The Bantu language Swahili (or Kiswahili) belongs to the Sabaki subgroup of Northeast Coast Bantu. Its nearest genetic relatives include Comorian, spoken on the Comoro Islands, and the Mijikenda languages of Kenya, among others.
First-language (L1) speakers of Swahili, who probably number no more than two million, refer to themselves by such names as Washirazi, Wabajuni, Watumbatu, or Wamakunduchi; relatively few L1 speakers use the term “Waswahili” (wa- Class 2 animate plural prefix, -swahili < Arabic sawāḥil ‘coast’, thus ‘people of the coast’). At least seventy-five million people speak Swahili as a second language, ranging from standard varieties to highly pidginized ones. (For reference, see Prins 1961, Hinnebusch 1979, Wald 1987, and Nurse and Hinnebusch 1993.)
1. Dialects and variants
Swahili is dialectally diverse. Its major divisions, with some of their individual members, include the following (areas where each is spoken are in parentheses; see also Map 1):
(i) ChiMwiini (Barawa, Somalia)
(ii) Northern Swahili: KiAmu, KiPate (Lamu Archipel-ago, Kenya); KiTikuu (Lamu Archipelago, Somalia coast); KiMvita (Mombasa, Kenya)
(iii) Southern Swahili: KiUnguja, and dialects on Zanzibar Island and Pemba, and others spoken along the Tanzanian coast
(iv) KiMwani (Cabo Delgado and Kirimba Archipelago, northern Mozambique)
(v) KiNgwana, etc. (eastern Democratic Republic of the Congo, hereafter “Congo”)
(vi) Standard Swahili (East Africa)
There are growing communities of L1 ‘standard’ speakers in East African urban areas, and in eastern Congo. There are also pidginized variants throughout East Africa and Congo. Some of these have acquired their own names, e.g. KiSetla, associated with European settlers and their laborers in Kenya, or KiHindi, spoken by Indian immigrants. KiSerikali ‘government language’, and other pejorative terms have been used by L1 speakers for standardized Swahili.
Mutual intelligibility is dependent on the usual range of variables: geographic location, exposure to other communities, knowledge of standard variants, age, sex, etc. Thus, while most East African and Congolese speakers of L1 dialects of Swahili tend to understand Standard Swahili, speakers of Standard Swahili may have difficulty in understanding geographically isolated dialects like KiTikuu. ChiMwiini and KiMwani, spoken at the extremes of the continuum, are different enough to impede intelligibility; this leads some to set them apart as closely related, but non-Swahili, dialects of Sabaki.
Several stages in the development of Swahili have been identified (Nurse and Spear 1985):
(a) Pre-9th century: Pre-Swahili developed from earlier Sabaki roots in its northern homeland; communities were small farming/fishing settlements on the Lamu Archipelago, with some trading contacts with non-Swahili Islamic seafaring peoples.
(b) 9th–12th century: Swahili-speaking settlements of farmers, fishermen, and traders spread southward, where they worked iron and built towns with mosques and buildings of coral-rag and lime mortar; this period established the foundations of Swahili society and culture.
(c) 12th–16th century: the Indian Ocean trade expanded; its eventual demise resulted from Portuguese interference in the 16th century.
(d) 19th century: Zanzibari entrepreneurs expanded trade to the coastal hinterland as far west as the Congo. During this period, migrants from Arabia, Persia, and India settled on the coast, adding their cultural and historical contributions.
(e) European colonization began in the second half of the 19th century, and ended beginning in the early 1960s with independence.
(f) The post-colonial period continues today in East and Central Africa. Swahili serves as the official and/or national language.
Each of these periods had its impact on the language—most obviously in extensive borrowing from Arabic, Persian, Portuguese, English, and Indian languages.
3. Standard Swahili
is based on the KiUnguja dialect of Zanzibar Town. In the 19th century, with growing trade controlled from Zanzibar, KiUnguja spread first to adjacent coastal areas and islands, and then along the caravan trading routes through Tanganyika to the Congo. The earliest Europeans to arrive found KiUnguja or closely related variants already well established. In parts of Kenya, the closely related KiMvita dialect came into use. German colonists' choice of Swahili as an administrative language in Tanganyika reflected the fact that it was already a viable lingua franca; the British continued similar language policies in both Kenya and Tanganyika, for similar reasons. By the 1930s, a committee had been established to codify and elaborate a standard, to develop a modern orthography, and to supervise the publication of books for use in education. It also published traditional texts and descriptions of dialects. This work continued until the early 1960s. After independence, it was replaced by the National Swahili Council in Tanzania, the Institute of Swahili Research at the University of Dar es Salaam, and a similar body in Zanzibar. Their efforts, among others, have concentrated on developing modern terminology for education and government. Proposals for a similar institute have been discussed in Kenyan academic circles.
Standard Swahili is diverse. An influential form is used throughout Tanzania. A continuum of varieties can be found, ranging from the Tanzanian and Kenyan standards, which dominate print and broadcast media, to morphologically simplified, pidginized forms.
4. Literature and orthography
Swahili has an important literary tradition that goes back at least four centuries, with surviving manuscripts in verse from the 1700s (Harries 1962, Whiteley 1969:38, Abdulaziz 1979, Knappert 1979). The tradition began in the northern dialect area, and then spread south to Mombasa and elsewhere. Though it owes some debt to Arabic poetic forms, it is not a mere imitation of those models, either thematically or formally; it has adapted and changed in form, style, and content. Much of the poetry written in the classical styles is not accessible to many L2 speakers of Swahili, but poetry—often political and satirical—is a regular and popular feature in Kenyan and Tanzanian Swahili newspapers. There is also a growing popular literature.
Swahili was first written with a modified Arabic script. This is still used by older people, Koranic teachers, and some Swahili poets; however, since the 1930s, it has generally been supplanted by a Roman orthography.
Arabic, Persian, and other languages of the Indian Ocean basin, as well as several European languages, predominantly English, have contributed words. The distribution of these borrowings reflects the cultural impact of the lending communities: from Arabic have come terms of religion, seafaring, jurisprudence, trade, and non-indigenous flora and fauna, e.g. dini ‘religion’ or mshahara ‘wage’; from English, modern technology, government, etc., e.g. motokaa ‘automobile’ or gavana ‘governor’. Loans, with some exceptions, are incorporated into the native structure. Thus consonant clusters are usually disallowed, as in brush > burashi. Nouns are classified morphologically, with most assigned to N/N [9/10] or ∅/MA [5/6] classes. (Noun class numbers are given in brackets.) Verbs are inflected (Arabic -silimu ‘become a Moslem’, -silim-ik-a ‘be converted’, -silimish-w-a ‘be made a Moslem’; Zawawi 1979). Borrowing has increased the phonological inventory, and has added some adverbials and conjunctions (kabla ‘before’, baada ‘after’, au/ama ‘or’), but the basic Bantu structure is intact, with little indication of creolization.
Standard Swahili has the vowels i e a o u, and the consonants shown in Table 1 (see Polomé 1967).
Table 1. Swahili Consonant Phonemes. Parentheses indicate elements introduced in Arabic loanwords. Angle brackets enclose orthographic equivalents.
ɲ ǰ 〈nj〉
The aspirated stops are not found in all L2 dialects; they can be morphophonemically derived from underlying sequences of N plus C̥(NC̥ → NC̥h → C̥h). Some L2 speakers also neutralize the distinction between /l/ and /r/, e.g. /mahari/ ‘bride wealth’ and /mahali/ ‘place’ → [mahari]. NC clusters are generally realized phonetically as prenasalized stops: [mb,nd,ɲǰ,ŋg]. Swahili is non-tonal, and has penultimate stress. Some dialects depart from this inventory; e.g. the northern dialects have /t̯/ and /nd̪, instead of /č/ and /nǰ/.
The Standard orthography does not mark aspiration. Except for adding aspiration where relevant, examples hereafter are cited in standard orthographic form.
Swahili is rich in morphophonemic alternation. Examples follow (for a discussion of class, see section 7 below):
(a) Nominal prefixes frequently show alternation between forms which occur with C-initial and with V-initial stems. This involves various processes, including vowel rules (contraction, assimilation, devocalization, etc.), consonant rules, and others; e.g. vi-∼vy- in vi-tu vy-angu  ‘my things’.
(b) Non-aspirated consonants may alternate with aspirated ones, e.g. ukuta  ‘wall’, khuta  ‘walls’; the aspirated variants are derived from underlying sequences of /NC̥/.
(c) Other stem-initial alternations are sensitive to specific nominal prefixes, or to underlying morphological features: thus /ɗ∼r∼nd/ in defu ‘tall’ , warefu , ndefu [9/10].
(d) A type of vowel harmony exists between verb stems and their suffixes: thus the suffix /-i(1)/ is realized differently in -fik-i-a ‘arrive at’, -amb-i-a ‘say to’, vs. -som-e-a ‘read for’, -elek-e-a ‘point to’; the reversive suffix /-u(l)-/ in -fung-u-a ‘open’ vs. -chom-o-a ‘pull out’; or the stative suffix /-ik-/ in -vunj-ik-a ‘be broken/breakable’ vs. -som-ek-a ‘be read/readable’.
(e) Stem-final consonant alternations are associated with the causative inflectional suffix (-y- < PB *ı̨), and with the agentive nominalizing prefix: -ogop-a ‘be afraid’, -ogof-y-a ‘frighten’; -lew-a ‘be drunk’, -lev-y-a ‘make drunk’; -tot-a ‘be soaked’, -tos-a ‘make wet’ (<*-tot-<ı̨>-a); -lip-a ‘pay’, m-lif-i ‘payer’; -ib-a ‘steal’, mw-iv-i ‘thief’; -pik-a ‘(to) cook’, m-pish-i ‘(a) cook’.
(f) ∅ alternates with /l/ in certain verbal affixes conditioned by stem-final V or C: -ruk-i-a ‘jump at’, -nunu-li-a ‘buy for’; -chom-o-a ‘draw out’, -chom-o-le-a ‘draw out for’.
Swahili classifies nouns into formal, semi-semantically defined noun classes or genders. These are typically arranged in singular/plural pairs, and are numbered accordingly. (The numbering of classes reflects an overall Bantu system; classes 12–13 are absent in Swahili.) A gender is defined in terms of an underlying constellation of phonological, morphological, and syntactic/semantic features; these determine surface forms, and, through agreement in the shape of a “concord,” are spread to other sentential constituents. Typical examples of gender prefixes with their major alternates are shown in Table 2. The shape of a prefix is either ∅, C, or CV-; roots are C-initial or V-initial, and may be monosyllabic, disyllabic, or polysyllabic.
In Swahili studies, the genders are referred to either by the shape of the noun prefix—e.g. M/MI Class [3/4], N/N class [9/10]—or by the shape of the verbal concord they control, e.g. I/ZI [9/10]. Though pairwise noun sets are common, not all genders have plurals (e.g. abstract nouns in Class 14, such as u-zuri ‘beauty’, u-huru ‘freedom’, u-toto ‘childhood’); nor do all plurals have corresponding singulars (e.g. mass nouns in Class 6, such as ma-futa ‘oil, fat’, ma-ji ‘water’, ma-ziwa ‘milk’).
In clausal and sentential environments, gender features are spread to other constituents. Adjectives, demonstratives, possessives, interrogatives, and verbs are brought into obligatory agreement with their controlling noun. In addition, subject nouns spread concord to verbs; objects also do this, but non-obligatorily:
(1) Wa-le wa-kulima wa-zuri wa-na-(ya)-lima
2-those 2-farmers 2-good 2-are-(6)-cultivating
ma-hindi mashamba-ni kw-ao.
6-maize 17-farms-loc 17-their
‘The/those good farmers are cultivating maize at their farms.’
Bantuists distinguish nominal and pronominal/verbal concord. Nominal concord, marked by a prefix on the noun, is spread to adjectives, including some numerals; nearly everything else is pronominal concord (see Table 2).
Table 2. Swahili Noun Classes
m- / mw-
m-tu ‘person’, mw-alimu ‘teacher’
a- / yu-
wa- / w-
wa-tu ‘people’, w-alimu ‘teachers’
wa- / w-
m- / mw-
m-ti ‘tree’, mw-itu ‘forest’
u- / w-
mi-ti ‘trees’, mi-itu ‘forests’
i- / y-
∅ / ji-
tunda ‘fruit’, ji-cho ‘eye’
li- / l-
ma- / m-
ma-tunda ‘fruits’, m-eupe ‘white things’
ya- / y-
ki- / ch-
ki-tu ‘thing’, ch-akula ‘food’
ki- / ch-
vi- / vy-
vi-tu ‘things’, vy-akula ‘foods’
vi- / vy-
N- / ∅ / ny-
m-bwa ‘dog’, thembo ‘elephant’, ny-ama ‘meat’
i- / y-
N- / ∅ / ny-
m-bwa ‘dogs’, thembo ‘elephants’, ny-ama ‘meats’
zi- / z-
u- / w-
u-limi ‘tongue’, w-imbo ‘song’
u- / w-
N- / ∅
n-dimi ‘tongues’, ny-imbo ‘songs’
zi- / z-
u- / w-
u-zuri ‘beauty’, w-eupe ‘whiteness’
u- / w-
ku- / kw-
ku-pika ‘cooking, to cook’, kw-enda ‘going, to go’
ku- / kw-
pa- / p-
p-angu ‘at, on my place/s’
pa- / p-
ku- / kw-
kw-angu ‘at, to my place/s’
ku- / kw-
m(u)- / mw-
mw-angu ‘in my place’
m- / mw
Three types of gender can be distinguished: inherent, derived, and propositional. Inherent gender is defined by lexically specified features, and is marked in each dictionary entry for a noun. This entails the assumption that nouns have a basic gender specification, e.g. m-tu  ‘person’, m-ti ʒ ‘tree’, ki-ko  ‘pipe’, etc. A syntactic rule, sensitive to higher-level structure, adds a plural feature to lexical entries; assignment of concord then proceeds appropriately.
Derived gender is acquired by inherently genderless entities, such as adjectives and verbs, through various derivational processes of nominalization, e.g.:
(2) -za-(l) (genderless) ‘give birth to’ >
m-zaz-i  ‘parent’
ki-zal-ia  ‘offspring’
u-zaz-i  ‘parenthood’
ku-za-a  ‘giving birth’
Nouns which have lexically specified features also participate in derivational processes, and can have their lexical gender submerged by another gender feature:
(3) m-tu  ‘person’ > ji-tu  ‘giant, large person’
ki-ji-tu  ‘dwarf, small person’
u-tu  ‘humanity’
Both inherent and derived gender are sensitive to certain semantic considerations, and can be submerged in two cases:
(a) If a noun is [+animate], regardless of its inherent gender, it governs animate concord, i.e. of Class 1/2 (e.g. m-toto yu-le  ‘that child’, wa-toto wa-le  ‘those children’). Thus the noun phrases m-toto  ‘child’, m-ungu ʒ ‘God’, ki-j-ana  ‘youth’, n-dugu  ‘relative, brother’, and ji-tu  ‘large person’ all take the pronoun yu-le  ‘that’ and the predicate a-mefurahi  ‘is happy’, meaning ‘That child (etc.) is happy.’ Animate concord is generally not characteristic of other Bantu languages; where it exists in East Africa, it can be attributed to areal spread from Swahili.
(b) Animate concord does not operate when animate augmentatives and diminutives are considered derogatory; in this case, animate concord is submerged and concord is determined accordingly:
(4) ji-tu li-le [5, derogatory] ‘that giant/ogre’
Cf. ji-tu yu-le [5, non-derogatory] ‘that large person’
(5) ki-twana ki-le [7, derogatory] ‘that brat’
Cf. ki-twana yu-le [7, non-derogatory] ‘that young slave’
(6) nyumba-ni pa-ngu ‘at my house’
nyumba-ni kw-angu ‘to my house’
nyumba-ni mw-angu ‘in my house’
8. Noun-class semantics
The rules which operate to spread and specify the shapes of concord on constituents are generally formal, and are sensitive to the presence (or absence) of specific prefixal shapes. Semantics plays only a limited role, e.g. as noted above. However, there is internal and comparative historical evidence that gender was once semantically defined. The outlines of such a system are still discernible in Swahili, as shown in Table 3.
Table 3. Meanings of Swahili Noun Classes
[1/2]: names of human beings; a few animals
[3/4]: plants, things made from plants or wood (mat, handle), natural phenomena (fire, smoke), the spirit world (God, spirit), some animate-like body parts (foot, arm, heart)
[5/6]: fruits, paired body parts
: liquid mass nouns (water, oil, saliva); collections (groups: friends, womenfolk, hours, hyena pack)
[7/8]: artifacts (tool, chair, pipe), defective humans (deaf person, lame person)
[9/10]: kinship terms, most animals, insects
[11/10a]: linear objects (tongue, wall, fence, sword)
: abstracts (freedom, beauty, humanity)
Animacy completely dominates any other semantic feature that might still play a role; it is the only semantic feature to which concord is sensitive, except for the “derogatory” characteristic mentioned above. However, parts of the system play a role in assigning loanwords to classes: thus words for trees and plants are placed in Class 3/4, e.g. mw-embe, mi-embe ‘mango tree’ (< Hindi), m-nazi, mi-nazi ‘coconut palm’ (< Persian or Arabic), m-ndimu, mi-ndimu ‘lime tree’ (< Persian and Hindi). Most borrowed words fall into either Class 9/10 or 5/6—because, like native nouns in these classes, they have no overt prefix. Loans whose initial syllables can be reinterpreted as identifiable prefixes are assigned accordingly: ki-tabu, vi-tabu [7/8] ‘book’ (< Arabic kitāb), or m-sumari, mi-sumari [3/4] ‘nail’ (< Arabic musmār) (cf. Zawawi 1979, Whiteley 1969:138).
9. Verb morphology
Swahili is an agglutinating language; this is seen in the pivotal role played by the verb root in supporting affixes. The verb carries subject/object concord, anaphoric pronouns, and tense/aspect affixes; it also can be marked for negative and relative, as shown in (8–9) respectively:
(7) (Hasani) a- ta- m- pend-a (Adija).
H. 3.sg.subj fut 3.sg.obj love-indic A.
‘He (Hasani) will love her (Adija).’
(8) Ha- tu- ta- wa- pend-a.
neg 1.pl.subj fut 3.pl.obj love-indic
‘We will not love them.’
(9) wa- na- o- tu- pend-a
3.pl.subj pres rel.3.pl. 1.pl.obj love-indic
‘they who love us’
Table 4. Swahili Verb Inflection
‘not fasten, tie’
‘be not fastened’
‘fasten for, to, with’
‘cause to fasten’
‘cause to be fastened’
‘cause to fasten to’
‘cause to be fastened to’
applied passive causative
‘be fastened together’
‘carry out work together’
causative of reciprocal
‘be opened by’
passive applied reversive
Though the rules are complex, a hierarchy is apparent: modal suffixes are always final, and of the non-modals, passive is always final.
10. Derived nominals
Nominalizing rules derive agentive, instrumental, stative, and other nominals by adding suffixes to verb stems. Some examples are given in Table 5.
Table 5. Swahili Derived Nominals
-i [ +agentive]:
-shinda ‘win, overcome’
-o [ +instrumental]:
ki-zib-o ‘stopper, cork’
-o [ +resultant]:
-u [ +stative]:
-tukuka ‘be exalted’
m-tukuf-u ‘exalted person’
-ku-u ‘important, chief’
-e [ +patient]:
-peta ‘bend round’
Other rules move nouns from one class to another, changing their inherent gender, e.g. m-tu  → u-tu  ‘humanity’.
Compounding is not common in Bantu languages, but some old patterns exist in Swahili. Thus, from mwana ‘offspring’ come mwana + funzi ‘student’ (-funzi < -fund- ‘learn’) and mwana + chama ‘member’ (chama ‘organization’). Noun + noun juxtaposition, formerly uncommon, is increasingly used to create new terms: thus askari kanzu ‘plainclothes policeman’ < askari ‘police’ + kanzu ‘caftan’.
11. Aspects of syntax
Word order is typically Subject Object Verb, but different orders are possible under topicalization and relativization (of objects, but not of subjects):
(10) Mwanafunzi a-li-(ki)-soma kitabu.
student he-past-(it)-read book
‘The student read the book.’
(11) Kitabu a-li-ki-soma mwanafunzi.
book he-past-(it)-read student
‘The book, the student read.’
(12) mwanafunzi a-li-ye-(ki)-soma kitabu
student he-past-who-(it)-read book
‘the student who read the book’
(13) kitabu a-li-cho-ki-soma mwanafunzi
book he-past-which-it-read student
‘the book which the student read’
Noun modifiers follow the noun, but demonstratives can be preposed to mark definite nouns: mtoto mzuri ‘good child’, mtoto mmoja ‘one child’, mtoto wangu ‘my child’, mtoto yule ‘that child’, but yule mtoto ‘the child’. Possessives are formed with an associative morpheme -a, subject to concord, which links the possessed with the possessor: mtoto w-a Juma ‘Juma's child’.
Prepositional phrases are formed with a number of devices:
(a) Noun + associative -a + noun, where the first noun is relational, e.g. juu ‘top’, nyuma ‘behind’, mbele ‘front’, etc.: mbele ya nyumba ‘in front of the house’
(b) Noun + noun, where the first noun can be verbal or relational, e.g. (ku)-toka nyumba-ni ‘from the house’ (ku-toka ‘to come from’), mpaka nyumba-ni ‘up to the house’ (mpaka ‘boundary’)
(c) Phrases with na ‘and/with/by, etc.’ and the associative kw-a (ku-  locative), e.g. -enda na Juma ‘go with Juma’, -kata kw-a kisu ‘cut with a knife’.
Many English prepositional phrases are translated into Swahili by the applied verb form with a prepositionless object, e.g.,
(14) Hamisi ali-m-fungu-lia (Juma) mlango.
H. he-past-him-open-for J. door
‘Hamisi opened the door for him (Juma).’
Adverbials are expressed by a number of devices:
(i) Locative kw-a constructions, e.g. kw-a uangalifu ‘carefully (lit. with care)’
(ii) Locative nominals, e.g. kw-ingine  ‘in/by another direction’, p-engine  ‘sometimes’
(iii) Nominals with ki- or vi-, e.g. ki-dogo ‘a little amount’, vi-zuri ‘well’, vi-le ‘also’ (-le ‘that’ demonstrative)
(iv) Loans from Arabic, e.g. sana ‘very’, karibu ‘near, nearly’, kabisa ‘absolutely’
(v) Ideophones, e.g. mw-eusi t̲i̲t̲i̲t̲i̲ ‘very black person’, A-me-lala f̲o̲f̲o̲f̲o̲ ‘He is fast asleep.’
(a) With conjunctions: kw-amba ‘that’ [+factive] (< -amb- ‘say’), ati ‘that’ [−factive] (<*-ti ‘say’), kuwa ‘that’ (< -wa ‘be, become’), e.g. A-li-sema ati a-ta-kuja ‘He said (=claimed) that he would come.’
(b) By the use of -amb- (‘say’) in relative clauses: mtu amba-ye a-me-kuja ‘person who has come’
(c) By pronominal relativization: mtu a-na-ye-kuja ‘person who came’, mtu a-ja-ye ‘person who comes’
(d) By concatenation of verbs marked with tense/aspect and/or subjunctive mood markers:
(15) Nenda u-ka-ni-nunu-li-e.
‘Go and (i.e. so that you might) buy me something.’
(16) N-li-mw-ona a-ki-ja.
‘I saw him coming.’
(17) Ni-me-vunjika mguu.
‘I have a broken leg.’
See also Bantu Languages.
Abdulaziz, Mohamed H. 1979. Muyaka: Nineteenth century Swahili popular poetry. Nairobi: Kenya Literature Bureau.Find this resource:
Harries, Lyndon. 1962. Swahili poetry. Oxford: Clarendon Press.Find this resource:
Hinnebusch, Thomas J. 1979. Swahili. In Languages and their status, edited by Timothy Shopen, pp. 209–294. Cambridge, Mass.: Winthrop.Find this resource:
Knappert, Jan. 1979. Four centuries of Swahili verse: A literary history and anthology. London: Heinemann.Find this resource:
Nurse, Derek, and Thomas J. Hinnebusch. 1993. Swahili and Sabaki: A linguistic history. Berkeley and Los Angeles: University of California Press.Find this resource:
Nurse, Derek, and Thomas Spear. 1985. The Swahili: Reconstructing the history and language of an African society, 800–1500. Philadelphia: University of Pennsylvania Press.Find this resource:
Polomé, Edgar C. 1967. Swahili language handbook. Washington, D.C.: Center for Applied Linguistics.Find this resource:
Prins, Adriaan H. J. 1961. The Swahili-speaking peoples of Zanzibar and the East African coast (Arabs, Shirazi and Swahili). London: International African Institute.Find this resource:
Wald, Benji. 1987. Swahili and the Bantu languages. In The world's major languages, edited by Bernard Comrie, pp. 991–1014. London: Croom Helm. New York: Oxford University Press.Find this resource:
Whiteley, Wilfred. 1969. Swahili: The rise of a national language. London: Methuen.Find this resource:
Zawawi, Sharifa M. 1979. Loan words and their effect on the classification of Swahili nominals. Leiden: Brill.Find this resource: