Dictionaries and the Standardization of Spelling in Swahili ·

The paper discusses the role of the dictionary in standardizing the spelling of Swahili. The discussion begins by defining key terms in this paper: spelling and standardization of spelling. It surveys lexicons written in Swahili between 1811-1990 and records the efforts made to establish spelling conventions for Swahili words in Roman characters, pointing out variant spellings of words written by different authors. The paper focuses on the role played by different lexicons in setting orthography for Swahili 'words, viz. Steere (1870), Krapf (1882), Nettelbladt (1891), Madan (1903), Sacleux (1939), etc. It observes how the lexicons established nonns for words of a language which was hitherto not written in Roman characters. It also shows how lexicons helped to standardize the spelling of words to its current fonn especially after a standard dialect had been selected, pointing out lexicons which exclusively recorded words of the standard language with minimal variants. Finally the paper emphasizes the significance of the dictionary to adhere to the standard orthography.


Introduction
The observation by Krumm (1940: 3) that "the question of orthography of Swahili written in Roman characters is not yet definitely solved", is as valid today as it was more than half a century ago when it was made. Dictionaries compiled between 1882 and 1991 prove this fact despite the efforts made in 1925 by a Committee for the Standardization of Swahili (better known as the Chiponda Committee (see Mbaabu 1995)) and the Interterritorial Language Committee founded in 1930. The objective of this paper is to discuss the role of dictionaries in standardizing orthography of written Swahili in Roman characters. In this paper we shall (1) examine efforts made to establish spelling conventions in Swahili from the time when Swahili was first written in Roman script, (2) discuss the role of dictionaries in standardizing spelling of Swahili words from 1870 to 1990, and (3) discuss why a dictionary should promote a standard orthography. But first of all let us briefly explicate what spelling and standardization of spelling mean.
Spelling, according to Gove (1961), is the art and technique of forming words by letters according to accepted usage or standard spelling as formulated by language planners. Standardization of spelling is concerned with unifying underlying linguistic diversity in order to make the standard language fit to be a national language. Standard orthography is the chosen spelling of a word out of the competing variants. This is the spelling that will be put into use in schools, government offices, the press and other areas. Standardization-of a language has basically two steps: (1) creation of a model for imitation and (2) promotion of this model over rival models (Ray 1968). Having put into perspective what standardization of spelling means, we shall now examine how Swahili was introduced into writing systems.

Writing Systems in Swahili
Swahili' was already a written language in Arabic script in the 13th century (Eastman 1983: 21). For almost five centuries this script dominated Swahili writings which were mainly songs and poetry, e.g. the work of Furno Liyongo, Aidarus bin Uthman, Binti Lembe, Mwengo bin Athuman, Muyaka Haji, etc.
By the beginning of the 19th century, Roman script was introduced by European travellers and missionaries who visited East Africa and began to learn Swahili. They prepared Swahili word-lists, conversation books, grammar books, teaching manuals and dictionaries. In this paper we shall confine our discussions to spelling conventions in Roman script because it has a considerable literature from which we can draw examples for illustration. Let us now look at how Swahili was first written in Roman characters.

Transcription of Swahili in Roman Characters
When a hitherto unwritten language is chosen as a standard language, a spelling system is developed for it and conventions are set up on how it is to be written (Eastman 1983). First attempts to establish orthography of Swahili words were made by European visitors, explorers and adventurers between 1811-1849. This involved (1) the transcription of Swahili phones in order to determine the Swahili phonetic alphabet and (2) establish spelling conventions for Swahili words as in Table I

Swahili Alphabet
The alphabet used for writing Swahili is the same as the international Roman alphabet, with letters chosen to represent the Swahili phones. It also uses digraphs i.e. a sequence of two letters to represent single sounds. According to Steere (1870) the Swahili alphabet has five vowels: a, e, i, 0 and u, and sixteen consonants: b, ch, d, f, g, h, j, k, I, m, n, p, r, s, t, V, w, y and z. The vowels are pronounced as in Italian, and the consonants as in English. He also included the following digraphs: gh is pronounced as the Dutch g, kh is pronounced as the German ch, th is pronounced as the English th in thing, kw is pronounced as the English qu in queer, ny has the sound of French ni or English ni as in companion, ng' resembles English ng when it occurs at the end of a word, e.g. bang, king, etc., ng resembles English ng as in engage, and sh resembles the English sh.

Influence of Foreign Languages on Swahili Orthography
A critical examination of the data in Table II shows (1) variant spellings for the same words and (2) variant representations of a phoneme. Variants of a word reflect the existence of more than one dialect in use in one speech community: Cf.
(1) afadali, afadhali, afasali, afathali, afazali and afzali, or (2) asubuhi, assubui, assubukhi, essubukhi, subuhi, ussubui, ussubukhi, etc. The variants cannot be said to belong to the various Swahili dialects because they are loanwords. In this case we are obliged to concur with Krumm's point of view (Krumm 1940: 50f£.) that the different Arabic variants we see in these word-lists originated from different Arabic dialects whose speakers came to East Africa and introduced them in Swahili: It seems natural that some Arabian words have penetrated into Swahili not in their classical but in their dialectic or foreign form ... Often we find in Swahili besides the usual form of the Arabic or Persian word a dialect form, for instance, dhaifu and daifu; dhoruba, doruba and zoruba; rejea and regea, jambia and gambiawhich prove that these words have found their way into Swahili from different Arabian sources.
However, we should also note that some of the variants are creations of the authors whose languages influenced the spelling of the words. For example, the Germanfl or 55 which is pronounced as lsi may have influenced German authors to use a double 5 instead of a single 5 which is pronounced as Iz/. Thus in order to articulate kisa correctly, they wrote the word with a double 5, i.e. qissa. One could also argue that afasali in Nettelbladt's list was his creation to suit his German Swahili learners so that they could pronounce [afazali]. We note the same for all Nettelbladt's words with 5, e.g. asuuri for azuuri. The influence of foreign languages as a source of variants in Swahili is also seen in j or dj pronounced as Id31 in French as in ngoja, ngodja. We observe the same for ch and sh each of which has three variant representations of the same phoneme: ch, t and tch (French) for It I I, e.g. cheka, teka and tcheka, and sh, ~ and sch (German) for I I I, e.g. ishirini, isirini and ischirini.
Another influence of foreign languages on Swahili as evident in early writings is the double consonance in Arabic which has phonological and se-, mantic signification in that language but has no bearing on Swahili. For example: katala he killed kattala he caused to kill alima he knew aUanta he caused to know salim he was well saUam he made or wished somebody to be well The double consonant in kattala above has caused the meaning to change and differ from that of katala. Phonologically a double consonant in Arabic signifies an emphatic consonant. These features are not found in Swahili. Unfortunately they have been introduced in Swahili. For example: bure and burre hata and hatta tena and tenna mwalimu and mwallimu tayari and tayyari feza and fezza

Vowel Harmonization of Loanwords
Swahili syllables are open and their structure is basically C + V. Some Arabic words adopted in Swahili consist of group consonants (which are not double consonants) and lack the C + V structure. For example: bakhti, barza, arbain, ashrini, hutba, etc.
These words were entered in some of the early lexicons and also appear in some Swahili literature from this time. However, in course of time the group consonants were separated by vowels whose selection was based on the rules of vowel assimilation by the same vowel that precedes the consonant (progressive assimilation) or by the same vowel that follows the consonant (retrogressive assimilation).
When lexicons were compiled during this time, Swahili had no standard dialect to which to refer for the standard spelling of words. Consequently, the compilers of these lexicons recorded every sound they heard, because it represented a lexical item. Sometimes they could not give a good transcription of the words pronounced for them either because they failed to capture the sounds correctly, or because their resource persons failed to articulate the words clearly. Cf. youmba (instead of nyumba), or che-mo-je, mouya, moya (instead of moja).

Creation of a Model to Follow
So far we have seen the beginning of writing Swahili in Roman characters and noted the reasons for variant spellings of the same words. Efforts were therefore made to standardize the language and hence harmonize the spelling of its words. In 1925 an education conference attended by members from Tanganyika and Zanzibar was held at Dar es Salaam and the Zanzibar dialect was selected as Standard Swahili for the two dependencies. Among other things it also proposed how to standardize the spelling of Swahili words. It gave a list of words which had variant spellings and recommended the standard orthography to be adopted (Mbaabu 1995: 47 Your committee came into being on 1st January 1930. Its terms of reference were to standardize the orthography and the grammar of the Swahili language, to write a new standard dictionary and to scrutinize the books in use in schools and to certify that they conformed with the standards laid down.
A new dictionary was necessary because the dictionaries which were already compiled had not aimed at standardizing the language. This is why each had a different orthography for the same words. The new dictionaries attempted to standardize the language.
(1) The double consonant feature which was dominant in many loanwords was abandoned. Cf. bara, basi, huba, asubuhi etc. in Johnson (1935 and on the one hand and Steere (1870), Krapf (1882) and Velten (1910) on the other. (2) One variant was selected out of the competing ones and was adopted as the standard orthogra-phy except if it was felt that more than one fonn had wide usage. Dialectal forms ~ere reduced substantially but not eliminated completely, as shown in Table III below:

Contribution of Swahili Lexicons towards the Standardization of Swahili Orthography
The first Swahili lexicons, though fragmented, were compiled between 1811 and 1848 (see Whiteley 1967: 50 Velten (1910), Madan (1903) and Brutel (1928). Coincidentally all except Krapf (1882) were in Kiunguja, the dialect which later became Standard Swahili. Krapf was aware of Kiunguja and entered some words from this dialect and marked them. Despite the discrepancies we noted above, e.g. double consonants and superfluous variants, the lexicons laid a good foundation for the standardization of Swahili. One of the criteria used to select a language or dialect to be a standard language is the availability of published literary works, grammar and lexicons in that language. Kiunguja had all these. The lexicons greatly influenced the standardization of spelling because the language planners could use them to select the forms with a higher degree of occurrence and simple spelling i.e. without double consonant or aspiration. Therefore, although it is agreed that the pre-Johnson dictionaries did not endeavour to standardize Swahili and that their contents reflected that fact, they laid the base for the standardization process to take place. Let us now examine the role of the dictionary in standardizing orthography.

The Role of the Dictionary in Standardizing Orthography
A dictionary whose objective is to describe the standard dialect has to record the standard orthography of a lexical item where there are competing forms (Mdee 1990). A lexicographer should therefore decide which form is considered standard and which is not. In this regard a standard dictionary has only one mission: to propagate the standard orthography.
A dictionary is regarded as the authority on spelling, grammar, meaning and usage of a language. It records the standard orthography of the norm, and if it includes items of another norm or other dialects of the same language, the social and geographical areas where each is spoken, are marked accordingly. A dictionary shall command authority over its users if it convinces them that it is adhering to the standard. Otherwise it will lose credibility as an authoritative reference of the standard language.
We have seen how the standard dictionaries compiled by the Interterritorial Language Committee Qohnson 1935 and 1939) attempted to standardize orthography of Standard Swahili. It was therefore expected that subsequent compilers of Swahili dictionaries would adhere to the standard. In what follows we shall examine dictionaries compiled between 1960, namely Rechenbach (1968, TUKI (1981) and Feeley (1990), i.e. thirty years after Swahili had been standardized. Rechenbach is a 641 page dictionary with 401 catchwords under letter A, TUKI, a dictionary of 325 pages, has 528 catchwords under letter A, and Feeley with its 179 pages has 298 catchwords under letter A. All three dictionaries describe Standard Swahili and owe much to their predecessors, especially to Johnson (1935 and. Above we noted the double consonant and variant spellings as main features found in pre-Johnson dictionaries. Johnson (1935) eliminated both of these. The post-Johnson dictionaries adhered to the noninclusion of double consonants in Swahili words but not to the exclusion of v~riant forms. As can be seen from Table IV Table IV Catchwords with Variant Spellings in post-Johnson Dictionaries    Rechenbach (1968) has the highest number of variant forms, higher even than the pre-Johnson dictionaries as was shown in Table ill. The author adopted a retrogressive approach to dictionary making by recording every plausible variant that Swahili speakers could form. Rechenbach was not interested in continuing the process of standardizing Swahili started by Johnson (1935). TUKI's 10% is equally large for a dictionary of Standard Swahili. Feeley's number is acceptable if one bears in mind that in any language there are some words with variant spellings which are all accepted as standard. Rechenbach and TUKI picked most of the variants from the pre-Johnson dictionaries which Johnson (1935) had dropped in his effort to standardize Swahili. Both of them undermined the role of the dictionary in standardizing a language and eroded the whole concept of Standard Swahili as Table V below shows.   Table V Variant Spellings in Rechenbach (1968) and TUKI (1981) Rechenbach(1968) aheri, akheri ahiri, akhiri ahsante, asante asubuhi, asubui, asubukhi atamia, tamia, latamia TUKI (1981) afandi, afande, afendi agenda, ajenda aghlabu, aghalabu alimradi, alimuradi, ilimuradi, mradi, muradi angalau, angalao, angaa; ngaa The above sample upholds Krumm's observation made in 1940 and quoted in the introduction of this paper: "The question of the orthography of Swahili written in Roman characters is not yet definitely solved." It shows the following: (1) Both dictionaries seem undecided on what is more accepted and widely used (i.e. the standard form) to such an extent that every sound is recorded. See atamia and asubuhi in Rechenbach (1968), and angalau and alimradi in TUI<I (1981).
(2) Both of them record a loanword in two forms: (a) as it is pronounced in the source language and (b) as it is written in and adapted to the Swahili alphabet. Cf. eksidenti and aksidenti; ateri and arteri; ajenda and agenda.
(3) Both dictionaries record Arabic loanwords first as they are pronounced in Arabic (with grpup consonant or aspirated kh), and then adapt it according to the syllabic law of Bantu languages, i.e. a consonant is always followed by a vowel. Cf. akrabu, akarabu; abwabu, abuwabu; aghlabu, aghalabu; arbaini, arubaini etc. and the deletion of the aspiration and the stop /k/, e.g. akhera, ahera; akheri, aheri etc.

Relevance and Realities of the Decisions of the Interterritorial Conference of 1928
The adoption of the Zanzibar dialect as Standard Swahili for East Africa and all the resolutions of the Dar es Salaam Conference implied that the Mombasa dialect (Kimvita), which had already literary works written in it, would be confined to spoken and informal communication. In literary works it was to be abandoned in favour of the standard dialect. It was for this reason that the East African Literature Bureau was established in order to publish Swahili works which had been certified by the Interterritorial Language Committee as written in Standard Swahili. However, some writers continued writing literature in Kimvita and as a result of this, dialectal forms continued to find their way in both literary works and dictionaries. This seems to explain why Johnson (1939) failed to adhere to the standard orthography as he attempted in his previous dictionary Oohnson 1935). Consequently he recorded more variants, as can be seen from Table ill above which shows that 10% of the catchwords under letter A in Johnson (1939) were variants in contrast with Johnson (1935) which had only 1%. Likewise, the editors of TUI<I (1981) who endeavoured to compile a standard Swahili dictionary, Kamusi ya Kiswahili Sanifu, recorded 10% dialectal forms.
In essence, the Chiponda Committee created words that did not exist when it selected forms of the standard dialect which were alien to speakers of other dialects. For example: asili (instead of asli), ishirini (ashirini), tisini (tisaini), sheria (sharia), lakini (ilakini), hasa (haswa) etc. Since the later spelling was preferred to Swahili under Arabic influence, the former were resisted in favour of the latter.
Dialectal variations are still generated in Swahili, even for words which entered the language in recent years. Cf. ateri and arteri, eproni and aproni, edita and editori, pension and pencheni, hospitali and spitali etc. The variants reflect different sections of the speech community using the different forrhs. The essence of these variants is that the loanwords are adopted in two different ways, namely as they are pronounced e.g. edita, eproni, ateri, pensheni, etc. and as they are written in their original language with some adjustments to conform to Swahili structure, e.g. editori (from the English word editor), aproni (apron), arteri (artery), hospitali (hospital) etc. There is also the problem that the loanwords are not pronounced correctly. For example, pencheni (pension) or spitali (hospital).
Indeed one could argue that we are not yet to see the end of dialectal words in Swahili dictionaries because many of these are found in contemporary literature read in schools. Cf. A. Abdala's Sauti ya Dhiki, or A. Nassir's Malenga wa Mvita. All these show that the exercise of standardizing orthography is an ongoing process and lexicographers and writers have to be active participants in promoting the standard.

What dict~onaries should do to promote the Standard Orthography
We noted in the introduction that a standard language is a medium of communication in government administration, education, the press and all forums where a degree of formal communication is needed. In order to ensure that communication is not hampered, a standard language and indeed the standard orthography have to be adopted and consistently used so that it can spread within and even without the speech community. Standardization of a language is an arbitrary decision made by language planners (1) to choose a language or dialect of a language as the norm, (2) to simplify spelling of words of the norm and (3) to adopt one form 'where a word has variant forms, and popularize it.
For the dictionary to be able to execute this responsibility, it is recommended that lexicographers should record only the form that is considered to be standard or accepted by the majority of the Swahili speakers. It is however possible for a dictionary to include words of other social and regional dialects provided that (1) it states that objective unambiguously, and (2) it marks geographical and social status of the dialectal words. Contrary to this, the dictionary would confuse and mislead the users, and would be condemned by the readers and the reviewers. This is what befell Gove and his dictionary, Webster's Third New International Dictionary of the English Language in 1961 when the critics called the dictionary "a calamity, an exponent of anarchy in language and a disaster" (Lodwig ar.d Barrett 1967: 56). The readers and reviewers criticized the dictionary because according to them it failed in its responsibility to set forth and maintain standards of usage for the language.
Such criticisms could be made against Rechenbach (1968) '4'd TUKI (1981 because of the superfluous variant spellings of catchwords. See Tables IV and V. It is important for dictionaries to adhere to the standard language and what is considered to be the most accepted spellings of words because it has a role of standardizing, disseminating and teaching the norm.

Conclusion
In this paper we endeavoured to examine the contribution of Swahili dictionaries in setting an orthography for Swahili words which culminated in standardizing the orthography of Standard Swahili. In doing so, we traced the beginnings of writing Swahili in Roman characters and noted (1) the variant spellings in different dictionaries, (2) the creation of the norm and (3) how the standard orthography was realized. Finally we examined the role of the Swahili dictionaries over the last 100 years in setting Standard Swahili and standard orthography and why it is important for dictionaries to uphold standard orthography.