Printer Friendly

The adaptation of Swahili loanwords from Arabic: a constraint-based analysis.


This paper touches on two languages; Kiswahili and Arabic. Kiswahili is a Bantu language spoken by more than 80 million people mainly in East and Central Africa (Bosha 1993:45); while Arabic belongs to the Semitic group and is mainly spoken in the Middle East. There has been an interaction between Arabs and East Africans for hundreds of years leading to linguistic interferences on both sides. Though so, the most affected language between the two seems to be Kiswahili and the most affected domain of the language is phonology. Kiswahili has borrowed heavily from Arabic. Zawawi (1979:73) notes that "a collection and collation of loanwords in Johnson's Standard English-Swahili Dictionary yielded a total of 3,006 words of foreign sources out of which 2,354 (80%) were of Arabic origin".

This paper seeks to analyze how Kiswahili loanwords from Arabic have been nativized by the recipient language. I have replicated the data used here from Bosha (1993). A few words have been chosen (see Appendix) for the purpose of this analysis. In choosing the words, care was taken to include words with various types of syllables, for example, words that do not change phonologically, words with consonant clusters in the initial, mid and final positions, words with consonant geminates, vowel hiatus and long vowels.

The syllable, being a major component of phonological organization, will be the focus of this paper. The arrangement and rearrangement of the phonemes in the syllable in the recipient language will be explained by using constraints interaction. The purpose is to show how the recipient language repairs borrowed syllables coming into the word using constraints interaction.

Theoretical Background

The analysis in this paper is based on Optimality Theory whose central idea is that surface forms of a language reflect resolutions of conflicts between competing constraints. A surface form is "optimal" if it incurs the least serious violations of a set of constraints, taking into account their hierarchical ranking (Kager 1999). The following comprise the core principles of Optimality Theory:

(a) Violability: Constraints are violable, but violation is minimal.

(b) Ranking: Constraints are ranked on a language-particular basis; the notion of minimal violation is defined in terms of this ranking.

(c) Inclusiveness: The constraint hierarchy evaluates a set of candidate analyses that are admitted by very general considerations of structural well-formedness.

In the analysis that follows, the following constraints are used:

*COMPLEX = no complex syllable margins

*COMPLEXVOW = no strings of vowels

DEP-C = output consonants must have input correspondents (no C epenthesis)

DEP-IO = output segments must have input correspondents (no epenthesis)

IDENT-IO (F) = the specification for the feature of an input segment must be preserved in its output correspondent

IDENT-IO (place) = the specification for place of articulation of an input segment must be preserved in its output correspondent

MAX-IO = input segments must have output correspondents (no deletion)

MAX-V = input vowels must have output correspondents (no deletion of vowels)

NOCODA = syllables are open

ONSET = syllables must have onsets


SON-SEQ = complex onsets rise in sonority, and complex codas fall in sonority

Before delving into the analysis, it is necessary to explain the kind of syllable structures found in Kiswahili language.

Kiswahili Syllable Structure

Kiswahili, just like other languages, has its words divided into syllables according to the principle of increasing sonority. The CV syllable is the most common in Kiswahili.
(1)      CV CV
        / k i. t i /    'chair'
        / f i. k a /    'arrive'
        /ka.ta/         'cut'

The first word can be represented by the following syllable tree.


This syllable structure presupposes two faithfulness constraints: ONSET and NOCODA. Put on a tableau, this will look as follows:
Input: / kiti /                 NOCODA        ONSET
a. [thumbs left] ki.ti
b.               k.i.ti                        *!
c.               ki.t.i                        *!
d.               kit.i             *           *!

Also, Kiswahili has single vowel syllables. Examples of such syllables are:
(3)     oa           /o.a/   'marry'
        pia          /pi.a/  'also'
        fua         /fu.a/   'wash' (like in washing clothes)

This shows that you can get one vowel syllable sequences in Kiswahili as illustrated in (4) below:


In terms of OT, we can use the following constraints to show this kind of syllabification: DEP-C,

Input: / oa /                  PEAK         NOCODA      DEP-C

a. [thumbs left] o.a
b.                                          *!
c.               ol.a                          *!         *

Also, a preconsonantal nasal functions as a syllabic peak. The nasals that function as peaks in Kiswahili are /m/ and /n/. Examples are shown below:
(5)      mtu       /m.tu/      'person'
         nne       /       'four'
         nchi      /        'country'

In syllabifying the above nasals as PEAKS we violate ONSET but avoid NOCODA violation. Using the first word /m.tu/, it is possible to show that sometimes syllable and morpheme boundaries coincide; for example, /m/ is a noun class prefix in singular form for people and animals while /tu/ is the root of the word that means 'person'. The syllable therefore functions in the demarcation of morpheme edges and in defining the position and shape of affixes.

It is important to state that a number of scholars (Polome 1967:50; Myachina 1991:12; Mohammed 2001:11) have observed that Kiswahili has open syllables. This seems to suggest that there is NOCODA in Kiswahili syllables because the morphological structure of the language does not permit it. When loanwords are adapted into this language, it is difficult to maintain this position. The incorporation of loanwords into Kiswahili has resulted in an expansion of the Kiswahili syllabry, that is, it has increased the use of other syllable structures which are not very common in this language such as CCV and CCCV.

In that vein, we can say that another syllable structure that is found in Kiswahili is a cluster of two consonants with a vowel. This type is usually restricted to syllables with either a nasal as the first consonant or the approximants /j/ or /w/ as the second consonant. Here are examples of words:
(6)       ngamia     /nga.mi.a/      'camel'
          mwezi      /mwe.zi/         'moon'
          twanga     /twa.nga/        'pound'

Although Kiswahili does not frequently have consonant clusters, when they occur they are to be found at the beginning of words or syllables. This is particularly common with the borrowed words. This shows that the language follows the principle of maximum onset and minimum coda.

Lastly, the language also has the syllable structure CCCV, mainly in borrowed words, where the first consonant is a nasal or the last consonant is an approximant /w/, for example:
(7)    chungwa               /cu.ngwa/         'orange'
       ngwena                /         'crocodile'

The last two forms of syllables show that *COMPLEX is violated in Kiswahili because of the initial consonant cluster. Having seen the types of syllables found in Kiswahili, I will now proceed to show how loanwords are adapted into the language.

Syllable Repair Process

The idea of loanword adaptation or nativization at the phonological level is governed by syllable well-formedness in the recipient language. When a word is borrowed from one language to another, in most cases it violates some constraints of syllable well-formedness. The recipient language moves fast to fix the problem. For example, many languages try to avoid complex onsets and codas. The typical avoidance strategies that Kiswahili uses to repair the nonconforming syllables of the borrowed words are:

(a) vowel epenthesis

(b) consonant deletion

(c) cluster tolerance

(d) feature change

These procedures are pursued in more detail below with relevant examples.

Vowel Epenthesis

Batibo (1996:38) notes that this is by far the most common method of consonant cluster nativization in Kiswahili. It involves the insertion of a vowel between two consonants or after a consonant in a syllable final position. Epenthesis involves a violation of faithfulness because the epenthetic segment has no counterpart in the input. For example:
     Input: asl
    Output: asili

/asl/--> asili
Input: / asl /            *COMPLEX   NOCODA    DEP-IO   ONSET

a.               asl         *!        *                  *
b.               a.sil                 *!       *         *
c. [thumbs left]                        * *       *
d.                    *!       * *       * *

Although Polome (1967) and Batibo (1996) acknowledge the existence of consonant clusters at the onset of borrowed words and give examples like stempu /ste.mpu/ "stamp' and stovu / "stove', I did not get any such examples of word initial onset consonant cluster in the data. Also, procedurally we know that an epenthetic vowel breaks up clusters of two consonants at the beginning of a word, as well as clusters of three consonants in medial position. A medial cluster of two consonants is not broken up by epenthesis, as this can be split between two syllables without the need for a complex margin: where the first consonant syllabifies as a simple coda, and the second as a simple onset. This has proved problematic in Kiswahili when it adapts a loanword with a medial consonant cluster (CC) because it prefers open syllables to closed syllables. In fact, Polome (1967:50) says that in words of Bantu stock, consonant sequences are always tautosyllabic, as in / 'crocodile' and /ku.bwa/ 'big'. He, however, adds that in loanwords the syllable boundary usually lies after the first two consonants, for example., between /l/ and /t/ in sultani , but in colloquial speech, this syllable-final consonant is often released with a short vowel, thus tending to re-establish the Bantu pattern of syllabification, for example., labda 'perhaps' /la.bu.da/ or ratli 'pound', also becomes / I will use the word sultani to illustrate my point.
(8) Input: sulta:n
    Output: sultani

Let us try to syllabify the word sultani.

(i) Write the word "sultani" and link the letters to C or V forms as appropriate.
(9) C V C C V C V
      :   :   :  :   :  :   :
      s  u   l  t  a  n  i

(10) Link each V element to a syllable symbol


(11) Link C-elements to the V on their right.


(iv) Link C-elements to the V preceeding them so long as the resulting sequence is allowed in the language.



In this example of the word sultani, we favour (a) against (b). Although (b) maintains the open syllable principle which is basic in Kiswahili, it violates the sonority principle in the second syllable; that is, the SON-SEQ constraint which requires complex onsets to rise in sonority and complex codas to fall in sonority. In (b), /l/ and /t/ are in onset position yet /l/ is more sonorous but is at the outer edge than /t/. This is illustrated in (12):


The syllabified word, /, can be arrived at using the following constraints.
Input: /sulta:n/             *COMPLEX SON-SEQ  NOCODA  DEP-IO

a. [thumbs left]                       *        *
b.          *!       *
c.                           * *!     *

Although some borrowed words with word-medial consonant clusters are heterosyllabic, e.g.,/, other words do not take this option but prefer to break up the cluster by epenthesis, for example.,
(13)  ibli:s 'devil' - / /
      ibri:q 'kettle' - / bi.ri.ka /

Suffice to say that the unmarked situation in Kiswahili is for syllables to lack coda. The syllabification of a word like kubwa 'big' will differ from that of sultani because the sonority between the two consonants which appear mid-word is arranged differently. It is possible to maximize the onset and have open syllables like /ku.bwa/; but vowel epenthesis as in (c) below will not work. This is how its constraints can be ordered on the tableau.
Input:/ kubwa /           SON-SEQ   NOCODA  DEP-IO   *COMPLEX
a. [thumbs left] ku.bwa                                 *
b.               kub.wa               *!
c.               ku.bu.wa                     *!

Kiswahili treats loanwords from Arabic with medial consonant cluster, final consonant cluster and just a final consonant in a similar way. This is because the language insists on maximum onset and minimum coda. In these cases, a vowel is inserted in between the consonant cluster or word finally. Look at these examples:
(14) adhuhr --> adhuhuri / ri/ 'midday'
     ahd    --> ahadi    /a.ha.di/     'promise'
     urs    --> arusi    /     'wedding'
     milk   --> miliki   /    'property',
     waqt   --> wakati   /wa.ka.ti/    'time'

It is possible to predict what kind of vowel will be added in word final position during epenthesis. Words borrowed from Arabic which end in a consonant acquire additional vowels whose type is determined by the nature of the final consonant; after labials, /u/ or /o/ is added, and after coronals and dorsals, /i/ or /e/ is added.
 (15) /o/or/u/
        amm         -->  amu          'uncle'
        asquf       -->  askofu       'bishop'
        ibd Adamu   --> binadamu      'human being'
       aawa:m       --> awamu         'inception'
       ibn amm      --> binamu        *'cousin'
       iara:b       --> irabu         'vowel'
      tab           --> taabu         'trouble'

      /i/ or /e/

      waqt          -->  wakati       'time'
      ahd           -->   ahadi       'promise'
      ars           -->   arusi       'wedding'
      milk          -->    miliki     'property'

Consonant Deletion

Segment deletion is another way to avoid violation of the NOCODA condition which Kiswahili tries to uphold. Segment deletion is a violation of faithfulness because an input segment has no counterpart in the output. The constraint that enforces the preservation of input segments in the output is MAX-IO. Careful analysis of the data shows that consonant deletion will occur when there is a geminate consonant. In that case, one of the consonants of the geminate is dropped, for example,
(16) ammar    --> amiri   /a.mi.ri/    'begin'
     amm      --> amu     /       'uncle'
     assubh   --> asubuhi / 'morning'
     ibn amm  --> binamu  /   *'cousin'
     budd     --> budi    /bu.di/      'alternative'
     iddaaa   --> dai     /da.i/       'claim', 'demand'
     ghass    --> ghasia  /    'confusion', 'bustle'
     hadd     --> hadi    /ha.di/      'until'
     saffa:   --> safi    /      'become clear' 'clean'

Consonant deletion to resolve violations of the coda condition will involve:
       ha.di   > hadd

Input: / hadd /           *COMPLEX   NOCODA   MAX-IO   DEP-IO
a.               hadd        *!        *
b.               had.di                *!                *
c. [thumbs left] ha.di                          *        *
d.               ha.ddi      *!                          *

As can be seen in the example in the tableau, consonant deletion alone is not enough to make these syllables well-formed. It must work hand in hand with vowel epenthesis.

Cluster Tolerance

There are a few cases where Kiswahili maintained clusters that were in the borrowed Arabic word. In Optimality Theory terms, this indicates faithfulness. Below are examples of consonant cluster tolerance.
(17) unwan  -->  anwani    /   'address'
     usquf  -->  askofu    /as.ko.fu/   'bishop'
     izz    -->  enzi      /en.zi/      'rule', 'power', 'domain'
    sultan  -->  sultani   /  'king', 'ruler', 'chief'
 amr    -->   amri         /am.ri/      'command'

The faithfulness constraint militating against epenthesis is DEP-IO, and against syncope is MAX-IO. This can be represented on a tableau with a word like amri.
/amr/--> amri

Input: / amr /            *COMPLEX  SON-SEQ   DEP-IO   NOCODA
a. [thumbs left] am.ri                         *        *
b.               amr        *!                          *
c.               a.mri      *!        *        *
d.                            * *!
e.               amr.i      *!                 *        *

So far it seems there is a free variation in the syllabification of CC clusters in words. It can either be heterosyllabic like / or tautosyllabic like /ku.bwa/.

The examples listed above indicate that Kiswahili does accept consonant clusters. On the consonant cluster tolerance, Batibo (1995:39) notes that the language has become unique among Bantu languages due to its insensitivity to consonant clusters. This, he explains, is due to three reasons:

(a) Its long association with foreign languages, particularly Arabic.

(b) Its susceptibility to borrowing foreign words.

(c) Identification with Arabicism and Anglicism has prevented nativization because of popular use of the borrowed items.

There are also a few examples of vocalic complexes which have been tolerated in Kiswahili. These are listed below:
(18) aib    -->  aibu   /a.i.bu/   'shame'
     baia   -->  bei    /be.i/     'price'
     iddaaa -->   dai   /da.i/     'claim', 'demand'
     kaid   -->  kaidi  /ka.i.di/  'obstinate', 'disobedient'
     naam   -->  naam   /naam/     'yes', 'certainly'
     taab   -->  taabu  /ta.a.bu/  'trouble'
     za:id  -->  zaidi  /za.i.di/  'more', 'besides'

It seems that where a borrowed word has three or more vowels in a row, some are deleted so that at most two remain. This is what we see in /iddaaa/ --> dai 'claim', /aib/ --> aibu 'shame' and /baia/ --> bei 'price'. What then emerges is that Kiswahili can comfortably accommodate two vowels occurring in a row; it is not in the tendency of employing consonant epenthesis to break vowel clusters. In that case, when a loanword has a hiatus, either vowel syncope can be employed to reduce the cluster or the word is adapted the way it is with no changes. The constraint that is violated in these examples is ONSET.
/ baia /--> bei
Input: / baia /          *[COMPLEX.sup.VOW]  ONSET  IDENT-IO(F)  MAX-V
a.               ba.i.a                        * *!
b. [thumbs left] be.i                          *         *          *
c.               bai.a           *!            *

/aib/ --> aibu
Input: / aib /            *[COMPLEX.sup.VOW]   NOCODA   DEP-IO   ONSET
a. [thumbs left] a.i.bu                                     *      * *
b.               ai.bu           *!                         *      *
c.               a.ib.u                            *!       *      * * *
d.               a.ib                              *!              * *

Feature Change

Other changes noted in this data have got to do with the phonemes themselves. There are some phonemes that are found in Arabic but are not in the Kiswahili phonemic inventory. If a loanword has such a phoneme, then it has to be changed to the nearest equivalent in the recipient language. Below are examples where Arabic /q/ and /kh/ are realized as /k/ and /h/ respectively in Kiswahili.
         /q/ -->  [k]
(19) waqf        -->  wakfu   /wak.fu/   'religious endowment'
    wa:faq       -->   afiki  /  'to agree with'
    aqd          -->   akidi  /  'celebrting a wedding'
    aql          -->    akili /  'intelligence'
    usquf        -->   askofu /a.sko.fu/ 'bishop'
    ibri:q       -->  birika  /bi.ri.ka/ 'kettle', 'cistern
    ta:ri:kh     -->  tarehe  / 'date', 'chronology', 'annals'
    waqt         -->  wakati  /wa.ka.ti/ 'time'
    ya:qu:t      -->  yakuti  /ya.ku.ti/ 'ruby', 'sapphire'

      /kh/  -->  [h]
   akhar        --> ahirisha  /a.hi.ri.sha/  'postpone'
   bakht        --> bahati    /ba.ha.ti/     'luck', 'fortune'

We can use the constraint IDENT-IO (place) together with the others introduced earlier to show the changes that have taken place here.
Input: /waqt/             *COMPLEX   NOCODA   IDENT-IO(place)   DEP-IO

a.[thumbs left] wa.ka.ti                           *              **
b.              waqt         *!        *
c.              wakt         *!        *           *
d.              wak.ti                 *!          *               *


This paper has done an analysis of Kiswahili loanwords from Arabic using the collection of words in Bosha (1993). A point that has clearly manifested itself is that Kiswahili prefers vowel epenthesis to vowel syncope or apecope in the resyllabification of loanwords. The analysis has also shown that Kiswahili has two main epenthetic vowels, /i/ and /u/, which to some extent are contextually predictable. Though Kiswahili is an open syllable language, I have shown that loanwords have forced it to take up closed syllables so as not to violate the sonority hierarchy within the syllable. It is difficult to explain this in phonology without going into sociolinguistics. It seems that Kiswahili as a borrowing language was at the substratum in comparison to the lending languages which were at the superstratum. This made it prestigious to retain the phonotactics of the lending language. This means that, although the language has a mechanism of adapting new words, sometimes not all syllables or words are repaired; there is a high level of tolerance.

Generally, this paper has also shown that Optimality Theory can be effectively used to analyze nativization of loanwords without having to resort to rule based phonology. This has been done by looking at the interplay between faithfulness and markedness in the syllable repair process.


Batibo, Herman M. 1996. "Loanword clusters nativization rules in Tswana and Swahili: a Comparative study" in South African Journal of African Languages Vol. 16.2

Bosha, Ibrahim A. 1993. The Influence of Arabic Language on Kiswahili with a Trilingual Dictionary (Swahili-Arabic-English). Dar es Salaam. Dar es Salaam University Press.

Kager, Rene 1999. Optimality Theory. Cambridge. Cambridge University Press.

Mohamed, M. A. 2000. Modern Swahili Grammar. Nairobi. E.A.E.P.

Myachina, E.N. 1981. The Swahili Language: A Descriptive Grammar. London. Routledge & Kegan Paul.

Polome, E.C. 1967. Swahili Language Handbook. Washington. Center for Applied Linguistics.

Zawawi, Sharifa M. 1979. Loanwords and Their Effect on the Classification of Swahili Nominals. Leiden. E.J. Brill.

Kiswahili words with Arabic and English translations replicated
from Bosha (1993)

Arabic Word  Swahili Word   Formal         English Word

adhuhr       adhuhuri       adulhuri       midday
wa:faq       afiki          alfiki         to agree with
ahd          ahadi          alhadi         promise
ahl          ahali          alhali         relatives, family
akhar        ahirisha       ahilrisa       postpone
aib          aibu           alibu          shame
aqd          akidi          alkidi         celebrating a wedding
aql          akili          alkili         intelligence
alasr        alasiri        alalsiri       afternoon
ammar        amiri          almiri         begin
amm          amu            lamu           uncle
unwan        anwani         alnwani        address
urs          arusi          alrusi         wedding
asl          asili          alsili         root, source
usquf        askofu         aslkofu        bishop
assubh       asubuhi        asulbuhi       morning
bakht        bahati         balhati        luck, fortune
baia         bei            lbei           price
ibni Adamu   binadamu       binaldamu      human being
ibn amm      binamu         bilnamu        * cousin
ibri:q       birika         bilrika        kettle, cistern
budd         budi           lbudi          alternative
iddaaa       dai            ldai           claim, demand
dain         deni           ldeni          claim, debt
izz          enzi           lenzi          rule, power, dominion
fihris       faharasa       fahalrasa      index
ghass        ghasia         yalsia         confusion, bustle
hadd         hadi           lhaldi         until
ibli:s       ibilisi        ibillisi       devil, satan
iara:b       irabu          ilrabu         vowel
jaash        jasho          ljaso          sweat
ja:su:s      jasusi         jalsusi        spy
kaid         kaidi          kalidi         obstinate, disobedient
maadin       madini         maldini        mine, mineral
du:d         mdudu          mldudu         insect
milk         miliki         milliki        property, possession
naam         naam           lnaam          yes, certainly
raad         radi           lradi          thunder
saffa:       safi           lsafi          become clear, clarify
sulta:n      sultani        sulltani       king, ruler, chief
taab         taabu          talabu         trouble
ta:ri:kh     tarehe         talrehe        date, chronology
waqt         wakati         walkati        time
ya:qu:t      yakuti         yalkuti        ruby, sapphire
za:id        zaidi          zalidi         more, besides
waqf         wakfu          lwakfu         religious endowment


Leonard Chacha Mwita, Ph.D.

chachamwita8@yahoo. com

Kiswahili Department, Kenyatta University
COPYRIGHT 2009 Journal of Pan African Studies
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Mwita, Leonard Chacha
Publication:Journal of Pan African Studies
Article Type:Report
Geographic Code:60AFR
Date:Mar 1, 2009
Previous Article:Actualizing free primary education in Kenya for sustainable development.
Next Article:The case for Kiswahili as a regional broadcasting language in East Africa.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters