal assimilation rule.
d) /N/ occurs as [m] when it preceeds a bilabial obstruent as the following examples show: (163) /n+bari/ → [mbari]
clan
/n+bingu/ → [mbingu]
heavens, sky
/n+buzi/ → [mbuzi]
coconut grater
/n+b ni/ → [mb ni]
eye pupil
/n+binu/ → [mbinu]
method
172
/n+barika/ → [mbarika]
castor nuts
/n+b l k / → [mb l k ]
baby carrier
Although, as mentioned, most of these nouns belong to class 9-10, there are also others from different classes like [mbuzi] (goat) and [mbag ] (warthog) from class 1-2. In this case, /n/ is harmonized with /b/, a bilabial and is thus realized as /m/, the bilabial nasal. This process can be summarized thus: (164) /N/→[m]
— [+bilab]
However, homorganic nasal assimilation is not unique to the alveolar nasal. It is also evident in the articulation of the bilabial nasal especially when it precedes the voiced labio-dental obstruent. This happens to nouns of most classes as shown below: NOUN
GLOSS
(165) /m+vivu/ → [ vivu]
lazy person, idler
CLASS 1-2
/m+vuvi/ → [ vuvi]
fisherman
,,
/m+vj l ] → [ vj l ]
old person
,,
/m+vi/
grey hair
→ [ vi]
9-10
/m+viza/ → [ viza]
an evergreen tree
/m+vul / → [ vul ]
wooden bowl
,,
/m+vungu/ → [ vugu]
empty space
,,
/m+vuk / →
[ vuk ]
vapour
,,
/m+vua/ →
[ vua]
rain
19
173
3-4
Here, the bilabial nasal changes place of articulation to be realized as a labio-dental nasal. Of course, it is noted that for the class 1/ 2 nouns given as examples above, the vowel /u/ which occur between the nasal and the voiced labio-dental is first deleted in order to form the sequence NC. The change maybe shown thus, (166) /M/ → [ ]
— +lab-dent. +voice
So far, concerning this process, this study has given examples mostly from nouns but it is important to note that whenever these sound segments (the nasals) precede voiced obstruents even in other word categories, the same rules apply as exemplified in the following verbs:
(167) /ungan ni/ → [ugan ni]
unite
/ nd a/ → [ nd a]
remove
/nikun i ] → [niku i ]
fold (it) for me
/n
/→[
]
come
/mvami /→[ vami ]
attack him/her
/mvizi /→[ vizi ]
waylay him/her
The rule governing homorganic nasal assimilation in Kiswahili has been written in slightly different variations by different scholars, for example: Habwe (2004):
[+nas]→[ place] — C
174
Njuguna (1992): [+nas]→ [ place]
C [+nas] →[ place]
Bakari (1982):
+const. — -syll. place
— [ place]
Although all these versions mean the same thing, some like Njuguna (1992) are more easily understandable and clear though not specific enough. However, this study also has a different variation of the rule, thus:
(168)
[+nasal]→ [ place] — +const -nas place
The variation here is that this rule indicates that the consonant which assimilates the preceding nasal itself should not be a nasal.
It has been indicated that this is regressive assimilation since the obstruent influences the nasal coming before it. However, this study notes the presence of progressive assimilation in Standard Kiswahili and in which case the nasal also plays a central role. This usually happens when a resonant precedes the nasal. The resonants, which are essentially continuants, acquire the feature [-CONTINUANT] and consequently are harmonized with the nasal (which in most cases is the alveolar one). Since in this case the resonant is influenced by the nasal before it, then this is progressive assimilation. Examples include:
175
(169) /n+ an / → [nd an ]
fish hook
/n+ imi/ → [ndimi]
tongues
/n+ a/ → [nd a]
marriage
/n+ t / → nd t ]
dream
/n+r fu/ → [nd fu]
long, tall
It can therefore be summarised thus:
(170)
l r
→ [d]
/n/ —
This leads to the formalisation of the following rule: (171)
+const + son + cont + ant + cor
→
- son - cont
+ const + son - cont + ant + cor
—
Notable in all these cases is the simplification in articulation of neighbouring sounds. As Mberia (1993) states, homorganic nasal assimilation results from the early adjustments of the articulators in anticipation of the following consonant, consequently, the points of articulation of the nasal and the following consonant are harmonized leading to this articulatory simplification.
176
In conclusion, as noted by Katamba (1989), the nasal prefix /N/ in Kiswahili undergoes homorganic nasal assimilation whereby the nasal class prefix adjusts to the place of articulation of the noun root if it begins with a voiced consonant. Consequently, as has been illustrated, the nasal is either labial, labio-dental, alveolar, palatal or velar depending on whether the first consonant of the noun root is bilabial, labio-dental, alveolar, palatal or velar. Based on this observation, he posits the following general rule:
C → [+nasal]
ant cor back
—
ant cor back
Where the Greek letter variables alpha, beta and gamma mean that the nasal and the consonant after it share the same features, for example both maybe [+ant] or [-ant]. Clark & Yallop (1995:146) even extend this general rule on assimilation further by inclusion of the feature [HIGH] and consequently the addition of the Greek delta to cater for the additional shared feature. They posit the rule as:
# +cosnt. +nasal
→
ant cor high back
—
ant cor high back
This rule is not essentially different from the one posited in this study to denote (regressive) homorganic nasal assimilation in (168) above.
177
3.6
Consonant Deletion
To some scholars, deletion is a form of weakening. Sommerstein (1977:228), for example, writes of deletion as being the extreme case of weakening. Katamba (1989: 105) on his part notes: Of course, having no obstruction at all and dropping a sound altogether is the ultimate form of weakening… In some cases, before being deleted, a voiced stop (for example) may go through a fricative phase. (Emphasis added)
This study, however, treats deletion separately because unlike in the previous section on weakening (3:1) where phonemes are realized differently (which maybe at the fricative or any other phase), in deletion the phonemes are realized as / /. According to Habwe & Karanja (2004) this process may be viewed using two perspectives as regards Standard Kiswahili, these are: a)
Deletion as a result of historical reasons and
b)
Deletion as a result of Ganda Law.
However, this study notes the existence of more parameters in Standard Kiswahili, for example there are rules governing trill deletion as well as nasal deletion which are different from those mentioned above.
3.6.1
Historical Reasons
In their exposition, Habwe & Karanja (2004) argue that the consonant mostly affected by the historical phenomena is /l/ in the environment of occurring between two vowels, thus,
178
(172) /l/→/ / v—v Examples of lexical items that exhibit this process are: (173) /l la/ → [ a]
marry
/kil l / → [ki :]
mirror/glass
/kip g l / → [kip g :] aspect The above rule may be written formally as: (174)
+ const + son + cont + ant + cor
→ [ ]
+syll — - const
+syll - const
Drawn from: (174a)
+const +lat -back
→/ /
[+syll] — [+syll]
This rule is however not exclusive and it does not seem to be any more productive in Standard Kiswahili. There are many similar lexical forms to which the rule cannot be extended. Such lexical forms include: (175) /lala/ → [lala]
sleep, lie down
/kalamu/ → [kalamu]
pen
/ki+l ma/ →[kil ma]
disabled person
/ki+l mba/ →[ kil mba]
turban
/ki+l l / → [ kil l ]
peak, apex
/lalama/→[lalama]
complain
179
The rule is, however, quite productive in the Kiamu and Kitukuu dialects of Kiswahili language [see Bakari (1982)].
3.6.2
Ganda Law
The other process denoting consonant deletion is the Ganda Law which is also referred to as ―Meinhof‘s Law‖ (for example by Meeuseen: 1963; Herbert: 1977). This is because the law was formulated by the outstanding Bantu languages linguist Carl Meinhof when he was analysing the Luganda language. In the law, Meinhof postulates that a steminitial consonant is usually deleted when it follows a nasal consonant and is itself followed by a sequence of a vowel and a nasal consonant. This rule may be indicated simply as follows: (176) C → [ ]
N + —VN
The examples that follow show the existence of this phenomenon in Standard Kiswahili: (177) /n+gamba/ → [amba]
shiny part of turtles‘ shell
/n+g mb / → [ mb ]
cow/cattle
/n+gamb / → [amb ]
across
/n+ganda/ → [anda]
pip of playing cards
/n+g nda] → [ nda]
dried fish
As is evident, most of these lexical items belong to class 9-10. However, there are a few from other different classes (like [ mb ], cow/cattle and [ nda], dried fish from class 180
1-2). One thing is common to all these items; in the morphological classification of nouns, their singular and plural morphemes are always represented by the morph { }. In all the above nouns, the nasal /n/ is followed by the consonant /g/ (which is [-nasal]). This consonant is then deleted. However, before the deletion occurs, /g/ is harmonized with /n/ through the process of homorganic nasal assimilation (see section 3:5) and consequently, phonetically, /n/ is realized as []. This, therefore, is a case that denotes the feeding rule where homorganic nasal assimilation ―feeds‖ the deletion process.
In this case, just as Katamba (1989) observed of Luganda, the implications for rule ordering are obvious. For this surface realization to occur, homorganic nasal assimilation must precede deletion. This is the only way that both rules may occur. 10 Katamba (1989:125) observes: Technically, this kind of rule relationship where one rule opens the door to the application of another rule is called FEEDING ORDER. As a result of these rules, /g/ is completely deleted leading to ―absolute dissimilation” (Mberia: 1993) where the phonetic structure changes thus: (178) NCVN→NVN In Kiswahili, the phonemes involved are mostly /g/ and /n/. The process may thus be shown as: (179) /g/ →
N+ —VN
and therefore:
181
(179a) C →
C +— V C
-nas
+nas
+nas
Informally, therefore, the rule may be re-written thus:
(180)
C + back +high +plos
# +const +alv +back +nas
→
+—V [+nas]
+ voice
This rule may then be simplified as follows:
(180a)
# +const - nas →
+const +const +nas +— [+syll] +nas
Written formally and in full, the Ganda law then is: (181)
# +const - son - cont → - ant - cor + voice
+const + son - cont
+ syll + — -const
+ const + son - cont
This rule reduces a nasal-consonant sequence into a nasal. This single nasal has no effect on the vowel. 11 As is evident, it is the root initial which is deleted.
182
One more remark concerning Ganda Law in Standard Kiswahili is in order. It is noted that there are items that appear marked for the rule but which deviate from it. Examples include /gan/ (wheat; fable), /gawira/ (booty), /g/ (scorpion), /giri/ (wild-boar; hernia), /gg/ (catfish), /gumi/ (fist) and /gwena/ (crocodile) among others. Since such items are many and diverse, they cannot be said to be exceptions or historical residues.
One may then ask: Is Ganda Law still productive in Standard Kiswahili? One viable suggestion under NGP is that Ganda Law may have been blocked at some point in time and is, therefore, no longer productive in Standard Kiswahili. This argument would only be strong if the process is ignited by a P-rule. P-rules, as earlier stated, are said to be natural, unsuppressable and exceptionless. A second argument would be that the rule is still productive but it is an MP-rule in Standard Kiswahili and is, therefore, only evident in lexical items marked for it. Since many lexical items do not subscribe to Ganda Law in Standard Kiswahili, this second suggestion appears to be more justifiable and plausible.
3.6.3
Trill Deletion
However, it is not only the velar plosive that is deleted in Standard Kiswahili. There are instances where other consonants are also deleted. A good example is where the alveolar trill is deleted in the environment of following a nasal and preceding a vowel-nasal sequence, thus: (182) /r/→ / N+—VN This is the case in the noun:
183
(183) /n+r mb /→[n mb ] tattoo, beauty marks Here, the rule maybe indicated formally as: (184) +const + res + cont + ant - cor
# →
+ const + son - cont + ant - cor
+nas + —V
drawn from: (184a) +const - nas →
#
+const +nas + —[+syll]
+const +nas
It is clear that rule (184a) is similar to rule (180) used to depict the deletion of /g/ in Ganda Law. In fact, some scholars, inter alia Armstrong (1967), Herbert (1977) and Mberia (1993) discuss trill deletion under Ganda Law. Safe for the segments being deleted (the trill and the voicelless velar plosive), all other variables are similar in both cases.
Mberia (1993) further notes the limited scope of the rule in Kitharaka. This observation is also true of Standard Kiswahili. He goes ahead to give two probable reasons for this limited scope first of which sees the rule as being no longer operational and therefore the few lexicons evident being cases of historical residues. This argument is perhaps given impetus by the fact that at times there seem not to be a clear dividing line between synchrony and diachrony. Concerning this particular issue, Katamba (1989:137) observed that:
184
Phonological processes which might have been regular, natural and motivated at an earlier period in the life of a language may only survive as sporadic alternations in a few isolated forms. However, in the case of Kiswahili, it is the second of Mberia‘s (1993:105) reasons which appears more appealing; he states that: ….. (alternatively), the rule is synchronically operational in the language but that it is not a P-rule (see Hooper 1976). Rather, it is a morphophonemic rule affecting only those morphemes that are positively marked for it. We could then extend this argument to the effect that only very limited number of morphemes are positively marked for the rule.
This argument seems very valid in the case of Standard Kiswahili. In NGP, according to Hooper (1976), a morphophonemic rule applies to a morphological class or category thus changing phonological features but this is not done in a purely phonetic environment. The theory states that morphophonemic rules may (and do actually) have exceptions.
3.6.4 Nasal Deletion Investigation has also revealed that there exists a nasal deletion rule in Standard Kiswahili. In the previous section (3:5), it was noted that homorganic nasal assimilation occurs when a nominal base begins with a voiced consonant. However, where a nominal base is preceded by a nasal prefix begins with a voiceless stop or any other obstruent, be it a fricative or an affricate, homorganic nasal assimilation does not occur (Katamba 1989:126). The rule posited in (168) above is blocked. Instead of being assimilated, in
185
this case, the nasal is dropped. It is deleted. A few examples from class 9-1012 nouns will be useful in understanding this process:
(185) /N+sufuria/→[sufuria] metal pan (not [nsufuria]*) /N+silaha/→[silaha] weapon (not [nsilaha]*) /N+filimbi/→[filimbi] whistle (not [filimbi]*) /N+fimb /→[fimb ] stick (not [fimb ]*) /N+tai/→[tai] neck tie (not [ntai]*) /N+tabu/→[tabu] misery (not [ntabu]*) /N+p r m nd /→[ p r m nd ] peppermint (not [mp r m nd ]*) This phenomenon is also evident in other word categories like adjectives as in:
(186) /N+sufuria N+kubwa/ → [sufuria kubwa] a big pan /N+ p r m nd N+tamu/ →[ p r m nd tamu] sweet peppermint The rule used, therefore, reflects the fact that the alveolar nasal is deleted in the environment of occurring before a voiceless obstruent, thus: (187)
+ const + son - cont → + ant + cor The general rule for the above formal rule is:
—+
(187a) N→
—+
186
-cont -voice
+const -sont - cont - voice
This rule, however, does not appear to be exclusive as, for example, it does not apply to the following forms: (188) /N+ a/ → [ a]
tip
/N+ i/ → [ I]
country, land
/N+s / → [ns ]
kidney
/N+ta/ → [nta]
wax
/N+ti/ → [nti]
ear ornament
/N+tw ] → [ntw ]
stigma
One thing though is evident from these items; in the absence of the N prefix, they are all mono-syllabic. One is, therefore, tempted to argue that the exemption to the rule are mono-syllabic lexical items, but again this is rendered null by the following three examples: (189) /N+timbi/→[ntimbi]
water shore
/N+kindiza/→[nkindiza] /N+
r /→ [
r ]
low tide a small canoe
Now, the issue of syllables may appear to be far fetched but a few remarks are in order. Firstly, in spite of broad research, these three are the only items that this study could identify in Standard Kiswahili that are not mono-syllabic but do not adhere to the nasal deletion rule despite the fact that they theoretically seem positively marked for it. There is no apparent phonetic (or morphological) justification for these items retaining the nasal. Infact, when one adds a semantic angle – the fact that all three items are associated with the coast - have something to do with water- the issue becomes even more interesting, and complex. 187
Since the rule still seems to be productive in Standard Kiswahili, then the only reasoning left is that the three items are historical residues which were by-passed by the rule and have since ―refused‖ to adapt to it. For a rule to be productive, the assumption, usually, is that if new forms enter the lexicon of a language, then chances are that the rule will be observed, meaning it is still operational. This study observes that insofar as the nasal deletion rule is concerned, productivity is almost guaranteed. In fact, there are some items that, according to this study, are still in the process of adopting the rule, for example, (190) /N+si/→ [nsi] [fish] may also be articulated as [isi]13 However, the best example is: (191) /N+swi/ → [nswi] or [swi]
(fish)
Here, the nasal may (or may not) be dropped. Chances are that with time, the nasal could be completely dropped. Of course syllabification factors are also very important when processes such as deletion occur in language.
Secondly, since the lexical items that do not subscribe to the deletion rule are few (and almost negligible), one may be tempted to dismiss their existence but this would be a grave mistake. The question that begs is: is the nasal deletion rule a P-rule or an MPrule? Strictly speaking, at least according to Hooper (1976), this may not be a P-rule since it has exceptions. However, by putting a caveat, that for the rule to apply, the lexical items involved should not be mono-syllabic, it may be possible to tie the rule within the P-rules, safe for the three lexical items in (189) above that are not mono-
188
syllabic but still have retained the nasal. This is where NGP views such items as historical residues, a view that seems very useful in this particular case. In fact, NGP generally encourages the existence of residues as this is seen as incontrovertible evidence that a certain change did actually occur, hence the rule. It goes further to strengthen the NGP claim of naturalness as opposed to the ―abstractness‖ evident in the standard model of GP.
3.7
Concluding Remarks
In this chapter, consonant sound changes evident in Standard Kiswahili have been analysed. The changes discussed include lenition, fortion, palatalisation, homorganic nasal assimilation and consonant deletion. The general and formal rules governing each sound change have been posited. Reasons giving rise to the the various changes have been suggested. Discussion and analysis in the chapter are informed by the NGP theory. Notes
1.
This matter of exceptions is explained theoretically later in the chapter, just before positing the rule that governs weakening.
2.
For example, the nominalised form of the word /taka/ (need) is not /matak /* as may be expected but /matakwa/ (needs). One basic explanation for this deviation from the norm of the language‘s morphology is the existence of the noun /matak / which means buttocks.
The connotation here makes the
deviation imperative as any ambiguity would have a negative impact. Of course many more examples may be given. The diminutive and augmentative forms of /kiti/ (chair) are /ki ikiti/ and / ikiti/ respectively due to the existence of the terms / iti/ (the augmentative form of tree) and /ki iti/ (stick). /kiatu/
189
(shoe) is not /tatu/ because that word means python. The augmentative form of /mi/ is not / i/ as that would refer to a cattle dip. 3.
Although many Kiswahili linguists such as Mgullu (1999:87) place [m this group, this study sees it differently. [m would be expected from /m+ sense of /m+
4.
i] in
i] is not ―one who laughs‖ (as
k+i/) but ―one who makes others laugh‖ in the
k+ +a+ i/.
See section on Theoretical Framework, NGP, for a deeper exposition on these rules.
5.
This process is discussed later in the chapter, in section 3:5.
6.
In some other Bantu languages related to Kiswahili the process stops at this point thus forms such as [kj nd ] (basket), [kj r ] (restroom), [kjandaroa] (net) [kjama] (party), as evident in Gikuyu language.
7.
Palatalization is a dorminant process in the Chi-Jomvu dialect especially as far as nouns in class 7-8 are concerned.
8.
It is shown later in this section (in the sub-section on progressive assimilation) that at times resonants such as /r/ and /l/ may occur but these segments are realized as [d]. The process involved here is progressive assimilation and denotes consonant strengthening as discussed in section 3:3.
9.
For the other two alveolar consonants (resonants), see comment 8 above.
10.
This statement is however not meant to imply that this study proposes limitless adherence to linear phonology. It only proposes the use of linearly ordered rules to the extent that they are intrinsically ordered. Extrincally ordered rules are not relevant to this study.
Most proponents of NGP
including Hooper (1976) have argued strongly against extrinsic linear ordering, proposing instead that intrinsic ordering is the only kind of rule interaction that should be allowed [Katamba (1989)]. However, even she, Hooper, argues a case for proper arrangement of rules (1976:53). As Katamba (1989:129) points out, this proposition is made out of the realisation that: A (total) ban on linear rule ordering effectively means that only those underlying representations which require rules
190
interacting in a straight forward way can be successfully mapped on phonetic representations. Of course there are many parameters that may be used to describe and explain intrinsically ordered rules and these include Simultaneous Rule Application, Random Sequential Rule Application and Elsewhere Condition (Kiparsky (1973). 11.
That is, no effect, safe for the vowel nasalisation process which is discussed in the next chapter, section 4:6.
12.
It is noted that this study prefers the use of an eclectic approach in nominal classification. Consequently, examples such as /N+pang / (gadfly) – class 1/2 and /N+ umvi/ (salt) – class 19 as given by Katamba (1989:126/127) are avoided. However, the rule used is similar to the one posited by him.
13.
Here, it seems, apart from nasal deletion, there is another process involved which gives rise to [i]. This other process is most likely epenthesis which leads to the insertion of /i/.
191
CHAPTER IV Standard Kiswahili Vowel Morphophonology 4.1 Preliminary Remarks In a way, this chapter is an advancement of processes discussed in the previous chapter. Issues expounded in it as pertaining sound change are still very relevant to the current discussion. However, instead of dealing with consonant processes, this chapter engages in an in-depth analysis of the various vowel phonological processes evident in Standard Kiswahili. In addition to the discussion and analysis, adequate examples are offered and rules underlying each process are posited generally and formally. Although the major theory used in this chapter is that which is applied to this whole research, namely NGP, a second supplementary theory (APT) is used in sections of the research to enhance NGP where it might appear inadequate. This is especially the case in the analysis of two processes, namely vowel harmony and nasalisation.
The first process discussed is vowel coalescence. The process is expounded and issues pertaining to rule formulation are then tackled. Next to be discussed is the vowel deletion process. It is observed that this is a form of weakening and is discussed under the tenets of vowel cluster and redundancy. Glide formation (itself a form of fortition) is discussed as a regular and productive process; both orthographical and phonological glide processes are expounded.
192
Vowel harmony is studied under both segmental and supra-segmental parameters especially insofar as rule formulation is concerned. Two categories of words are used: the demonstrative pronouns and the verbs with extension suffixes. The supra-segmental theory that is applied is APT. The same theory is used to posit more rules for the vowel nasalisation process. This process is seen as a form of assimilation and the issue of naturalness and markedness are revisited. Instances when nasalisation occurs are pointed out and exemplification done.
4.2
Vowel Coalescence
Vowel coalescence is a process that occurs when two different vowels occur in neighbouring environments. It so happens that due to major articulatory differences between the two vowels at the phonological level, they coalesce and thus they do not occur at the phonetic level; instead, a third vowel replaces the two. It may be argued that the two different vowels influence each other leading to what is referred to as reciprocal assimilation – a situation where each of the vowels assimilates the other; leading to the loss of both and the realization of a new one.
A good example in Standard Kiswahili is when the low vowel /a/ appears in neighbouring environment with the high front vowel /i/. Both vowels assimilate and are phonetically realized as / /. What happens, with the express aim of easing articulation, is that the low vowel is attracted upwards while the high vowel is attracted downward hence leading to a mid-vowel, thus:
193
(192)
/i/
/u/
/ /
/ /
/a/ (192a)
Where /a/↔/i/→[ ]
There are many lexical items in Standard Kiswahili which display this process, like: (193) /wa+igi/→[w gi]
many (of people)
/ma+in /→[m n ]
teeth
/ma+ik /→[m k ]
kitchen, cookers, stoves
/wa+izi/→[w zi]
thieves
/pa+igin /→[p gin ]
another (of place)
/wa+itu/→[w tu]
ours (e.g. children)
In all these case, /a/ and /i/ coalesce and are realized as [ ] phonetically. This eases articulation since the movement of the tongue from [-HIGH] position to [+HIGH] in quick succession is avoided.
Now, formulation of a rule governing coalescence using the generative phonology convention is quite problematic. Mberia (1993: 181-182), while describing a similar process in Kitharaka observed that:
194
Formulating a rule to govern the change from /a+e/ to [ ] presents difficulties … In such rules, whereas the S.D. (structural description) changes into the S.C. (structural change), the environment remains unchanged. In the coalescence process … each of the two vowels in the sequence /a+e/ is simultaneously an S.D. as well as an environment.
Of course, as Mberia (1993) correctly observes, the environment ultimately coalesce with the S.C and hence both are phonetically realized as one.
When all this is put into consideration, it is evident that it is not easily possible to formulate a rule that indicates the environment as is done with other processes. Moreover, this process necessitates the positing of a rule that has two parts on the left side of the arrow instead of one as is common in generative phonology. Consequently, this study posits the following rule to describe vowel coalescence: (194) + syll +syll - const + - const + low + high - back + tense
→
-high -low -back - tense
It is, however, worth mentioning that Njuguna (1992:29) did posit a rule depicting the environment as follows: +syll -const +low
+
+vowel -const +high -low
# -high → -low
+const _____ -vow
195
However, due to the reasons given above, this research holds the view that this should not be the case.
Moreover, there actually are some items that are not preceded by a
consonant but which nevertheless undergo the process as in: (195) /a+itu/→[ tu] /a+inu/→[ nu]
ours yours
Such forms denote that the rule posited by Njuguna (1992) is not inclusive enough. Although a rule is posited in this study, a suggestion is offered that this issue of rule formulation on coalescence requires further investigation.
It is noted that the coalescence rule is not as exclusive as is required of phonological rules in NGP. It appears that the rule is more of an MP-rule (MP-rules are not exclusive and allow for alternatives) than a P-rule. This is because there is a large number of lexical items with the structure /a+i/ but which negate the coalescence rule though they appear marked for it as in:
(196) /uraibu/
addiction
/u aibu+ni/
abroad
/mai a/
life
/maiti/
corpse
/aibu/
shame
/wa+igiz+a+ i/
performers (in a play)
/wa+imb+a+ i/
musicians, singers
196
The data above shows that the coalescence rule in Standard Kiswahili has many exceptions, nevertheless, the rule is widely spread and as has been shown, it can be explained phonetically.
4.3
Vowel Deletion
As mentioned in section 3:6, deletion is at times described as the ultimate form of weakening. This scenario is even more interesting in relation to vowels. Being the weakest phonemes possible, vowels cannot weaken further. They can only be deleted. Due to the nature of articulatory anatomy, there are instances where the presence of vowels complicates articulation and in order to simplify it, then, naturally, vowel deletion occurs.
The other reason that causes deletion is avoidance of redundancy.
Vowel
deletion, as explained below, occurs under different environments.
4.3.1 Deletion Occasioned by a Vowel Cluster Firstly, deletion is most pronounced in Standard Kiswahili in the environment of a vowel cluster.
According to Schane (1973), deletion in such a case does not only cause
simplicity in articulation, but also enables attainment of the preferred syllable structure. In Standard Kiswahili, the vowel most affected is the low one, for example:
(197) /wa+anafunzi/ → [wanafunzi] /wa+ usi/
→ [w usi]
students black (for example people)
/ha+utaki/ → [hutaki]
you do not want
/pa+ak / →
his/her place
[pak ]
197
/wa+ ma/ →
[w ma]
good, nice (of people)
/wa+ k vu/→ [w k vu]
born again, saved (ones)
Hooper (1976:110) states that deleted vowels must be phonologically predictable on the basis of universal principles. This predictability seems obvious here. The deletion causes the CV sequence and even tempts one to formulate the rule: (198) →
+ syll - const + low
# [+const]— + + syll - const.
This rule is derived from the sequence: (198a)
/a/→Ø # [+cosnt] — +[+vowel]
However, further scrutiny will enlighten that it is not only the low vowel that is deleted as shown in the following examples:
(199)
/zi+ tu/→ /li+
[z tu]
w /→[l
w ]
/ i+umba/→ [ umba]
ours (eg. houses) itself (e.g. a stone) a big house
In these forms, it is the high front vowel that is deleted. In the examples cited in (197) and (199) above, it is safe to say that it is the vowel found in the prefix (the first one in the cluster of double vowels) that is deleted while the one found in the word-initial position is retained. Before forming the deletion rule, further examples will suffice.
198
Presently, in (200), instead of giving examples using prefixes, suffixes will be used. The following examples which posses the referential {-o} denote the rule:
(200) /hi+zi+ /→[hiz ]
those(eg. of houses)
/hi+li+ /→[hil ]
that (eg. of a stone)
/ha+pa+ /→[hap ]
there (of a place)
/ha+ja+ /→[haj ]
those (e.g. of stones)
In these last examples, again, it is noted that of the two vowels occurring in a cluster, it is the first one that is deleted consequently leaving the referential {-o}. This vowel that is deleted is actually the extension suffix, which, in most cases, is similar (at least as far as the feature [TENSE] is concerned) to the vowel found in the prefix.
From the foregoing, this study suggests that the deletion rule should, as a matter of fact, show that deletion involves all Kiswahili vowels when they occur in neighbouring environment 1, consequently, (201) [+vowel] → ø
# [+const]— + [+vowel]
Formulated as a rule, this information would appear as: (201a) +syll → ø # [+const] — + +styll - const - const Of course this is a general rule but which can well cater for the specific changes that occur; that is:
199
(202)
a /a/ → ø # [+const] — + u
and; (202a) /i/→ ø # [+const] — + u
The choice between which vowel to delete and which to leave does not seem to be optional; it is predetermined. One interesting aspect is that although the deletion of one vowel is purely phonological, the choice of which one to delete seems more morphological. Given the choice between deleting a prefix vowel and a root initial, as has been evident, the language favours the root initial to the prefix vowel, hence the frequent deletion of /a/ which is a prevalent Standard Kiswahili prefix vowel. Mberia (1993) noted a similar phenomenon in Kitharaka and wondered why /a/, a supposedly least marked vowel is deleted in favour of /e/ and /o/ which are more marked. His view is that it is the other vowels which should be deleted if the rule of markedness is followed.
In GP and natural phonology, to digress a little, markedness theorists - at least according to Hyman (1975) - argue that markedness captures linguistic generalisations, and one of those generalisations is that unmarked sounds are generally acquired earlier than marked sounds and that they are also generally required in the inventory of sounds of a language before marked sounds can be added. In fact, Kiparsky (1974), a proponent of the theory,
200
attempted to reduce the linguistic factors that motivate phonological change to three; i.e. ease of production, ease of perception and ease of acquisation.2 In his exposition on the issue Mberia (1993: 172) argues: If (the above) claims are founded – and they seem to be – one would expect /a/ to be either the easiest vowel to articulate, or the easiest vowel to distinguish auditorially, or probably both. A vowel that combines both qualities would be preferred in speech and therefore, we would expect it to be retained when it occurs in a sequence of two vowels that is affected by a vowel deletion process … This raises the question as to whether naturalness of sounds is an absolute and universal phenomenon, or whether it is dependant on individual languages, or perhaps, specific contexts.
At a glance, the question that Mberia (1993) raises may seem rhetorical but in retrospect it is an issue that has been the subject of phonological debate for a long time. One thing that should be appreciated though, is that even major proponents of generative grammar, viz. Chomsky and Halle, did at some point in time (e.g. in SPE) insist that their features were basically meant for Indo-European language family and may not apply universally to all languages. Consequently, there are very strong indications that ―naturalness of sounds (as) an absolute and universal phenomenon‖ may not be a linguistic reality.
But again, the theory may be true and this is when the explanation given in this study gains more weight; that the choice of the vowel to be deleted is determined by extra phonological determinants which include the morphological reasons given above.
201
4.3.2 Deletion Occasioned by Redundancy Another reason that causes vowel deletion in Standard Kiswahili is the presence of redundant features in a certain sound sequence. Hooper (1976) observes that the vowel deleted in such cases is always either the minimal vowel or a vowel with features identical to those of a nearby segment. In cases where the neighbouring segment is a consonant, then the vowel must share some features with that consonant.
A good example in Standard Kiswahili involves classes 1 and 3 nominal prefixes which happen to be {MU-}3. Now, both /m/ and /u/ are resonants, and moreover, both subscribe to the feature [+LABIAL]. While /m/ is a bilabial nasal, /u/ has the feature [+ROUNDED]4. Bakari (1982) indicates that nasals and vowels are all resonants. It is, therefore, phonetically viable that two neighboring resonants delete one in order to minimize feature redundancy. When it gets to this, it is the weaker phoneme - in this case the vowel /u/ - that has to go. Deletion occasioned by redundancy is evident in the following examples: (203) /mu+tu/→
[mtu]
a person
/mu+kaguzi/→ [mkaguzi]
an inspector
/mu+kulima/→ [mkulima]
a farmer
/mu+g ni/→
[mg ni]
visitor
/mu+ti/→
[mti]
tree
/mu+ al →
[m al ]
arrow
/mu+kuki/→ [mkuki]
spear
A rule for this process may be posited thus:
202
(204)
# + syll - const → ø + back + high
+ nas + lab — +[+const]
It is noted that when /u/ precedes a vowel and is itself preceded by /m/, instead of this deletion rule, a different rule, that of glide formation (section 4:4) , comes into operation. Instead of being deleted, /u/ is strengthened and is thus realized as a glide.
4.4 Glide Formation (Vowel Hardening) In section 4:3, it was observed that vowel deletion is a natural process that is aimed at reducing the occurrence of vowel clusters. Glide formation is such other process, consequently, in Standard Kiswahili, glide formation is a process that occurs quite regularly and is among the processes that are productive. In fact, it might not be farfetched to argue that in the absence of glide formation, the deletion rule may apply in certain words in an attempt to rule out vowel clusters, therefore, glide formation helps to avoid deletion. Further, as observed in the previous chapter (see section 3:2:1 & 3:3), glide formation is a form of fortition. In the phonological strength hierarchy, glides are at a higher level of strength in comparison to vowels. Consequently, the process of a vowel changing to a semi-vowel is evidently hardening.
In Standard Kiswahili, vowels harden into two glides; /j/ and /w/ as is evident in the following forms:
203
(205) (a)
(b)
/mu+afrika/→ [mwafrika]
an African
/mu+amk /→ [mwamk ]
awakening, awareness
/ku+ tu/→ [kw tu]
our place
/mu+ ga/→ [mw ga]
coward
/mu+ nza/→ [mw nza]
companion
/mu+ugwana/→ [mwugwana]
child of a noble birth
/vi+al /→ [vjal ]
sugarcane farms
/→[vj
ranks, positions
/vi+
]
/vi+ nd /→ [vj nd ]
baskets
/vi+ugu/→[vjugu]
cooking pots (earthen)
/vi+amb /→[vjamb ]
baits
In the above data, it is noted, both high vowels /u/ and /i/ glide when they are immediately followed by another vowel (not /i/), thus; (206) /u/ → [w]
—+
a
u
And:
204
(207) /i/ → [j]
—+
a
u
While discussing this process in the Chichifundi dialect of Kiswahili, Bakari (1982:75) posits the rule:
(208) + syll + high back
V [-high] → [-syll]
— + V [+high] [ back]
In words, the rule states that a high vowel becomes a glide when immediately followed by a non-high vowel with the opposite value for the feature [BACK]. This rule, as posited by Bakari (1982), would at best be partially true for Standard Kiswahili. As shown above, it is only the front high vowel that does not trigger the process. As for the back high vowel, there are changes such as:
(209)
/mu+ugwana/→[mwugwana]
child of a noble birth
/vi+ugu/ → [vjugu]
cooking pots (earthen)
/vi+u / → [vju ]
colleges
205
Mberia (1993) argues that the formulation of one general rule for glide formation in Kitharaka would result into a highly crowded and clumsy rule, one that would be undesirable. However, as regards Standard Kiswahili, it is possible to formulate a clear and easily understandable rule on the process. Firstly, we note that /i/ and /u/ glide into /j/ and /w/ respectively in the environment of preceding vowels /a, , ,u /, therefore, (210) /i/ /u/
→
[j] [w]
___ +
a
u
Thus: (210a) +syll - Const + high
→ [-syll]
___ +
condition:
+ syll - const condition
+syll should not be +high - const -back
This rule states that a vowel becomes a glide when immediately followed by another vowel provided that the other vowel is not a high front vowel.
It is however important to note that the gloss used in this study as relates to glide /j/ is all in the plural forms. The surface structure of their singular co-relates does not have the glide as it is usually deleted at the same time as palatalization occurs in those items. 5 Further, it is also observed that the palatalization of the alveolar nasal is preceded by the glide formation process.6
206
In all the cases cited above, the glide occurs both orthographically and phonologically. However there are cases where the glides do not occur orthographically but are, as a matter of fact, realized phonologically. This is evident in forms such as:
(211) (a)
/ki+ama/→[kjama]
doomsday
/ki+ l l z /→[kj l l z ]
illustration
/mi+ g ni/→[mj g ni]
amongst
/ki+uka/→[kjuka]
step over
/ti+ara/ →[tjara]
kite
/mi+ nd k /→ [mj nd k ] gait, style (e.g. of walking) (b)
/mu+aama/→[mwaama]
eminence
/mu+amana/→[mwamana]
trust
/u+amilifu/→[wamilifu]
function
/afu ni/
recovery (from sickness)
→ [afw ni]
/fu+ata/ →
[fwata]
/u+aminifu/→[waminifu]
follow loyalty
Once again, in the gloss above, the rule in operation is: (212) + syll - const → [-syll] — + + high
+ syll - const Condition
Condition:
207
+syll should not be - const
+high -back
It is, however, important to note that where the glide does not occur orthographically, it is usually realized in normal, continuous speech7. Where the speech is slow and halting or where a word is mentioned in isolation, the glide may not be clearly audible. All Standard Kiswahili vowels, with the exception of /i/, are involved in the realisation of this process. However, where the [+high, -back] vowel appears the process realized is zero as evident in the following examples:
(213) /ki+imb /→[kiimb ]
intonation
/ki+ini/→ [kiini]
neucleus
/ki+ima/→[kiima]
subject
/mi+ik /→ [miik ]
taboos
/ i+igiz /→[ iigiz ]
indulge in
4.5 Vowel Harmony Vowel harmony is more of a morphological process and less of a phonological one. It is a process where the influence of one vowel is reflected on another to the extent that the two vowels exhibit similar articulatory features. According to Kenstowicz (1994), it is a phonological state in which the vowels in a given domain share or harmonise for a given feature. He further observes that unlike in other processes, all vowels of a particular language participate in the harmonic constraints.8
One general observation that can be made in relation to this process is that many suffixes (especially in reference to verbs) vary the types of their vowels in accord with the root.
208
Kenstowicz (1994) remarks that if a certain word contains more than one suffix, then the harmonic effect propagates from the root through the suffixes to the end of the word. Put in other words, the process of vowel harmony is realized when vowels of a given domain (mostly the word) share a phonological property or properties.
It is not possible to deal exhaustively with this process in Kiswahili here. This arises out of several reasons. First, as Bakari (1982) observed, vowel harmony in Kiswahili is grossly unsystematic in comparison to other languages. This would not be surprising since Kiswahili has not been classified as a harmonic language per se; all the same, the language displays some harmonic tendancies that are hard to ignore. Secondly, the process is further complicated by the fact that it is half segmental and half suprasegmental. A further complication arises from the fact that being an MP- process, the outcome is mostly determined by the morphology of particular words and more so that of the affixes. Bakari (1982: 78) states that: Descriptions of vowel harmony have posed the most intractable problems for scholars working within the Natural Generative paradigm propounded by Vennemann and Hooper. No adequate approach for handling this process has so far offered itself.
To date, this problem still persists. However, there have been several attempts to offer solutions but none seems to fit well within the NGP since, it appears, all tend to lean on abstractness - the very concept that NGP claims to fight.9 Be as it may, this theoretical quagmire does not impact negatively on this study‘s description of the process as is evident in Standard Kiswahili. However, there may appear to be a problem in rule
209
formulation.
To exemplify the process, two word types have been used: the
demonstratives and verbs.
In Standard Kiswahili, demonstratives of proximity exhibit morph copying. Hyman (1975: 233) refers to this as ―complete harmony‖. Consequently, the vowel used in the subject prefix is reduplicated or copied in the demonstrative marker.
Mohammed
(2001:172) observes that: (In Standard Kiswahili) h- is used for proximity. This demonstrative marker takes an appropriate subject prefix and then, between the two, the same vowel occurring in the subject prefix is repeated. This is quite clear when examples are used: (214) Jiwe hili
this stone
Mawe haya
these stones
Mutu huyu
this person
Watu hawa
these people
Kijiko hiki
this spoon
Vijiko hivi
these spoons
Apart from the subject prefix vowel being reduplicated in the demonstrative marker, it also appears that the same vowel is always used as final prefix. More examples include: (215) [huu] this [hizi]
these
[haya] these [hapa] here (definite)
210
[humu] here (within) [huku] here (indefinite) [hii]10 this
The consequence is that the harmonizing vowels are similar to one another. This is a clear form of vowel harmony in Standard Kiswahili. The discussion of vowel harmony in this light does not denote a phonological process per se but this is expected since, as Mgullu (1999) states, vowel harmony does not necessarily occasion change in phonemes. It is merely a phenomenon where certain phonemes appear together in specific environments. For this particular process, as concerns demonstratives, it has proved very difficult to posit a phonological rule using the NGP.
In the demonstratives, the harmony is more or less straightforward. A less clear form of harmony is displayed in Standard Kiswahili verbal suffixes. In such cases, the vowel in the verbal base (though not always reduplicated or copied) has a direct effect on the vowel that appears on the suffix. It has been found out that the high vowel /i/ is realized in the suffix if the verbal stems have vowels /a,i,u/. Examples are:
(216) (a) Verbal stems with /a/: [katia]
cut for
[andikia]
write to /for/ on behalf of
[zalia]
bear for
(b) Verbal stems with /i/:
211
[pikia]
cook for
[imbia]
sing for
[pitia]
pass through
(c) Verbal stems with /u/: [fumia]
weave for
[rudi ia]
return, give back
[fugia]
lock in, lock for
A further observation is that when the stem contains vowels / , /, the mid-vowel / / is realized in the verbal suffix, for example:
(217) (a)
(b)
verbal stems with / /: [p k a]
receive
[s m ka]
be readable
[k p a]
loan to
verbal stems with / / [p nd za]
likeable
[
make laugh
k a]
[ nd a]
go for
From the above examples, it is easy to conclude that the vowel of the verbal base always determines that of the extension suffix. From this research, it is evident that the main phonetic feature that determines the vowel to be realized in the suffix is the feature
212
[TENSE] as it appears in the verb root. In Hyman‘s (1975: 233) words, this is ―partial harmony‖. Katamba (1989) states that the validity of the feature [TENSE] has always been controversial; however, it is not the desire of this study to delve into polemics concerning the issue. This research takes the definition given by Halle and Clements (1983) for the feature as quoted by Katamba (1989: 48), that: Tense vowels are produced with a tongue body or tongue root configuration involving a greater degree of constriction than that found in their lax counterparts; this greater degree of constriction is usually accompanied by greater length. Given this definition, it turns out that Kiswahili language has three vowels that may be described as being tense and these are: /i, u, a/. The remainig two, / , /, are closer to the feature lax.
Now, it is noted that the extension suffix in Kiswahili (be it applicative, stative or otherwise) is usually / i /. It is also evident from the above examples that / i / has two allomorphs, viz. {i} and { }. / i / is a tense vowel while / / is a lax vowel. The examples in glosses 216 and 217 show that when the verbal base has a vowel with the value [+] for the feature [TENSE], then the suffix takes the allomorph which has a similar value for feature. Inversely, when the verbal base has a vowel with the opposite value for the feature, then the suffix takes the allomorph with a similar value. Put simply, the high vowel [i] is realized in the suffixes of the verbal base that contain vowels /i, u, a / and the mid-vowel [ ] will be realized in the suffixes if the verbal base contain vowels / / and / /. This process, it is noted, is very systematic.
213
As mentioned earlier, rule formation for vowel harmony is quite problematic. While writing a rule for the harmony evident in Kiswahili verbal suffixes, Massamba (1996:116) gave the following description:
+ syll - const
→
[+high] [-back] [-high] [-back]
[+syll] [-mid] [+syll] [-high] [-low]
____ Co
This rule does describe the process well, however, it seems too bulky and though it captures the generalizations that are evident, it does not appear to achieve economy and simplicity.11
Bakari (1982:80, 93) posits a more economical rule and one that is much clear and general. The rule is: V [-low]→[ high]
high - low
C— stem
The rule simply states that a suffix vowel will have the same value for the feature [HIGH] as that of the verb stem.
While appreciating the fact that the two rules do basically define vowel harmony, this study posits the rule differently. The major reason for doing this is that the findings of this research are that the feature that determines which amongst the two allomorphs ({i}
214
and { }) appears in the suffix is neither [LOW], [BACK] nor [HIGH]. It is the feature [TENSE]. Before forming the general rule, the specific rules are first posited, thus: (218) /i/→ [ ]
Co —
This may be formally stated as: (218a) + syll - const → [-tense] - back + tense ________ extension
+syll - const Co —
_________ verbal base
That is, the tense vowel /i/ is realized as the lax vowel / / when it occurs in the environment of a suffix whose base has a lax vowel followed by zero or more consonants. The second rule may be posited as:
(219)
/i/ →/i/
i u a
Co —
This rule may formally be stated thus: (219a) +syll - const → [+tense] - back - low ______ extension
+syll Co — - const + tense _________ verbal base
215
This means that in the event of appearing in a suffix whose verbal base has a tense vowel followed by zero or more consonants, the suffix allomorph that is realized has the feature [+TENSE]. Although these two rules appear nominally different; a general rule may be posited to cater for both, that is: (220) + syll - const → [ tense] - back - low tense ______ extension
+syll - const Co — tense _______ verbal base
This rule denotes that the vowel that is realized in the extension suffix of a verb attains the same value for the feature [TENSE] as that of the vowel that appears on the verbal base. Posited this way, this rule captures the necessary generalizations and is also easier to understand. It is clear and economical.
As has been stated, since vowel harmony is viewed as both a segmental and suprasegmental process, this study will in the next sub-section apply APT to rule formulation of the process.
4.5.1 An Application of APT to Vowel Harmony in Standard Kiswahili In the discussion on theoretical framework, this research stated that the APT will be applied in the study of two segmental processes; vowel harmony and nasalisation. The main justification is that the two processes are partly segmental and partly suprasemental. It has actually been found out that it is, at times, almost impossible to posit a
216
rule using NGP to capture all aspects of vowel harmony. This further justifies the application of APT to this study. In APT, a sequence of identical autosegments is usally collapsed to form a simplified rule. Although it is basically a suprasegmental theory it has, time and again been used to analyse these two segmental processes.
According to Katamba (1989: 212), vowel harmony can be studied under the framework of APT by using several principles which he states as: (i)
(ii) (iii)
(iv)
Identification of the set of harmonizing features which are suprasegmentalised and placed on a separate tier; Identification of the elements (i.e. vowels) which bear the harmonizing features; Identification of the set (possibly empty) of opaque segments. These are vowels that ought to obey the harmony rules but fail to do so because they are specified in the lexicon for the harmonizing feature and are therefore exempt from vowel harmony rules which fill in blanks for the harmonizing feature during a derivation; Mutatis mutandis, harmonizing features are associated with vowels in accordance with the well formedness condition.
The feature which is harmonized and put on different tiers in the case of Standard Kiswahili has been identified as the feature [TENSE]. The vowels that bear this feature (i.e. [+TENSE]) are /i, u, a/ while those which bear the value [-TENSE] are / , /. Now, going back to the demonstratives, it is shown how the vowel used in the subject prefix is reduplicated in the demonstrative in what was referred to as morph copying. Using the APT, the rule may be indicted as:
217
(221) (a) .
→
[+TENSE] [+TENSE]
hi
li
[+TENSE]
hi
li
/hili/
[hili] (this)
(b) .
[+TENSE] [+TENSE]
hu
[+TENSE]
→
mu
hu
mu
/humu/
[humu] (here; within)
(c) .
[+TENSE] [+TENSE]
ha
wa
[+TENSE]
→
ha
wa
/hawa/
[hawa] (these; people)
In the case of the demonstratives, no real phonological change is evident though the harmony is clear. It is however worth noting that while it was problematic formulating a rule using NGP; it is easy to posit one using APT. The second analysis involved Kiswahili verbal suffixes. It was concluded that the underlying verbal suffix
218
/i/ has two allomorphs: {i} and { }. /i/ is reaslised when tense vowels appear in the verbal stems and / / is realized when lax vowels appear in the stems. Consequently: (222) (a) [+TENSE]
[+TENSE] [+TENSE] [+TENSE]
→ ka
ti
a
ka
ti
a
[katia] cut for
/katia/
(b) [+TENSE]
[+TENSE] [+TENSE] [+TENSE]
→ pi
ki
a
pi ki
a
[pikia] cook for
/pikia/
(c) [+TENSE]
[+TENSE] [+TENSE] [+TENSE]
→ fu
mi
a
fu mi
a
[fumia] weave for
/ fumia/
All the above rules indicate that when tense vowels /a,i,u/ appear in the verb stem, then the tense allomorph /i/ is realized.
219
(223) (a) [-TENSE] [+TENSE]
[-TENSE] [+TENSE] [+TENSE]
→ p
ki
a
p
k
/p kia/
a [p k a] receive
(b) [-TENSE] [+TENSE]
[-TENSE] [+TENSE] [+TENSE]
→ k
pi
a
k
p
a [k p a] loan to
/k pi a/ (c)
[-TENSE] [+TENSE]
[-TENSE] [+TENSE] [+TENSE]
→ p
ndi
za
p
nd
za
[p nd za] likeable
/p ndiza/
In such circumstances the tense vowel /i/ acquires the feature [-TENSE] and is subsequently realized as / /. Thus, this is yet another clear case of vowel harmony.
4.6
Vowel Nasalisation
The final vowel process to be examined in this research is that of vowel nasalisation. According to Clark & Yallop (1995:32), a vowel may be ―distinctively nasalized‖ when the velum (soft palate) is deliberately lowered to ensure substantial airflow through the nasal cavity. The nasal cavity is used together with oral and pharyngeal cavities leading 220
to the realization of an audible nasalised quality. Consequently, vowel nasalisation is a process whereby a non-nasal vowel acquires the feature [+NASAL] from a neighbouring segment. To produce an oral sound, it is imperative that the access to the nasal cavity be blocked. When an oral vowel follows a nasal in quick succession, it is almost impossible to achieve the blockage; consequently, as the velum is not fully raised, some air escapes through the nasal cavity thereby nasalising the vowel.
Massamba (1996:92) views vowel nasalisation as ―the most common type of assimilation‖ and observes that the process is usually simply referred to as ―nasalisation‖. Katamba (1989:93) states that historically, nasalisation is ―almost always a consonant feature‖ that is assimilated by vowels. He further states that in a synchronic description of language, it is possible to find vowels that are always nasal and which must be presumed to be underlyingly nasal.
The only justification for such occurrences, it
appears, is that at one time, historically, there would have been a nasal consonant which then conditioned the vowel nasalisation but which has since disappeared. Katamba (1989:102) goes on to advance an argument on the naturalness of vowel nasalisation in the environment of neighbouring a nasal, he observes: Nasal vowels … are marked. Indeed, we would be extremely surprised if we found a language which had only nasal vowels and no oral ones. However, between two nasal consonants, or before a nasal plus consonant cluster like [nd], nasalised vowels would be unmarked. It would be somewhat unusual for vowels occurring in those contexts to have no nasalisation.
221
Put differently, the normal tendency in languages is to have oral vowels. A case of a language having only nasal vowels would be grossly abnormal. However, it would be abnormal to have an oral vowel in the environment of NVN.
In such cases, the
realization of a nasalised vowel would be unmarked, i.e. normal.
Although this process is rarely discussed vis-à-vis Kiswahili language, it is this study‘s contention that the rule is much spread, and as in other languages, it is particularly unmarked; at least in three instances. First, it is unmarked when a vowel appears in the environment of immediately following a nasal, the vowel acquires the feature [+NASAL]; i.e; it is nasalised. Examples include:
(224) /mama/ → /mimina/ →
[mãmã]
mother
[mĩmĩnã]
pour into
/mumu a/ → [mũmũ ã]
suck
/ ani/ →
[ ãnĩ]
baboon
/ a ua/ →
[ ã ũa]
raise, lift
/nusu/ →
[nũsu]
half
/atuka/ →
[ãtuka]
retire voluntarily (borrowed from Zanaki language of Tanzania)
In all these cases it is evident that the nasal coming before the vowel has a direct effect on it and therefore: (225) / V/ → [+nasal]
N ____
222
This rule may be written formally as: (225a) + syll - const → [+ nas]
+ const + son - cont
________
The second instance (and one which is more or less like the above) is realized when a vowel appears immediately before a nasal consonant. Again, in such a circumstance, the vowel is nasalised as evident in the following examples:
(226) /amua/ → [ ãmũa]
decide
/amka/ → [ãmka]
wake up
/anuai/ → [ãnũai]
various
/ini/
liver
→ [ĩnĩ]
/imara/ → [ĩmãra]
firm
/uma/ → [ũmã]
fork
/unabii/ →[ũnãbii]
prophesy
/u ama/→[ ũ ãmã]
brutality
Looking at the gloss in (224) and (226) above, it is easy to conclude that the motivation for nasalisation lies with the anatomy of the articulators. In (224), the articulators, the velum mostly, do not move into place fast enough after nasal articulation in order to avoid the vowel following the nasal getting nasalised. In (226), nasalisation arises from anticipation and early adjustment. In anticipation of the articulation of a nasal consonant appearing immediately after a vowel; the articulators (mainly the velum, again) adjust
223
themselves early thus nasalising the vowel preceding the nasal. The rule for this may be stated as:
(227) /V/→[+nasal] — N
Formally, the rule may be posited as: (227a) +syll - const
→ [+nas]
—
+const + son - cont
Massamba (1996: 92) attempts what he deems as a general rule for nasalisation. It appears as a collapsed version of the two rules above. His rule appears thus:
+syll -cons → [+nas]
— +const. +nas +const. ---+nas
Although this rule aptly captures the two nasalisation processes discussed above, it is, however, not a general rule for vowel nasalisation in Kiswahili. This is because there is a third instance where Kiswahili vowels are nasalised. This is when the vowels appear after a cluster of a nasal and constant; more so if the consonant has been harmonized to the nasal. Examples to this end include:
224
(228) / vur / →[ vũr ]
wooden bowl
/ vuk /→[ vũk ]
steam
/ vi/→[ vĩ]
grey hair
/ga /→[gã ]
shield
/gariba/→ [gãriba]
circumciser
/ndumba/→[ndũmbã]
magic
/mbari/→[mbãri]
clan
For gloss (228) above, the rule may be formulated thus: (229) /V/→ [+nasal]
N+C — homorganic
When this rule is formalized, it may be posited as: (229a) + syll - const
→ [+nas]
+ cosnt + nas
+
+cosnt - syll
____
______________________ Homorganic Formulating one general rule for nasalisation under NGP may prove problematic because an attempt to do so would lead to an un-economical and clumsy rule. It has further been noted that this process has very obvious leanings towards suprasegmental phonology although in itself, it is a segmental process. Arising from the aforegoing, in the next subsection, this study attempts an application of the Autosegmental Phonological Theory to the process as it appears in Standard Kiswahili.
225
4.6.1 An Application of APT to Nasalisation in Standard Kiswahili Like vowel harmony, nasalisation is usually captured clearly using the APT. Going back to the gloss, the rules may be formulated as: (230) (a) [+nas]
[-nas]
[+nas]
[+ nas]
[-nas]
→ m
a
m
a
m a
m a
/mama/
[mãmã] mother
(b) [+nas] [-nas] [-nas]
[-nas]
[+nas] [-nas] →
m
k
m
k [m k ] hearth
/m k /
(c) [+nas] [-nas] [+nas] [-nas]
[+nas] →
a /
a a/
[
226
ã] suck
(d) [+nas] [-nas] [-nas]
[+nas] [-nas] →
ŋ
ŋ
a
a [ŋ a] uproot
/ ŋ a/
In the above examples, the nasal preceding the vowel influences the articulation of the vowel thereby nasalising it. In the examples below, the rules denote that the nasal following a vowel influences the preceding vowel; since in these cases the nasal is followed by another vowel, the second vowel is also influenced. As a result, the nasal influences two vowels simultaneously, thus: (231) (a) [-nas] [+nas] [-nas] [-nas]
a
m
a
u
[+nas] [-nas]
→
a
m u
a
[ãmũa] decide
/amua/
(b) [-nas] [+nas] [-nas]
[+nas] →
i
n
i
i
n
i
[ĩnĩ] liver
/ ini/
227
(c) [-nas] [+nas] [-nas] [+nas] [-nas]
[+nas] →
u
a
m
a /u
u
a
ma
[ũ ãmã] brutality
ama/
(d) [-nas] [+nas] [-nas]
[+nas] →
u
m
a
u
m
a [ũmã] fork
/uma/
The last rule involves the cluster of a nasal, followed by a harmonized consonant and a vowel. Both the harmonized consonant and the vowel acquire the feature [+NASAL] as in:
(232) (a) [+nas] [-nas] [-nas] [-nas] [-nas]
[+nas] [-nas] →
m
v
u
k
v u k [ vũk ] steam
/mvuk /
228
(b) [+nas] [-nas] [-nas] [-nas]
[+nas] [-nas] →
n
g
ŋ g
u
u [ŋg u] blood, red soil
/ ŋg u/
(c) [+nas] [-nas] [-nas] [+nas] [-nas] [-nas]
n
d
n
d
→
[+n a s ]
i
n
d
n d i [nd ndĩ] boxing
/nd ndi/
(d) [+nas] [-nas] [-nas] [+nas] [-nas]
m
b
n
→
i
[+nas]
m b
ni [mb nĩ] (pupil)
/mb ni/
Again, APT is able to posit rules for all the nasalisation processes evident. However, to paraphrase Katamba (1989:212), these last two processes; vowel harmony and nasalisation, constitute a theoretically fascinating phenomenon which can throw light on the nature of phonological representation especially due to the way they function, partly as segmental and partly as suprasegmental properties.
229
4.7 Concluding Remarks This chapter has a comprehensive review of the various vowel sound changes. NGP theory has been used in the analysis of vowel coalescence, vowel deletion, glide formation, harmony and nasalisation. While it was possible to easily posit NGP rules for most of the prosses, it was different in the case of vowel harmony and nasalisation. As a result, APT has been used to formulate non-linear rules.
Notes 1. Perhaps the exception to this rule involves the cluster of the low vowel and the high front vowel. As already demonstrated in section 4:2, such a cluster causes vowel coalescence and not deletion. 2. The markedness theory has been used for a long time. The Prague school circle, especially, was very fond of it. Even before Kiparsky (1974), it was used (with minor differences) by the likes of Chomsky & Halle (1968) and Lockwood (1969, 1972) among others. 3. Most linguists, who include, Bleek (1869), Jacottett (1896), Meinhof (1932) and Guthrie (1970/71), among others, generally agree that {mu-} is the actual prefix in proto-Bantu. {m-} is a phonetic realization. 4. Of course the feature [+ROUNDED] here is in reference to the rounding of lips. One may, however, wish to point out that though / / is also a rounded vowel, the sequence /m+ / is prevalent in Standard Kiswahili, and that it occasions no deletion. The justification for this state of affairs is that / /, being a mid-low vowel is much less rounded in comparison to the high vowel /u/, and, consequently, / / is less labial than /u/. 5. See section 3:4 on palatalization. 6. See the section on 5(above) for a broader exposition on this issue. 7. Sommerstein (1977:255) refers to rules governing such a process as fast-speech rules.
230
8. Kenstowicz further observes that vowel harmony exhibits many of the ―action-ata-distance‖ properties displayed by tone. Tone is a prosodic feature best studied under an autosegmental (non-linear) theory. It is for this reason that this study will apply APT vis-à-vis vowel harmony in the next sub-section. It is, however, important to note that there exist differences between vowel harmony and the suprasegmental properties like tone and stress. This, as Katamba (1989) points out, is because the harmonizing phonological features are normally part of the segmental representation of individual vowels. Further, it is worth noting that the process functions partly as a segmental and partly as a suprasegmental property, hence the application of the two theories. 9. The main cause of this problem is the suprasegmental nature of the process. This study‘s solution is an attempt at using the APT theory. 10. It is worth mentioning that as Kenstowicz (1994) states, in glosses (214) and (215); the harmonic effect of the verbal base vowel propagates from the base through the suffix to the end of the various demonstratives. 11. Katamba (1989) suggests that the major aim of using formal phonological rules is to capture generalizations, achieve economy and to enhance simplicity.
231
CHAPTER V Phonology of Borrowed Lexicon 5.1 Introductory Remarks Unlike in the previous chapters where this research concentrated on intra-language sound changes in Standard Kiswahili, the current chapter focuses on an inter-language description. As happens with all other languages and dialects, Standard Kiswahili borrows words heavily in order to keep pace with new developments occurring in areas where the language is used.
Borrowing is evidently a component of language growth and is thus a continuous process. Antilla (1972) rightly observes that it is one of the major factors that influence language change. As a result, as Mberia (1993) observes, it is studied in the realm of historical and comparative linguistics since it is a diachronic study. Although this research is largely synchronic, a study of borrowed vocabulary has been deemed necessary owing to several factors.
One such factor concerns phoneme inventory. It is appreciated that borrowed forms become part of Standard Kiswahili. Some forms may retain sound segments alien to Standard Kiswahili. Such segments would of necessity be added to the existing inventory. Since this study is interested in the segments, it is important to see what influence borrowing has had on the segment inventory.
232
Second is the issue of syllable structure. It was noted in chapter two that syllable structure is an important component in the study of phonological processes.
It is
therefore important to understand the original syllable structure of borrowed words and see whether any change has taken place; to explore whether the structure of one of the two languages influences the other or whether it is both ways. By and large, these issues have a direct bearing on the formulation and positing of phonological rules. Lastly, and to a lesser degree of importance, is the urge to know; the interest to understand the sound changes that occur as a result of borrowing.
It is, of course, noted that NGP, the key theory used in this study, advocates for the study of borrowed vocabulary. As noted in section 1.9.1, NGP‘s P-rules do apply to loan-words adaptation since they are not suppressable. These factors justify a chapter on diachronic study in a segmental research.
As indicated in section 1:8 the languages from which lexical examples are drawn from include Arabic, Persian, Indian, Portuguese, Turkish and English. The corpus analysed in this chapter is got from published works which mainly (but not exclusively) include Chiraghdin and Mnyampala (1977), Massamba (1987, 2004), Bosha (1993) and Tuki (1990, 2000, 2001 and 2004).1
233
5.2
Phonology of Consonants
The current study has found out that there are three major consonantal changes that occur when words are borrowed from other languages into Standard Kiswahili. These changes result from substitution, insertion and deletion of consonants.
5.2.1
Consonant Substitution
Consonant substitution is a very dominant feature of borrowed words in Standard Kiswahili. One of the major triggers of the substitution is the existence of a non-Standard Kiswahili phoneme in the borrowed lexicon. The phoneme is usually replaced by one which is phonetically close to it in Standard Kiswahili. This may be explained phonologically since using a very different phoneme would suggest a different word that is disimilar to the one that is borrowed. Of course the issue of phonological perception also comes up where language users perceive phonemes in terms of the ones that they know. A good example of a sound that is substituted is sound / / which is usually replaced by //. Sound /q/ is also substituted by /k/ a well as /h/ in some instances. Examples include:
(233) Kiswahili word
Source word
Source Language
Gloss
irni
a:rr
Arabic
neighbour
aribu
aarrib
Arabic
try
English
agenda
a nda
nd
234
a nti
e nt
English
agent
arida
:n
English
journal
kamani
qaman
Persian
tabak l
tabaqu ira
Persian
snuff box
kauli
qaw
Arabic
statement
kijama
qija:ma
Arabic
doomsday
iktisadi
iqtisa:d
Arabic
economy
akraba
aqa:rib
Arabic
relatives
baha i i
baq i
Arabic
tip/gratuity
main spring of a clock/watch
Some phonemes appear to be less favoured in Standard Kiswahili. Such phonemes are usually, but not always, substituted. As is natural, they are replaced by phonetically close ones. The two less favoured phonemes in Standard Kiswahili seem to be / / and /x/. It is however noted that these phonemes appear in most other Kiswahili dialects especially the Zanzibari ones. This may lead to the observation that dialectal forms appreciate borrowed words more. Even though these phonemes do occur in some borrowed standard words (such as /xa:/ - denoting anger or digust; /xalifa/ - caliph; / ala/ - storehouse and / asia/ commotion), they are replaced in others. It is thus not phonologically clear why they are less favoured given that they are easy to articulate and perceive. A few examples to illustrate this include:
235
(234) Kiswahili word
Source word
Source Language
Gloss
gari
adi
Indian
vehicle
gati
at
Indian
dock/wharf
baha a
bu a
Turkish
envelop
turuhani
turuxan
Persian
prestige/respect
alhamisi
alxami:s
Arabic
Thursday
hamsini
xamsi:n
Arabic
fifty
laki
lax
Indian
one hundred thousand
ta:hari
ta ax-xar
Arabic
be late, delay
Sounds /g/ and /h/ substitute / / while /x/ is usually replaced by /h/ and less frequently by /k/.
While the substitution of the above segments may be clearly explained; (given that both /g/ and /h/ are more native to Standard Kiswahili than / / and /x/) and is fairly systematic, in some other cases the substitution is neither frequent, consistent nor systematic. Such cases include the change of / / to / /, the weakening of /d/ to /r/; change of /s/ to /z/, /r/ to /l/, /s/ to //,/h/ to /r/, /n/ to /m/, /j/ to /i/ and even /p/ to /b/ as is evident in the following corpus of borrowed words:
236
(235) Kiswahili word
Source word
Source language
gurudumu
kwurdun
Persian
tyre
diri a
diri a
Persian
window
almari
armari
Portuguese
chest of drawers
m za
m sa
Portuguese
table
parafu
parafus
Portuguese
screw
bal zi
baljus
Turkish
ambassador
k r k r ni
k hk n
Turkish
gaol
af ndi
ef ndj
Turkish
adi
Indian
vehicle
gari
Gloss
Title used by soldiers to superiors
tarumb ta
tr mpIt
English
trumpet
kandarasi
kntrkt
English
contract
It is important to note that in most of the lexical items analysed, more than one process is evident. At this juncture however, only consonant substitution has been focused on.
5.2.2
Consonant Insertion
Consonant insertion is not a dominant process in Standard Kiswahili when compared to substution. However, it is a feature in a few lexical items. Given the nature of Kiswahili as a language which prefers the use of vowels to consonants, and has a preferred CVCV structure, it is not strange that consonant insertion is not widely evident. Examples of lexical items that denote the process include: 237
(236) Kiswahili word
Source word
Source Language
Gloss
Skurubu
skru:
English
screw
vi
Portuguese
wine
‗izz
Arabic
might/honour
vi nzi
Here, consonants /b/, / / and /n/, respectively, have been inserted. Even in these few instances, there does not seem to be any motivating phonetic reason for the insertion. Lexical structures such as /skru:/*, /vi
/* and / zi/* do appear to be acceptable in
Kiswahili phonology.
An interesting sort of consonant insertion involves the phoneme /g/. When // appears as a word final in the source language, a vowel is inserted. This epenthesis, as is now familiar, is aimed at inhibiting the occurrence of a closed syllable at the word final position. However, at the same time as the insertion of the vowel, consonant /g/ is also inserted thereby changing both the syllabic and phonetic shape of //. Instead of realizing the syllable /i/ (CV) for example, a syllable /gi/ (CCV) may be realized as is the case with /ra/ (Persian/ Indian, for colour or paint) whose borrowed form is /ragi/.
5.2.3 Consonant Deletion Consonant deletion is yet another of the rare and inconsistent phenomenon in Standard Kiswahili borrowed words. The only deletion that appears to be consistent is that of
238
phoneme /h/ when it appears at the final position of a word. 2 /j/ is also deleted fairly regularly. The process is also realised where a cluster of consonants appear in the borrowed word. The data below indicate the deletion of consonants /r/, /n/, /p/ /k/ and /x/ in addition to /h/ and /j/.
(237) Kiswahili word
Source word
Source language
Gloss
bafta
baftah
Persian
calico, lawn
randa
randah
Persian
plane
hab dari
hab rdari
Persian
beware!
bima
bijma
Indian
insurance
hundi
hundir
Indian
cheque
baia
baija
Indian
cake of lentils and pepper
daima
da:iman
Arabic
perpetually
ta:hari
ta ax-xar
Arabic
be late, delay
sukari
sukkar
Arabic
sugar
l s
l ns
Portuguese
a piece of cloth
bal zi
baljus
Turkish
ambassador
batiza
bptiz
English
baptize
Apart from the avoidance of occurrence of a closed syllable at the end of the word, the other major motivation for deletion seems to be avoidance of double and multiple consonant clusters. Such occurrences at times seem alien to Standard Kiswahili as is the
239
case with the word /ta hax-xar/ and /bptiz/. However, such argument is not full-proof as in most cases when such clusters appear vowel insertion is mostly favoured. A lexical structure such as /baputiza/* does not appear unfamiliar. What is clear is the fact that deletion occurs in an effort to ‗Swahilise‘ different lexical items. Of course, in regard to NGP, deletion in the corpus analysed above may be said to serve two purposes. The first one is to allow for ease of articulation by reducing the number of consonants and consonant clusters. The second reason is the attainment of economy in the various items. Deletion actually seems to occur at both the phonetic and orthographic levels.
5.3
Phonology of Vowels
As is the case with the consonants, the most prevalent of the processes in vowels seems to be an attempt to achieve the preferred syllable structure. The other reason that elicits phonological changes is the avoidance of the occurrence of closed syllable structures especially at the end of a word. As a result, unlike in the case of consonants where substitution was the most prevalent process, the most prevalent process in vowels is, naturally, insertion. Other evident processes, though to a much lesser degree, are substitution and deletion.
5.3.1
Vowel Substitution
Vowel substitution is generally not widespread in Standard Kiswahili. Where it occurs, the process is used to replace vowels that are non-existent in the language such as //, /o/,
240
/e/, //, // as well as the various diphthongs present in some source languages like English. In the case of Arabic (which has a three vowel system), Standard Kiswahili at times substitutes to form diversity. This is because Standard Kiswahili has a richer system of vowels in comparison to Arabic. However, there are substitutions that may not be explained since they lack obvious phonetic or phonological motivation triggering them. Such is the case involving the substitution of / / with /u/ in the item /pilau/ (pilaf) an item borrowed from Persian /pila /. However, when this particular case is examined critically under NGP, an explanation may be offered. The phoneme /i/ is near-tense while / / is a near-lax vowel. In the discussion in section 4.5 it was found out that [TENSE] is a harmonising feature in Standard Kiswahili. Consequently, it may be argued that the substitution of the / / with the /u/ is aimed at realising vowel harmony between it and the near-tense /i/. Examples of substitution include:
(238) Kiswahili word aki
Source word :k
Source language
Gloss
English
chalk
kal nda
klnd
English
calendar
pilau
pila
Persian
pilaf
r ani
ru n
Persian
balcony
r
r i
Portuguese
ace
s ti
s t
Portuguese
wah di
wa:hid
Arabic
241
seven (in playing cards) one
zaraa
zira:a
Arabic
agriculture
tafakari
tafakkara
Arabic
ponder
abiri
abra
Arabic
travel/sail
limu
ilm
Arabic
education
nzi
izz
Arabic
honour, power
The substitution evident in the above corpus does not seem to be consistent. Phonological sense would require that the vowel replacing the existing one be as close as possible to the one being replaced in terms of distinctive features. However, this is not always the case. There are instances where a low vowel is replaced by a high vowel (like in the case of /abiri/ from /abara/) or vice-versa (as in /zaraa/ from /zira: a/). The expectation would be that a high (or low) vowel would, at worst, be replaced by a midvowel. At times, reverse substitution is evident as in the case of /u/ and / / in /pila / which changes to /pilau/ and /ru n/ which changes to /r ani/. No motivation of such a change is fathomable. Phonological sense (at least under NGP) dictates that if one vowel is used to replace another (like in the case of / / for /i/), then the change should be consistent in all cases where the replaced vowel appears. This does not, however, seem to be the case. In most cases the source vowel is retained in the loanword. Given the theoretical framework used in this study, this inconsistency is baffling and does not seem to have any justifying factors to trigger it.
242
5.3.2
Vowel Insertion
Vowel insertion is a more dominant process in Standard Kiswahili borrowed words. It serves two main purposes. Firstly, it impedes the occurrence of a closed syllable at the end of a word and secondly it enhances the CV structure. It therefore realises the open syllable structure throughout the word. Examples include: (239) Kiswahili word
Source word
Source language
Gloss
bandari
bandar
Persian
harbour, port
k di
k d
Persian
rent
r ani
r n
Persian
balcony
turuhani
turxan
Persian
prestige/respect
bagiri
b agri
Indian
bracelet
gati
at
Indian
dock/wharf
laki
lax
Indian
a hundred thousand
ragi
ra
Indian
paint, colour
karata
karta
Portuguese
playing cards
bal zi
baljus
Turkish
ambassador
baha a
bu a
Turkish
envelop
s li
s l
Turkish
sergeant major
limu
ilm
Arabic
education
fahamu
fahm
Arabic
understand
243
kitabu
kita:b
Arabic
a book
daftari
daftar
Arabic
a notebook
rais
ras
Arabic
President
wah di
wa:hid
Turkish
one
aki
:k
English
chalk
daktari
d kt
English
doctor
futi
ft
English
foot
paipu
paip
English
pipe
rip ti
rip :t
English
report
skuli
sku:l
English
school
tani
t n
English
tonne
English
office
afisi/ofisi
fis
Vowel insertion is a more understandable process in Standard Kiswahili given that it is a Bantu language.
Bantu languages strongly prefer the CVCV syllable structure and
Standard Kiswahili is no exception. Most of the languages from which this study‘s corpus is drawn seem to favour a system where consonants dominate and which have closed vowel structures. This further explains why consonant insertion is so limited in Standard Kiswahili. One of the more consistent forms of insertion is prothesis. This is usually evident in the borrowing of names that refer to countries. It is noteworthy that most of them are borrowed from English (although some, especially those that came into contact with
244
Kiswahili early are borrowed directly from the source languages).
In both cases
however, the insertion of /u/ at the word initial position seem quite consistent as in: (240) Kiswahili word
Source word
Source language
Gloss
ufaransa
ƒra:ns
English
France
uturuki
t:k
English
Turkey
uhispania
spen
English
Spain
English
Germany
English
England
u rumani uig rza/wig rza3
:mn glnd
For those country names that entered Kiswahili in later years (especially African), the rule does not seem productive. Most of them retain their source forms such as /ka/, /k g /, /s malia/, /zambia/, and /afrika kusini/ among others.
However, the question as to why the insertion rule is not spread to all items while it seems fairly well spread still remains unanswered. In fact, further spread of the rule would mean more familiar structures and attainment of CVCV syllable structure as would be the case in: (241) Kiswahili word
Hypothetical word
Gloss
/daftari/
/dafutari/*
notebook
/daktari/
/dakitari/*
doctor
/alhamisi/
/alahamisi/*
Thursday
245
/bafta/
/bafuta/*
calico
/t kn l ia/
/t kin l ia/*
technology
/ inki afi/
/iniki afi/*
revelation
5.3.3
Vowel Deletion
As a process, vowel deletion is very limited in Standard Kiswahili. This is understandable since the language greatly favours the use of vowels. However, a few cases of vowel deletion have been noted. Examples include:
(242) Kiswahili word
Source word
Source language
Gloss
almari
armari
Portuguese
chest of drawers
bramu
biram
Portuguese
flag, banner
uru
Portuguese
diamond (in playing cards)
uru nadra
na:dir
Arabic
rare
In most of the above examples, deletion serves the purpose of impeding the occurrence of a double vowel cluster. This is understandable given Standard Kiswahili‘s preference to the CVCV structure.
However, the case of /nadra/ (rare) is both rare and unique. It therefore calls for special mention. This arises out of the following: Firstly, the deletion of /i/ is not motivated by the occurrence of a double vowel cluster and secondly (and more baffling to this
246
researcher) is the fact that the deletion goes against the syllabification preferences of the language. While /nadir/ has the CV$CV$C structure, the resultant /nadra/ deviates from the same and actually forms a closed syllable (i.e. CVC$CV).
No phonologically
partinent explanation may be offered using the selected theories to justify this process as regards this particular word. It would have been more natural to just insert a vowel as the word final to block the occurrence of a closed vowel. However, while this is done, the deletion of /i/ causes another closed syllable. These are some of the lexical items that NGP refers to as exceptions (and for which the theory allows room for). One may further add that this is clearly an Mp-rule which occurs in very few items and which items are actually marked for it.4
There is one other item that is worth mentioning here. The word /psa/ (money) is borrowed from Indian and its original form is /paisa/. Instead of deletion being realized, a different process, vowel coalescence, is. It means that the word was adopted in its original form and it consequently adapted the Standard Kiswahili rule which militates against the clustering of the low vowel followed by the high front vowel. Consequently, the cluster /ai/ coalesce to be realized as the mid-vowel / /. Again, this is a case of avoiding a vowel cluster amidst other phonetic factors that hinge on the anatomy of the articulatory organs.5
5.3.3.1 Vowel Shortening The issue of the existence of long vowels in Standard Kiswahili has not been handled conclusively so far. However, one thing that is clear from the cited corpus is that when
247
items with long vowels are loaned to Standard Kiswahili; the long vowel is usually (but not always) shortened. According to NGP, this is an opaque form of vowel deletion aimed at limiting the occurrence of vowel clusters. To exemplify vowel shortening, some words borrowed from Arabic and English are revisited below: (243) Kiswahili word
Source word
Source Language
Gloss
daima
da:iman
Arabic
perpetually
aifu
da i:f
Arabic
weak
nadra
na:dir
Arabic
rare
wahdi
wa:hid
Arabic
one
salamu
sala:m
Arabic
greeting
kitabu
kita:b
Arabic
book
rip ti
rip :t
English
report
aki
:k
English
chalk
skuli
sku:l
English
school
ufaransa
fra:ns
English
France
English
journal
arida
:nl
It appears that shortening of vowels in borrowed words is a rule occurring regularly and it is quite spread. It in fact allows for the attainment of the preferred vowel structure.
248
5.4
Remarks on Syllable Structure
It is important to note that this section does not portend to indulge into an in-depth study of the syllabification of borrowed vocabulary. Such a study would not be justifiable in this research given the theoretical approaches applied. In such a study, CV phonological theory would be a better choice in comparison to NGP and APT. In fact, as has been the case in the earlier parts of this supplementary chapter, no rules will be posited concerning syllabification.
As stated in the objectives, the purpose is to explore the sound adjustments that occur on loanwords and to investigate how the adjustments relate with Standard Kiswahili morphology. In view of the scope, and because of the continuous process of lexical borrowing, a systematised study in the area of borrowed vocabulary, its rules and syllabification is suggested.
It is, however, important to have an over-view of the syllable structure of borrowed words in this study. This arises out of several factors, core among them being the following: One, the syllable is a very important phonological/morphological unit. Syllables, according to Katamba (1989) regulate the combination of segments and also control the combination of features which make up segments. It is the unit in whose terms the phonotactic rules are stated. These are the rules which reflect the language users‘ understanding of which sound combinations are permissible in their language and which ones are not. Consequently, when new rules enter the language, (as part of language growth), it is phonologically justifiable to highlight them.
249
The second reason hinges on the chosen theoretical approaches. NGP relates to this chapter at two levels. The first level concerns the P-rules. These rules, it has been observed, are productive and cannot be suppressed. More important to this section, the rules influence foreign language acquisition and apply to loanwords adaptation. In this section therefore, the effect of P-rules on syllable structure and the notion of universality will be explored.
The second NGP level involves the MP-rules. Under these rules, the syllabification parameters are applied. These parameters assign syllable boundaries to phonological strings. The MP-rules are language-specific and they will be very helpful in this section. APT, as a suprasegmetal theory, also allows for the study of syllabification and is, in a way, a precursor to the CV theory.
A closer look at the syllable structure of loanwords in Standard Kiswahili reveals two important points: 1.
A great attempt is always made to maintain the original syllable structure of Standard Kiswahili in loanwords.
This has been done through the use of the processes studied earlier in this chapter. Such an attempt is not peculiar to Standard Kiswahili. Massamba (1987:142) observes that: Different languages have what one might call ―restricted‖ sound sequences and sound combination. Each language tends to arrange its sounds in its own fixed way. To this effect, within phonological systems of
250
languages, certain sound combinations and sound sequences are disallowed. True to the above observation, as is evident in section 2:2:4, many of the borrowed words have adapted to Kiswahili syllable structure. 2.
A few syllable structures alien to Standard Kiswahili have, all the same, entered into Standard Kiswahili and formed part of its structure.
This is not unique to Kiswahili. In the comment quoted above, Massamba (1987) goes further and adds that the comment is only true for languages that are either fully developed, and hence ―stable‖ or languages that have not had any contact with other languages. The latter scenario is becoming less and less likely. This is so because due to social mobility, education, scientific and technological advancements, language contact has become almost unavoidable. Apart from the effect such contact has had on other levels of languages, those on syllable and other phonological structures are glaring. Given the position of Standard Kiswahili in the world today (see section 1:1) it is even more pronounced in the language. A further observation by Massamba (1987:142) is: Languages that are just beginning to emerge as national or international languages seem to be affected by the turmoil of modernisation in the structure of their phonological systems. (Emphasis added) Standard Kiswahili is one such language and it has thus not escaped what Massamba (1987) refers to as ―the turmoil of modernisation‖. Although the occurrence of closed syllables is a rare phenomenon in native Standard Kiswahili words, the structure is widespread among loanwords as is evident below: (244)
251
/bafta/
calico
(CVC$CV)
Persian
/bah i i/
tip
(CVC$CV$CV)
Persian
/almari/
chest of drawers (VC$CV$CV)
Portuguese
/tkn l ia/
technology
(CVC$CV$CV$CV$V)
English
/inspkta/6
inspector
(VCC$CVC$CV)
English
/alfab ti/
alphabet
(VC$CV$CV$CV)
English
/akraba/
relatives
(VC$CV$CV)
Arabic
/halma auri/
board
(CVC$CV$CV$V$CV)
Arabic
/inki afi/
revelation
(VC$CV$CV$CV)
Arabic
/alhamisi/
Thursday
(VC$CV$CV$CV)
Arabic
The acquisition of a closed syllable system is one of the more dominant features that have been accepted in the phonological structure of Standard Kiswahili. In the above gloss (and in all other cases), what is put into consideration is the articulation of the various words and not their orthography. In other words, the phonetic information has been given preference over the written forms.
The other alien syllable structure that has entered Standard Kiswahili through loanwords is that of a cluster of three consonants and a vowel (CCCV) as evident in: (245) /sprigi/
spring
(CCCV$CCV)
English
/skrubu/
screw
(CCCV$CV)
English
252
But even as such borrowed words retain some of their original structure, there is a great attempt to give them the structure of the recipient language. In (245) above, for example, vowel insertion has occurred in the first item while in the second, there is consonant insertion. These processes are aimed at ―Swahilising‖ the loanwords.
It is, therefore, fair to note that while constraints on syllable structure serve as a filter, allowing only certain sound sequences to occur in a particular language, (Katamba 1989) the sequencing changes and, at times, even accommodates the sequences of the source language. This happens even though, as noted, each language has constraints specific to it where a well formed syllable in language X may not be so in language Y.
Even with the acquired syllable structures, loanwords are almost always nativised and this nativisation exemplifies the importance of syllables, their structures and rules in a language. 5.5
Concluding Remarks
From the above exposition, it has been evident that rules such as those involving substitution, insertion and deletion are involved in the process of nativisation of loanwords and the acquisition of the preferred syllable structure.
It is further evident that borrowing has had, though to a minor degree, some effect on the segment inventory of Standard Kiswahili. Segment /X/ is becoming more and more accepted as a Standard Kiswahili phoneme. On syllable structure, it has been found out
253
that borrowing has had effect at least at two levels: the realisation of closed syllables and the realisation of the CCCV structure. Although Standard Kiswahili has tried hard to maintain its syllable structure, nonetheless, it has had to accede to new structures. However, after ―breaking the structural norms‖ new norms are formed and they become part of the core syllabification norms of the language.
This best explains the strength of the P-rules on the syllable structure. Any rules in a language remain the only rules to the extent that no new rules and processes have entered a language at a given time. Otherwise, the rules of any language are very dynamic and are perpetually prone to change.
As has been evident, Standard Kiswahili is no
exception.
Notes 1.
In all these works, Standard Kiswahili dialect has been used.
2.
This is meant to check the realization of a closed syllable at the end of a word. It is noted that in most such words, instead of the occurrence of apocope, epenthesis is evident where a vowel is inserted thus leading to the preferred open vowel structure.
3.
In the case of /uigr za/, there is an initial vowel cluster involving the two high vowels where one has the value [+] for the feature [BACK] and the other has the opposite value for the feature. As has been mentioned (see section 4.4), Standard Kiswahili does not prefer such clusters; hence the phonetic realisation of the item as /wigr za/. The initial vowel /u/ is hardened into /w/ in the process known as glide formation.
4.
This rule seems to go against the grain. While in most loanwords Standard Kiswahili strive to avoid the occurrence of closed syllables (though they be
254
present in source words) and to maintain the CVCV structure; here is a case where the CVCV structure is destroyed in favour of the (unfamiliar) closed syllable. This is truly baffling. 5.
For a detailed account concerning this process and exceptions to coalescence rule, see section 4:2.
6.
Here is a case of a word having five consonants and only three vowels; quite an unfamiliar combination in native Standard Kiswahili words.
255
CHAPTER VI Research Conclusions 6.1
Introductory Remarks
This is the concluding chapter of the study. It attempts a synthesis of the conclusions that are evident in the research. The chapter has three sections. The first section details a summary of the findings of the research. These findings are offered on a chapter by chapter basis. In the same sections, comments are made concerning the hypotheses that were offered at the beginning of the research. Each hypothesis is analyzed vis-à-vis the relevant chapter or section and the outcome is laid down. The second section offers the contributions this research has made in the field of academia and linguistics in general and in the study of Standard Kiswahili morphophonology in particular. In the last section, the chapter offers recommendations on the areas that this research has established require further research and analysis. In order to avoid repetition, detail has been avoided in this chapter as much as possible.
6.2 A Summary of Findings and Remarks on Hypotheses At the phonological level, this study has established that Standard Kiswahili phonemes fall into three categories: vowels, semi-vowels and consonants. Standard Kiswahili has five vowels and various allophones based on the features concerning nasalization and height. A further observation is that the phonemes that Mohamed (2001:2) refers to as Kiswahili ―secondary vowels‖ are merely allophones of different phonemes based on vowel height. It has further been concluded that Standard Kiswahili phoneme /a/ is a mid, near back vowel.
256
As far as semi-vowels are concerned, it has been concluded that Standard Kiswahili has two semi-vowels; /j/ and /w/. These have been identified using articulatory parameters. The same parameters have been used to conclude that /h/ is not a semi-vowel in Standard Kiswahili but is clearly an obstruent and, therefore, a consonant. Concerning the consonants, this study has observed that the issue concerning the number of consonants in Standard Kiswahili is a contentious one and that figures rage from 23 to 27. Using mainly articulatory parameters, this study has identified 25 consonant phonemes. It has, however, noted that two of the 25 (/X/ and //) are contentious but advances reasons for their inclusion.
The distinctive features of all the 32 Standard Kiswahili phonemes identified by this study have been offered and a summary given in matrices. On the issue of aspiration, this research has concluded that it is not a distinctive feature in Standard Kiswahili though it is an important feature in other Kiswahili dialects such as Kimvita and Kiamu. It is, however, an area in need of further research since the users of the other dialects may gradually introduce the feature into Standard Kiswahili.
The research has further
concluded that allophones do exist in Standard Kiswahili but the parameters of their identification are different from those used in other languages like English. The main trigger of allophony in Standard Kiswahili has been found out to be nasalization (and not aspiration).
Concerning the issue of syllables, it has been concluded that Standard Kiswahili prefers the CVCV syllable structure. Nine syllable structures have been identified and two out of
257
the nine have been found out to be later entrants into Standard Kiswahili. This has been occasioned by borrowing following language growth. These two structures include the closed syllables and those with a CCCV structure.
It has been noted that syllable
structure has a major impact on sound change since sound changes result from neighbourhood of phonemes.
At the morphological level, two word types have been sampled. These are the nouns and verbs. The first observation made is that Standard Kiswahili subscribes to concordial agreement. Standard Kiswahili nouns have been found to belong to various types. These include proper, abstract, collective, common and concrete nouns. They may further be classified as animate or inanimate as well as countable or uncountable. Three major nominal structures are identified. The first include word roots which are complete in form and meaning without affixation. Then there are the prefixed nouns which comprise of a prefix and nominal root and the last structure results from nominalization thereby giving rise to nouns that contain prefixes, the root and suffixes.
This study notes that several approaches are used to classify Standard Kiswahili nouns; chief among them the morphological/structural approach and the syntactic approach. The semantic approach is not widely used. The study concludes that each of the approaches has its weaknesses. To counter these weaknesses, this study proposes the use of an alternative method which it refers to as the eclectic approach (section 2.3.3.4 and example 38). The proposed approach combines the morphological, syntactic, semantic
258
and numbering/numerical approaches. A further observation is that the borrowed words usually deviate from the aforementioned nominal structures, typology and classification.
Verbs have been identified as keywords in Standard Kiswahili. While all other word types may be deleted in a sentence, verbs are never deleted. A further observation is that, being an agglutinating language, Standard Kiswahili words have many affixes. Verbs are the word types that are most prone to affixation. Three verb types have been identified. These are the main, auxiliary and copular verb types. At the structural level, verbs comprise of roots, prefixes (where 9 morph positions have been identified) and suffixes (where 3 verbal suffixes are evident).
This research also explores consonantal sound changes evident in Standard Kiswahili. Five processes have been identified.
These are lenition, fortition, palatalisation,
homorganic nasal assimilation and deletion. It has been concluded that deletion may result from historical reasons or from the Ganda law. Trills and nasals are also deleted under a specific environment. Concerning this chapter, the first hypothesis stated that sound changes in Standard Kiswahili follow a specific, consistent and predictable pattern. This hypothesis has been found to be only partially true. While most sound changes do, some do not. Many exceptions have been evident. These exceptions deviate from the expected patterns. Examples and reasons have been offered. Items like [mtafiti], [mstahiki] and [m indi] evidently deviate from the lenition rule.
259
The second hypothesis stated that sound changes in Standard Kiswahili consonants are brought about by other consonants when the affected consonants occur in neighbouring environments. This hypothesis has also been found to be only partially true. While some consonant processes are caused by consonants, others are caused by vowels.
Consonants are responsible for fortition in some cases as is evident in rules (137) and (137a). In other cases, vowels are responsible for fortition as exemplified in (142) and (142a). In Standard Kiswahili, lenition is usually caused by vowels when in neighboring environments with plosives during nominalisation hence rules such as (131) and (131a). Palatalisation and some forms of deletion are other processes that have been found to be triggered by vowels at times.
However, homorganic nasal assimilation is caused by consonants as indicated in rule (168). Nasal deletion in Standard Kiswahili is also purely a consonantal process as is evident in rules (187) and (187a). There are, in fact, more consonant processes caused by a vowel and consonant neighbourhood than those caused by consonant-consonant neighbourhood.
Five vowel processes have been identified in this research. The processes are vowel coalescence, vowel deletion, glide formation, vowel harmony and vowel nasalization. The use of NGP on vowel harmony and nasalization has proved inadequate and as a result APT has been applied. This second theory clearly describes the two processes.
260
It is also noted that there is a problem in formulating the coalescence rule using NGP but an attempt has been made. It has, however, been observed that the process is quite natural and is spread in Standard Kiswahili although there are a few exceptions. Exceptions, it has been noted, are acceptable under NGP theory and are actually seen as a fortification of the theory.
A further conclusion is that vowel deletion is triggered by two factors; firstly to avoid the occurrence of vowel clusters and secondly as a result of redundancy. Glide formation has been found to be a natural rule which is widespread in Standard Kiswahili.
In vowel harmony, it has not been possible to posit a rule under NGP. In demonstratives, the reduplication of vowels is seen as a form of morph copying, hence complete harmony. In verbal extensions, the vowel found in the stem always determines that of the suffix. The vowel to be realized is always determined by the value of feature [TENSE] of the vowel occurring in the stem, this is partial harmony since the vowel is not always reduplicated.
Vowel nasalization is yet another naturally occurring process. It has been found to occur in three environments. Firstly, when the vowel occurs in the environment of immediately following a nasal; secondly, when the vowel appears immediately before a nasal and thirdly when the vowel appears after a cluster of a nasal and a consonant; especially where harmonization has occurred.
261
Concerning the first hypothesis, it is observed that, unlike in the consonant processes, most of the vowel processes are very consistent and do follow a specific pattern, and hence quite predictable. Vowel harmony is one of the most consistent and predictable processes. Vowel nasalization occurs quite naturally and is thus very consistent and predictable. Vowel processes follow a particular pattern. Coalescence, deletion and hardening are all aimed at avoidance of the occurrence of a vowel cluster. However, there are a few exceptions. Redundancy for example does not seem to be spread, consistent nor predictable. It has also been noted that there are instances where coalescence fails to take place while it would naturally seem appropriate.
The third hypothesis states that sound changes in Standard Kiswahili vowels are brought about by other vowels when the affected vowels occur in neighbouring environments. This hypothesis has been proved to be true to a very large extent. Unlike in consonants, most vowel processes arise as a result of being in the neighborhood of other vowels. Processes such as vowel coalescence, deletion occasioned by clusters, glide formation and vowel harmony are all occasioned by vowels sharing neighbouring environments. The only exceptions are deletion occasioned by redundancy (which lacks any clear guiding parameters) and vowel nasalization which is evidently caused by nasals, thus consonants.
This research has also studied the phonology of borrowed vocabulary. It has been found out that there usually is a major attempt to ―Swahilise‖ or nativise loanwords entering Standard Kiswahili. It is in the process of nativisation that sound changes occur. Three
262
changes have been identified in both vowels and consonants.
These changes are
substitution, deletion and insertion. The sound changes are motivated by three reasons. Firstly, they are aimed at avoiding the occurrence of a double consonant cluster, secondly they prevent the occurrence of closed syllables and lastly they obliterate the occurrence of a double vowel cluster. The sound changes are evidently an attempt at realization of a CVCV syllable structure but all the same some alien structures have now been acquired by Standard Kiswahili. The new syllable structures identified include CCCV and the closed syllable. Consequently, it is right to state that the hypothesis that the sound system of borrowed vocabulary in Standard Kiswahili is adjusted in such a way as to commensurate with the native Standard Kiswahili vocabulary is only partially true. There are times when loanwords enter Standard Kiswahili with their structures leading to the conclusion that nativisation is not always achieved even as sound adjustments almost always occur.
To sum up this section, a remark on the adequacy of the theoretical framework applied in this research is in order. On the whole, the use of NGP and APT as well as SPE‘s distinctive features has proved adequate for this research. It is, however appreciated that none of the theories would have been adequate on its own. NGP for example could not be competently applied to the analysis of vowel harmony and vowel nasalization. Under APT, however, an in-depth analysis was possible. On its part, APT could not be applied to the analysis of the other processes. Formulation of vowel coalescence rule under NGP also proved problematic as is evident in example (194). NGP, however, aptly describes the other processes.
263
6:3 Contributions of the Study This study has comprehensively and systematically studied Standard Kiswahili phonemes. Their distinctive features have been analysed leading to the positing of both vowel and consonant matrices. The phoneme inventory that has been developed will make the learning of the phonemes easy and fast and will ultimately lead to a better understanding of the language at the phonological level.
The research has further comprehensively studied nominal and verbal structures and typology. Since the issue of nominal classes has been found to be confusing; where each classification has been seen to have its weaknesses, this study has posited and recommended an eclectic classification. This sort of classification will be of help to both the morphologists and syntacticians in particular and to Standard Kiswahili linguists in general.
The analysis of vowel and consonant processes will lead to a better understanding of the language and the findings of these analyses may be very helpful to one attempting a diachronic study of Standard Kiswahili phonology.
It may also be helpful in
understanding the phonology of loanwords. Standard Kiswahili has not fully stabalised. As a result, knowledge of the phonological structures of both borrowed and native vocabulary is very important in understanding the language at this stage of its unprecedented growth. This is deemed of paramount importance given the rapid growth and spread of Standard Kiswahili. It is thus crucial to understand all its levels. This is what this study has assisted in doing.
264
To this researcher‘s knowledge, this is the first systematic and comprehensive study on Standard Kiswahili morphophonology. It would be a great asset in a comparative study of languages at the level of phonology, morphology and morphophonology. It puts Standard Kiswahili more conspicuously in the world‘s linguistic map.
The use of a suprasegmental theory to study segmental features is a pioneering endeavour in Standard Kiswahili. It has proved that some segmental processes which are not easily accounted for using segmental theories are very well analysed using suprasegmental theories such as APT. It has also been possible to test the validity of the two theories used vis-à-vis Standard Kiswahili morphophonology.
In fact, the
combination of a generative theory, generative distinctive features and a suprasegmental theory to form the theoretical framework used in this study indicates that there is an interface of such theories and that such a framework is more competent in an analysis such as the current in comparison to an individual theory.
This study therefore sets a paradigm and provides a comparative scale for other Kiswahili dialects in particular and African languages in general.
It is yet another
contribution to the understanding of African linguistics.
6:4
Recommendations for Further Research
One of the areas that this study suggests require further research concerns aspiration in Standard Kiswahili. Such a research would require use of respondents and even a comparative study vis-à-vis other Kiswahili dialects. It would be necessary to investigate
265
whether aspiration is present in Standard Kiswahili; and if it is, whether it is a distinctive feature leading to realisation of phonemes or whether it is a feature of complementary distribution leading to the realisation of allophones. This study identified nasalisation as the feature leading to allophony in Standard Kiswahili. It also mentioned that the feature [TENSE] may lead to allophony. This latter feature requires further research. Also necessary is an investigation to determine whether there are other features that cause allophony in Standard Kiswahili.
Related to the issue of aspiration is the issue of the existence of aspirated phonemes such as /ph/, /th/, /k h/ and / h/ in Standard Kiswahili. Other phonemes that needs to be investigated further are /x/ and / / - to establish if they are still productive in Standard Kiswahili or if they have been blocked.
The issue of double consonants, especially after harmonisation, also requires further investigation.
It is important to investigate if they are single phonemes or double
phonemes or whether their status changes. If the status change, then it would be important to investigate what triggers the change and the environment under which it occurs.
The issue of the morphological nominal classes 12 & 13 ({ka} and {tu}) also requires investigation. It is important to establish whether Kiswahili language has ever had the two classes. If so, then it would be enlightening to know what motivated their loss and if there are indications of their (re-)emergence in Standard Kiswahili. Maybe one way of
266
doing this is reference to classical Kiswahili texts. It is notable that the two classes are highly spread among other Bantu languages. Palatalisation process also requires further investigation to establish why many items that seem positively marked for the process do not palatalise.
On vowel harmony, it is important to investigate whether there are more harmonizing features other than the feature [TENSE] which was identified in this study.
More research also needs to be applied to the study of the phonology of borrowed vocabulary. It is important to lay down rules and to also investigate the environments under which various phonological processes occur.
6.5 Concluding Remarks This is a synchronic, segmental study. It would be exciting to have a research on diachronic morphophonology of Standard Kiswahili. The current study may be used as a base leading to a backward movement and thus the reconstruction of probable protoforms. Lastly, a study in the suprasegmental phonology of Standard Kiswahili is also suggested as necessary. This would mean that the whole spectrum of Standard Kiswahili phonology would be covered.
267
BIBLIOGRAPHY
Abdulaziz, M.H. (1971)
―Tanzania‘s National Language Policy & the Rise of Swahili Political Culture‖ in Whiteley, W (Ed) Language Use and Social change, Nairobi: OUP.
___________
Muyaka: The 19th Century Swahili Poet, Nairobi: KLB.
(1979)
_______& K.N. Osinde (1997)
―Sheng and English: Development of Mixed Codes Among Urban Youth in Kenya‖ in M.H. Abdulaziz (ed) International Journal of the Sociology of Language, Berlin: Mouton de Gruyter.
Abercrombie, D. (1967)
Elements of General Phonetics, Edinburgh: EUP.
Akmajian, A. et al (2001)
Linguistics: An Introduction to Language and Linguistics, New Delhi: Prentice-Hall of India,.
Amidu, A.A.
Classes in Kiswahili, Koln: Rudiger Koppe Verlag.
(1997)
Anderson, J.M &C.J. Ewen (1987)
Principles of Dependency Phonology, Cambridge: CUP.
Antilla, R. (1972)
An Introduction to Historical and Comparative Linguistics, New York: The Macmillan Comp.
Armstrong, L. (1967)
The Phonetic and Tonal Structure of Kikuyu, London: Downs of Hall Mall.
Ashton, E.O. (1944):
Swahili Grammar (Including Intonation), London: Longman.
_____________ (1982)
Swahili Grammar (Including Intonation), London: Longman: Green & Co. Ltd.
Bakari, M.
(1982)
The morphophonology of Kenyan Swahili Dialects, Nairobi: unpublished Ph.D Dissertation, University of Nairobi.
___________
(1985)
The morphophonology of Kenyan Swahili Dialects, Berlin: Dietrich Reimer Verlag.
268
Bosha, I. A.
(1993)
Taathira za Kiarabu Katika Kiswahili Pamoja na Kamusi Thulathiya (Kiswahili-Kiarabu-Kingereza), Dar es Salaam: Dar es Salaam University Press.
Bennet, P.R.
(1985)
―Dahl‘s Law and the Thagicu‖ in African Language Studies, Vol VII, pp 127-159.
Bleek, W.
(1869)
Comparative Grammar of South Languages, London: Trubner & Co.
African
Brain, J.L. (1966,1977)
Basic Structures of Swahili, Syracuse: Syracuse University Press.
Broomfield, G.W (1931)
Sarufi Ya Kiswahili, London: the Sheldon Press,
____________
(1975)
Sarufi ya Kiswahili, Nairobi: OUP.
Brown, G.
(1972):
Phonological Rules Cambridge: CUP.
Bubenik, V.
(2003)
An Introduction to the Study of Morphology, Muenchen: Lincom Europa.
Burt, F.
(1917)
Swahili Grammar and Vocabulary, 2nd Ed, London, S.P.C.K.
Bwenge, C.M.T. (1994):
―Historia Fupi ya Sera ya Lugha ya Kiswahili Nchini Tanzania‖ in Chaka, The Journal of African Kiswahili Association (CHAKA), No. 1/1994, pp 18-23.
Bybee, J.L. (1985)
Morphology: A study of the Relation between Meaning and Form, Amsterdam/Philadelphia: JBPC.
Chimerah, R.
Kiswahili: Past Present and Future Horizons, Nairobi: NUP.
(2000)
and
Dialect
Variations,
Chiraghdin, S. & M. Myampala (1977) Historia ya Kiswahili, 2 nd Ed, Nairobi: OUP. Chittick, H.N.
(1969)
―An Archaeological Reconnaissance of the Southern Somali Coast‖ in Azania, Vol.IV, pp 113131.
___________
(1974)
Kilwa: An Islamic Trading City on East African Coast, Nairobi: BIEA.
269
___________
(1975)
East Africa and the Orient: Cultural Syntheses in the Pre-Colonial Times ed by H.N. Chittick and R.I. Rotberg, New York: African Publishing Company.
___________
(1979)
―Manda and the Immigration of the ‗Shirazi‘‖, Discussion Paper No 119, Institute of African Studies, University of Nairobi.
___________
(1984)
Manda: Excavations at an Island Port on the Kenyan Coast, Nairobi: BIEA.
Chomsky, N & M. Halle (1968)
The Sound Pattern of English, New York: Harper & Row.
Churma, D.C.
Arguments from External Evidence in Phonology, New York: Garland Publishing.
(1985)
Clark, J. & C. Yallop (1995)
An Introduction to Phonetics and Phonology, 2nd Ed, Malden: Blackwell Publishing.
Clements, G. & S. Keyser (1983)
CV Phonology, Cambridge: MIT Press.
Crystal, D.
(1985)
A Dictionary of Linguistics and Phonetics, Oxford: Basil Blackwell.
___________
(1991)
A Dictionary of Linguistics and Phonetics, 3rd Ed., Oxford: Blackwell Publishers.
Durand, J.
(1990)
Generative and Non-Linear Phonology, London: Longman.
Ferguson, C.
(1961)
―Assumption About Nasals: A sample study in phonological Universals‖, in J.H. Greenberg (Ed) Universals of Language, Cambridge: MIT Press, pp53-60.
_____________
(1978)
―Phonological Processes‖ in J.H. Greenberg (Ed) Universals of Human Language, Vol. II, Stanford: SUP, pp 403-442.
Fromkin, V. & Rodman (1988)
An Introduction to Language, New York: CBS College Press.
Gibbe, A.G.
―Uchambuzi wa Nomino za Ngeli ya Kwanza‖ in Lugha ya Kiswahili, Dar es Salaam: TUKI.
(1983)
270
Goldsmith, J.
(1976)
Autosegmental Phonology, MIT, Doctoral dissertation, Bloomington: distributed by Indiana University Linguistics Club.
____________
(1979)
―The Aims of Autosegmental Phonology‖ in Dinnsen, D. (Ed) Current Approaches to Phonological Theory, Bloomington: IUP, PP 202222.
____________
(1990)
Autosegmental and Metrical phonology, Oxford: Basil Blackwell.
Greenberg, J.
(1949)
Studies in African Linguistics Classification: The Niger-Congo Family, Albuquerque: University of New Mexico.
___________
(1966)
The Languages of Africa, 2nd Ed, Bloomington: Indiana University Press.
Guthrie, M.
_________
(1967)
The Classification of the Bantu Languages, London: Dawson of Pall-Mall.
(1970/71)
Comparative Bantu, Vols 2,3 &4, Farnborough: International Publishers Ltd.
(1999)
Discourse Analysis of Kiswahili Political Speeches, Unpublished Ph.D dissertation, Nairobi: University of Nairobi.
____&P. Karanja (2004)
Misingi ya Sarufi ya Kiswahili, Nairobi: Phoenix Publishers.
Halle, M. & G. Clements(1983)
Problem Book In Phonology, Cambridge: MIT Press.
Hartman, R.
(1972)
A Dictionary of Language and Linguistics, London: Applied Science Publisher.
Hayes, B.
(1986)
―Inalterability in CV Phonology‖ in Language, Vol. 62, pp 321-351.
Habwe, J.H.
Heine, B.
(1970)
Status and Use of African Lingua Francas, Munchen: Weltforum Verlag.
271
(1977)
―Phonetic Analysis in Phonological Description: Prenasalised Consonants and Meinhof‘s Rule‖ in Lingua Vol. 43 pp 339-373
Hinnebusch, T. (1973)
Prefixes, sound change and the sub-grouping in Coastal Kenyan Bantu Languages, Ph. D. dissertation, Los Angeles: University of California.
___________ (1982)
Studies in classification of Languages, Sugia Supplement.
Hooper, J.
(1975)
―The Archi-Segment in Natural Generative Phonology‖ in Language, Vol. 51, pp 536-556.
___________ (1976)
An Introduction to Natural Generative Phonology, New York: Academic Press Academic Press.
___________ (1979)
―Substantive Principles in Natural Generative Phonology‖ in D. Dinnsen (Ed) Current Approaches to Phonological Theory, Bloomington: Indiana University Press, pp 106-125.
Hudson, R.
(1981)
Sociolinguistics, Cambridge: CUP.
Hyman, L.
(1972)
A Phonological Study of Fe Fe Bamileke, Los Angeles: University of California Press.
___________ (1975)
Phonology: Theory and Analysis, New York: Holt, Rinehart & Winston.
Jakobson, R et.al (1952)
Preliminaries to Speech Analysis: The Distinctive Feature and Their Correlates, Cambridge: MIT Press.
_______& M. Halle (1956, 1971):
Fundamentals of Language, The Hague: Mouton.
Jones, D.
(1957)
History and Meaning of the Phoneme, London: Longman.
Kapinga, M.C. (1983)
Sarufi Maumbo ya Kiswahili, Dar es Salaam: TUKI.
Katamba, F.
An Introduction to Phonology, London: Longman.
Herbert, R.
(1989)
___________ (1993)
Eastern
Morphology, New York: St Martin‘s Press
272
Bantu
Kenstowicz, M. (1994)
______& C. Kisseberth (1977)
Phonology in Generative Grammar, Cambridge: MA Blackwell. Topics in Phonological Theory, New York: Academic Press.
Kihore, Y et.al (2001)
Sarufi Maumbo ya Kiswahili Sanifu, Dar es Salaam: TUKI.
Kiparsky, P.
How Abstract is Phonology, Maas: MIT Press.
(1968)
________
(1973)
―Abstractness, Opacity and Global Rules‖ in O. Fujimura (Ed) Three Dimensions of Linguistic Theory, Tokyo: TEC.
_________
(1974)
―On the Evaluation Measure‖. In Papers From the Parasession on Natural Phonology, Chicago: Chicago Linguistic Society.
Kombo, D. & D. Tromp (2006)
Proposal and Thesis Writing: An Introduction, Nairobi: Paulines Publications Africa.
Krapf, L.
(1882)
Swahili-English Dictionary, Tubingen: Trubner & Co.
Ladefoged, P. (1962)
Elements of Acoustic Phonetics, London: University of Chicago Press.
Lass, R.
(1984)
Phonology, Cambridge: C.U.P.
Lockwood
(1969)
―Markedness and Stratificational Phonology‖ in Language, Vol. 45, pp 300-308.
__________
(1972)
Introduction to Stratificational Linguistics, New York: Harcourt Brace Jovanovich.
Loogman, A.
1965)
Swahili Grammar & Syntax, Pittsburg: Dequensne University Press.
Lyons, J.
(1968)
Introduction to Theoretical Linguistics, Cambridge: C.U.P.
Maganga, C.
(1990)
The Morphophonology of Standard Kiswahili, Kimakunduchi, Kitumbatu and Kipemba,
273
Unpublished Ph.D dissertation, Dar es Salaam: Dar es Salaam University. Maringa, E.
(1987)
A Comparative Analysis of Verbal Extensions in Kimbeere and Kiswahili, Unpublished M.A. dissertation, Nairobi: University of Nairobi.
Massamba, D. (1986)
―The influence of local languages on Kiswahili: The Case of Mara Region in Tanzania‖ in Kiswahili, Vol. 53/1, Journal of The Institute of Kiswahili Research, pp 67-83.
___________ (1987)
―The Effects of Language Modernization on the Phonological System of Kiswahili Language‖ in Kiswahili, Vol. 54/1, Journal of The Institute of Kiswahili Research, pp 142-151.
___________ (1996)
Phonological Theory: History and Development, Dar es Salaam: Dar es Salaam University Press.
___________ (2002)
Historia ya Kiswahili: 50 BK hadi 1500 BK, Nairobi: JKF
___________ (2004)
Kamusi ya Isimu na Falsafa ya Lugha, Dar es Salaam: TUKI.
_________ et. al (2004)
Fonolojia ya Kiswahili Sanifu, Dar es Salaam: TUKI.
Matthews, P.H . (1997)
Oxford Concise Dictionary of Linguistics, New York: OUP.
Maundu, M.P.
(1980)
Sound Change and the Reconstruction of Kikamba Consonant Sounds in Natural Generative Framework, Unpublished M.A. Thesis, Nairobi: University of Nairobi.
Mbaabu, I.
(1978)
Kiswahili: Lugha ya Taifa, Nairobi: KLB.
274
__________
(1985)
Sarufi ya Kiswahili, Nairobi: Kenya Publishing & Book Marketing.
__________
(1992)
Sarufi ya Kiswahili, Nairobi: Longman.
Mberia, K. wa (1993)
Kitharaka Segmental Morphophonology With Special Reference to the Noun and the Verb, Unpublished PhD dissertation, Nairobi: University of Nairobi.
Mdee, J.S.
Kiswahili: Muundo na Matumizi yake, Nairobi: Intercontinental Publishers.
(1986)
Meeuseen, A. (1980)
Bantu Lexical Reconstructions, Tervulen: MRLC.
Meinhof, C.
Introduction to the Phonology of the Bantu Languages, Berlin: Dietrich Reimer, pp 111-113.
(1968)
____&N.J. van Warundo (1932)
Introduction to the Phonology of the Bantu Languages, Berlin: Dietrich Reimer Ernst Vohsen.
Mfaume, G.E. (1984)
Misingi ya Isimu ya Lugha ya Kiswahili, Dar es Salaam: Utamaduni Publishers.
Mgullu, R.S. (1999, 2002)
Mtalaa wa Isimu: Fonetiki, Fonolojia na Mofolojia ya Kiswahili, 1st & 2 nd Eds, Nairobi: Longhorn.
Microsoft
―Encarta Africana‖ in MS Encyclopaedia
(1998)
Mohammed, M. (2001)
Modern Swahili Grammar, Nairobi: E.A.E.P.
Mreta, A. Y.
(1998)
An Analysis of Tense and Aspect in Chasu: Their form and Meaning in the Affirmative Constructions, Hamburg: LIT Verlag.
Mutahi, E.K.
(1977)
Sound Change and Classification of the Dialects of Southern Mt. Kenya, Nairobi: Unpublished PhD Dissertation, UON.
275
_________
(1980)
―Language and Politics in Kenya: 1900-1978‖, Staff Seminar Paper, Nairobi: UON.
_________
(1983)
Sound Change and Classification of the Dialects of Southern Mt. Kenya, Berlin: Dietrich Reimer Verlag.
Mwangi, P.I. (2002)
―Mofofonolojia ya Kiswahili: Nyufa Pana Katika Uchapishaji Wake‖, Nairobi: Unpublished paper presented at an International Kiswahili Conference.
_________
―Swala la Alofoni (Zipo au Hazipo) Katika Ufundishaji wa Fonolojia ya Kiswahili‖ in Njogu, K. (Ed) Kiswahili na Elimu Nchini Kenya, Nairobi, Twaweza Communications, pp 93-102.
(2007)
. Newman, P. (1968)
―The Reality of Morphophonemics, in Language Vol. 44, pp 507-515.
Njogu, K. et.al (2006)
Sarufi ya Kiswahili: Uchanganuzi na Matumizi, Nairobi: JKF.
Njuguna M. (1992)
Mofofonolojia ya Kiswahili Sanifu na Kikuyu Sanifu: Mathalani Kikuyu cha Kabete: Ulinganishi, Unpublished M.A. Thesis, Nairobi: University of Nairobi.
Nkwera, F.V.M. (1978)
Sarufi na Fasihi kwa Sekondari na Vyuo, Dar es Salaam: T.P.H.
Nurse, D.
―A Linguistic Reconsideration of Swahili Origins‖ in Neville, H. & J. Sulton, Azania: Journal of the British Institute in East Africa, Vol. XVII pp 127150.
(1985)
________&T. Spear (1985)
The Swahili: Reconstructing the History and the Languages of an African Society, 800-1500 AD, Philadelphia: University of Pennsylvania Press.
Odden, D.
The Phonology and Morphology of Kimatuumbi, Oxford: Clarendon Press.
(1996)
Okombo, D.O. (1982)
Dholuo Morphophonemics in a Generative Framework, Berlin: Dietrich Reimer Verlag.
276
__________ (2001)
Language Policy: The Forgotten Parameter in African Development and Governance Strategies, Nairobi: UON Press.
OUP & TUKI (1981, 2004)
Kamusi ya Kiswahili Sanifu, 1st & 2 nd Eds., Nairobi: OUP.
Polome. E .
(1967)
Swahili Language Handbook, Washington D.C: C.A.L..
Reusch, R .
(1954)
A History of East Africa, Stuttgart: Evang. Missions Vertag. .
Richards, J.et al (1985)
A Dictionary of Applied Linguistics, London: Harlow.
Ruo, K.R.
―Historia Fupi ya Maendeleo ya Sera juu ya Kiswahili Nchini Kenya (1900-1989)‖ in Chaka, The journal of African Kiswahili Association (CHAKA) No 1/1994, pp 13-14.
Schane, S.A.
(1994)
(1968)
―On the Non-Uniqueness of Phonological Representation‖, in Language, Vol. 44, pp 709ff.
__________ (1972)
―Natural Rules in Phonology‖ in R.P. Stockwell & R.K.S. Macaulay (eds) Linguistic Change and Generative Theory, Bloomington: Indiana University Press.
__________
Generative Phonology, New York: Prentice Hall, Eaglewood-Cliffs.
(1973)
Seth, J. & Dhamija (2004)
A course in Phonetics & Spoken English, New Delhi: Prentice-Hall of India.
Sommerstein, A .(1977)
Modern Phonology, London: Edward Arnold.
Steere, E.
A Handbook of Swahili Language as Spoken In Zanzibar, London: S.P.C.K.
(1870)
Thiong‘o, N. wa (1986)
Decolonising the Mind, Nairobi: Heinemann.
_____________ (1993)
Moving the centre: The Struggle for Cultural Freedoms, Nairobi: E.A.E.P.
277
Trubetzkoy, N.S. (1969)
Principles of Phonology, Los Angeles: University of California Press.
TUKI
(1990)
Kamusi Sanifu ya Isimu na Lugha, Dar es Salaam: TUKI.
__________
(2000)
English – Swahili Dictionary, 2nd Ed, Dar es Salaam: TUKI.
__________
(2001)
Kamusi Ya Kiswahili- Kingereza, Dar es Salaam: TUKI.
Vennemann, T. (1972,a)
―Rule Inversion‖ in Lingua, Vol 33, PP 137 – 156.
__________
―On the Theory of Syllabic Phonology‖, In Linguistsche Berichte. Vol 18, pp 1-18.
(1972,b)
__________ (1974,a)
―Phonological Concreteness in Natural Generative Grammar‖ in R. Shuy & C. Bailey (eds.) Towards Tomorrows Linguistics, Washington D.C: Georgetown University Press, pp 202 -219.
__________ (1974,b)
―Words and Syllables in Natural Generative Grammar‖ in Parasession on Natural Phonology, Chicago: Chicago Linguistic Society, pp 346-374.
Verma, S. & N. Krishnaswamy (1989) Modern Linguistics: An Introduction, New Delhi: OUP. Waihiga, G.
(1999)
Sarufi Fafanuzi ya Kiswahili, Nairobi: Longhorn.
Whiteley, W.
(1969)
Swahili: The Rise of a National Language, London: Methuen & Co Ltd.
Yule, G.
(1996)
The Study of Language, 2nd Ed, Cambridge: CUP.
278
