Kitaab-Afrikaans : The evolution of a non-white literary variety at the Cape ( 1856-1940 )

This paper presents the socio-historical and linguistic context of Muslim Cape Dutch/Afrikaans literature, using the same texts as in Stell, Luffin, and Rakiep (2007). It then focuses on the areas of greatest diachronic variation in the phonology, lexicon, idiomaticity, morphology, and syntax of the Cape Malay texts. Finally, it attempts to place that variation within the perspective of the evolution of Cape Dutch/Afrikaans.


Introduction
The first significant appearance of Cape Dutch/Afrikaans as a literary medium has been situated in the satirical dialogues published in the journalist Meurant's Cradocksch Nieuwsblad from 1860 onwards (Pienaar 1943: 73;Scholtz 1964: 170).The focus of the academic community on the White and Christian-dominated Afrikaans language movements means that little attention has been paid to the evidence of an older Cape Dutch/Afrikaans literary tradition cultivated among Muslim Cape Coloureds, often referred to as "Cape Malays".Descending mainly from the Asian slaves brought by the VOC, the Cape Malays developed from an early stage a distinct religious culture through their cultivation of Islam, as well as a distinct linguistic identity through their connections with the Dutch East Indies and the Islamic world.These cultural idiosyncrasies resulted in a relatively late nativization of Cape Dutch, as well as the appearance in the early 19 th century of a local Muslim literature that used as a linguistic medium a distinct variety of Cape Dutch/Afrikaans.This variety was known, among others, as "Kitaab-Hollandsch", a term which stressed its vernacular identity, as well as its literary and religious character (Davids 1992: 101).This variety initially used only the Arabic alphabet, until knowledge of the Roman alphabet spread among the Cape denominator, Malay was the Asian language most prestigious and widespread among the established Asian Cape Muslim community.This language enjoyed some prestige outside of that group as well (Valkhoff 1966).On the other hand, Asian Portuguese Creole may have served predominantly as an oral link among the constant inflow of Asians from geographic zones where Malay was not usual (Ponelis 1993).The fact that Malay was the medium of instruction at the early madaris (Islamic schools), and remained so until the early 19 th century (Davids 1987: 46;Ponelis 1993), suggests that it was in a position to subsume all locally represented Austronesian varieties from the Malay Archipelago.As a lingua franca among Nusantarans, Malay was bound to become more widespread than its mutually unintelligible Indian contenders. 4Its genetic proximity to Malagasy, the numerically strongest Austronesian language in the Colony in terms of native speakers, may have further strengthened its leading position.
From Kitaab-Hollandsch to Kitaab-Afrikaans 93 Knowledge of Dutch was probably not uncommon among the early Muslim community.
Dutch possessed a high social status in the colony, and its diffusion was probably stimulated by the close contact and miscegenation between Asians and Dutch-speaking Hottentots. 5   The turn of the 18 th century brought major social change that considerably broadened the ethnic and linguistic connotations of the label "Cape Malay".This social change originated in the promulgation of the 1770 amendment to the 1642 Statuten van Indië, which stipulated that Christians were to instruct their slaves in the Christian religion, and that, once baptized, those slaves were to be given the right of purchasing their freedom.The fear of Christian slave owners of losing their property encouraged a laissez-faire attitude that left the field wide open for Mohammedan proselytisation (Elphick and Shell 1979: 191;Shell 1994: 356-362).As the British began to abolish slavery during the early 19 th century, large numbers of "new" Muslims of non-Asian extraction were released from their rural bondage and re-settled in Cape Town, which, at the time, was the only Islamic centre in the Colony (Shell 1997: 275).
The fact that the label "Cape Malay" survived the ethnic diversification of Cape Muslims is probably due to Cape Town's Muslim social establishment remaining of Asian extraction.The Asian identity of that establishment could be cultivated, as it was perceived by new converts as worth identifying with (Franken 1953: 41-79, 116-122;Davids 1980Davids : 5-6, 1987: 42-43): 42-43).It seems, however, that the core of established Cape Muslims had to make cultural concessions to accommodate the new ethnic components of the community by gradually relinquishing its Asian linguistic connection in favour of Cape Dutch/Afrikaans.Among the factors contributing to this change was the British ban on slave imports in 1808 which severely reduced the functionality of Creole Portuguese and Malay as oral links with non-assimilated newcomers from Asia.A second factor was the rural origin of the non-Asian Muslims.Many of the new Muslims had been more subject to Dutch influence and had possibly been linguistically assimilated into the European community (Elphick and Shell 1979: 225-226).
However, despite the presumable linguistic barriers, the new ethnic groups that settled in the Bo-Kaap by the early 19 th century appear not to have been prevented from social interaction with the Asian group, although they may for a while have continued to be perceived as distinct. 6The disappearance of both Malay and Creole Portuguese by the turn of the 20 th century confirms that the components forming the Cape Muslim community may have merged into an ethnically all-inclusive Cape Malay group, of which a certain variant of Cape Dutch/Afrikaans was the most natural linguistic attribute (Ponelis 1993: 17).The rise of that Cape Dutch/Afrikaans as a lingua franca among the Cape Muslim community in the 19 th century is confirmed by its gradual introduction as both oral and written medium in madaris, which facilitated the perpetuation and transmission of its idiosyncrasies.These indiosyncrasies could be cultivated for as long as Coloureds were left in a state of relative exclusion from the government-sponsored education system, through which Standard (Std.) Dutch/Afrikaans and English could be learned (Horrell 1970: 10, 14;Davids 1980: 45;1987: 44-45, 49;Hoosain Ebrahim 2004: 54-57).

Overview of Cape Malay Dutch/Afrikaans literature and the corpus
The Cape Malay Dutch/Afrikaans literary heritage is not readily accessible.Efforts to trace it for academic research are quite recent, ranging from Van Selms (1951) to Davids (1992).Davids (1987: 49) puts the figure of discovered Cape Dutch/Afrikaans texts using the Arabic alphabet at 74, and the number of texts discovered in Roman alphabet, which can be called "Cape Malay" in terms of linguistic features, around 20.Most of these texts are devoted to strictly religious matters, and a few to linguistic matters revolving around the Arabic language (Muller 1962: 39-41).
With respect to Cape Malay Dutch/Afrikaans literature using the Arabic alphabet, Muller (1962: 39-40) suggests that the printing of the Bayanûddin marked the beginning of a period of inertia lasting until 1890.Davids (1987: 49), in contrast, proposes that "writing in Arabic-Afrikaans was one continuous process which started in 1815 and continued until 1957".The origin of "Arabic-Afrikaans" 7 as a graphic variety of Cape Dutch/Afrikaans can be traced back to the need for a comprehensible written medium of instruction for the use of Cape Town's madaris.Initially using Malay, madrasah students gradually replaced the Malay language with their Dutch-based vernacular, which they transposed into the Arabic alphabet in their notebooks (Davids 1987(Davids : 47, 1992: 109): 109).The cultivation of this written tradition can From Kitaab-Hollandsch to Kitaab-Afrikaans 95 be regarded as the Cape Muslim intelligentsia's response to the inaccessibility and/or unacceptability of English-dominated secular education amongst its members, i.e. the only environment in which the Roman alphabet could be learned formally (Horell 1970: 14;Davids 1992: 150-151;Hoosain Ebrahim 2004: 56-57).
The government's steps towards compulsory education for Coloureds, as well as a gradual change in community attitude, are the factors that favoured the introduction of the Roman alphabet into Cape Malay literature, inaugurated with the publication in 1898 of Imam Abdurakib ibn Abdul Kahaar's Kitâb Tarajomatarriyaadil Badîati (Kähler 1971: 196;Davids 1992: 96, 99) 8 .Texts in Cape Dutch, and later on in Afrikaans, have continued appearing ever since.
The most thorough diachronic survey of evolutionary trends affecting the Cape Malay literary varieties is that by Davids (1992), whose main focus of attention is restricted to orthographic matters.Davids (1992: 110) identifies two evolutionary stages in Arabic-Afrikaans literature.
The first stage is characterized by the use of the Jawi script, imported from Nusantara, as well as conflicting approaches to vocalisation.The later stage is characterized by a general compliance with the new orthographic conventions fixed by Abu Bakr Effendi in his Bayanûddin.The first texts in Roman alphabet seem to have been lacking in orthographic consistency, despite an observable pursuit of Dutch conventions.Dutch conventions were totally superseded by Afrikaans ones only from the 1940s onwards (Muller 1962: 40;Davids 1992: 110). 9 Other areas, such as lexicon, morphology and syntax, have so far received attention in Ponelis (1981), Kotzé (1984), Waher (1994), andDavids (1992).Despite their fragmentary character, Davids' (1992) observations in these respects provide interesting clues.Davids (1992) presents evidence of strong idiolectal differentiation between authors.For example, the variety used by Abu Bakr Effendi in his Bayanûddin displays syntactic features found nowhere else in Cape Malay literature.Further, the remarkably frequent use of the Malay lexicon by one author from Port Elizabeth leads Davids (1992: 156-159) to speculate on the existence of an Eastern Cape literary variety opposed to a Western Cape literary variety.
Regardless of idiolectal specificities and chronology, the author assumes the existence of a linguistic gap between Roman and Arabic texts.Cape Malay Afrikaans texts using the Roman alphabet would bear distinctive marks of Std.Dutch, whereas Cape Malay Afrikaans texts using the Arabic alphabet would form a more accurate reflection of real Cape Malay spoken usage (Davids 1992: 97-98;Muller 1962: 5).Putting his observations in a diachronic perspective, Davids (1987: 50-52) (Stell et al. 2007).The works of a total of 15 authors are involved in the study.
All the texts deal with religious issues related to the practice of Islam and the nature of faith.
All the texts are translations from Arabic originals.With the exception of Abu Bakr Effendi, all authors may be assumed to be native to Cape Town and surrounding areas.
The varieties used in the texts generally display such a strong Afrikaans character throughout the period involved, that the current norms of Std.Af. were used as a yardstick for identifying primary features, rather than those of Std.Dutch.For the sake of conciseness, an effort has been made to mention only statistically significant non-standard features.The extent of variation is only quantified for selected features that are deemed representative of general trends.Regardless of statistical significance, mention is also made of those non-standard morphological and syntactic features described by Ponelis (1981) in his account of the Bayanûddin's variety, or singled out by Deumert (2003: 221-222) for frequency analysis from the Corpus of Cape Dutch Correspondence (CCDC).Other useful points of comparison, especially in the realms of morphology and syntax, are the Straatpraatjes columns from the APO newspaper (1909APO newspaper ( -1922)), the early 20 th century platform of the Coloured political elite (Adhikari 1996;Pheiffer 1996;Ponelis 1996), and Di Patriot (1875-1905), an early mouthpiece of Afrikaner nationalism (Steyn 1931).

Phonological characteristics of the corpus
The matching of orthographic practices with dialect descriptions is the only means to speculate on the phonological patterns of Cape Muslim Dutch/Afrikaans.One can summarize the evolution of the Arabic-Afrikaans alphabet as a two-stage process, the first being experimental, and the second being marked by a higher degree of conventionalisation inspired by Abu Bakr Effendi's work (Davids 1991(Davids , 1992: 110-111): 110-111).Davids (1992: 301-302)  The Roman texts in the present study were ordered into a "Dutch period", represented by the texts covering the years 1898-1930, and an "Afrikaans period", represented by the work of Hanif Edwards.In the texts from the Dutch period, there is a certain degree of compliance with general Dutch orthographic conventions, as illustrated by the use of <ij>, <z>, and <ch>.
In a time when "Classical Dutch" and "Kollewijnian Dutch" were competing in South Africa (De Villiers 1936: 160-187), there is also substantial free -The graphemic combination <oe>, used where in Std.Dutch/Af.<o> or <oo> would be expected, as in soe (Std Af. so, "so"), hoegste (Std.Af. hoogste, "highest"), gloewe (Std.Af. gelowe, "beliefs").These orthographic irregularities may have been intended as a means to render the South Western Afrikaans raising of close mid vowels.
-An intervocalic <g> is encountered where Arabic texts have an intervocalic [kh].
Intervocalic <g> was described as optional by the early editions of the Afrikaanse Woordelys en Spelreëls (AWS), until it was eventually scrapped in the 1968 edition.
To visualize some of the orthographic/phonological trends over the timespan covered by the corpus, the evolution of the syllabisation of word-initial and word-final consonant clusters in the Arabic texts (respectively A and B), and the same two features in the Roman texts (respectively C and D), were reproduced, as reflected by Figure 1.As Figure 1 shows, Abu Bakr Effendi followed and even amplified a pre-existing graphic tradition of syllabizing word-initial and word-final consonant clusters (A and B).From Hanif Edwards onwards, syllabization of word-initial consonant clusters almost completely disappears, which may reflect the decline of that (possibly Malay) feature in real spoken usage, as observed by Kloppers (1983: 284-285).On the other hand, syllabization of wordfinal clusters remains stable, but is strikingly less represented in Hanif Edwards' work, even though this phonological phenomenon is still commonly affecting today's spoken Afrikaans.
In contrast, there are only marginal traces of syllabization in Roman texts (C and D in Figure 1).The observed mismatch between Arabic and Roman texts may confirm Davids' view that the Arabic and the Roman varieties were not intended to reflect Cape Malay Dutch/Afrikaans speech in similar terms.

Lexical characteristics of the corpus
Reflecting the linguistic background of the Cape Malay community, the corpus contains loanwords from Classical Arabic, Malay, and English.These loanwords are embedded in a Dutch/Afrikaans-based matrix lexical variety, which may differ in its characteristics from the Std.Dutch/Af.lexical varieties.
Part of the Dutch lexical stock, but never recognized as standard by Afrikaans lexical sources, there are a few lexical forms that appeared to be in widespread use across the Cape Dutch/Afrikaans speech community.Among these are the variant mut ("with", Std.Dutch/Af.The Arabic lexical component is mostly comprised of substantives/adjectives referring to Islamic concepts, such as h alāl ("any action or thing which is permitted or lawful"), ramad ān ("the ninth month of the Islamic calender during which fasting is required"), wājib (as substandard "obligation", "necessity", as adjective "obligatory", "necessary").Other than isolated words, Classical Arabic religious formulae regularly occur, such as muhammad sallā -llāhu "alaihi uasallama ("Mohammed, may Allah bless and hail him").
The English lexical component is diverse, but is mainly comprised of content words.
Examples are the adjective permit ("allowed", Std.Dutch/Af.toegestaan), the noun koort   As regards the English component, there is a slight increase in its frequency, as well as a diversification of its content, perhaps in reflection of the authors' improving knowledge of English.The English lexical component of the texts does not seem to be as frequent and diversified as that attested, but unfortunately not quantified, in Straatpraatjes (Ponelis 1996: 135-138;Pheiffer 1996: 154-158).
The Malay component remains stable, although it shows a trend towards impoverishment.both died out as spoken languages by the turn of the 20 th century.On the other hand, Malay lexicon was still used on account of its religious significance.As a final remark, no significant difference can be found between the lexical trends affecting the Arabic and the Roman parts of the corpus.

Idiomatic characteristics of the corpus
The term "idiomatic" is used here to include both idioms and habitual collocations (Fernando 1996: 33-37).Although idiomaticity implies a degree of conventionalization that cannot be systematically deduced from the data, the source of Cape Malay Afrikaans idiomaticity is considered here, i.e. patterns of literal or non-literal lexicogrammatical sequencing which are not reverberated in Std.Dutch/Af.Among the lexicogrammatical sequences which occur in the texts, separable verbs, prepositional verbs, sequences of verb + [+human]/[+animate] 104 marker vir, sentence adverbials involving a preposition, and finally semantic units which in terms of their composition could qualify as semi-clausal expressions (fixed collocational sequence noun/adjective/adverb + verb) were singled out for study (Fernando 1996: 41).
All separable verbs that occur in the corpus do exist in Std.Dutch/Af., albeit sometimes with a different meaning, as is the case with uitwys, which occurs with the meaning of "to refer" in association with verbs, which sometimes may theoretically be transitive (Raidt 1976: 84-93;Den Besten 1980: 141-179;Ponelis 1996: 272-275).This use of vir is illustrated by the sentences in (4). (4)

Allāhu ta 'ālā het duidelik gemaak het die vertel vir nabī Muhammad…
Allah explained the story of the creation to the Prophet Mohammed Allāhu ta 'ālā sal maak vir hom lewendig

Allah will call him into being
The prepositions used by the authors to introduce sentence adverbials, i.e. "semantic prepositions" underdetermined by the verb, may not reflect Std.Dutch/Af.collocation patterns.This is illustrated by the misuse of by in the prepositional phrase in (5), whose nominal element normally requires in as a head.
(5) die meening van die oemrah by die taal van die Arab is om te besoek the meaning of umrā in the Arab's language is to visit The bulk of non-standard semi-clausal expressions occurs in the form of paraphrastic formulations, such as twyfelheid maak in (Std.Dutch/Af.twyfel/twijfelen aan, "to doubt in"), maatskappy wees met (Std.Dutch/Af.begelei(den), "to accompany"), gelykens maak (Std.Af. vergelyk, "to compare").A number of semi-clausal expressions can incidentally bear the mark of Arabic, such as salaat maak ("to perform the Islamic ritual prayer").In other cases, we find loan translations from English, as in dra getuie van (Std.Af. getuig van, "to bear witness to").Prepositional verbs seem to undergo some measure of standardization.The high frequency of vir as a [+human]/[+animate] object marker in the dative case in the Dutch period of the corpus, compared with its relative underrepresentation in the CCDC, seems to confirm its previous status as a basilectal non-white speech marker (Raidt 1976: 92-93;Deumert 2003: 209-211).The use of non-standard semi-clausal expressions decreases only slightly.This can be ascribed to the persistence of a relatively stable core of semi-clausal expressions featuring an Arabic element.
Abu Bakr again shows his linguistic originality in an over-representation of non-standard prepositional verbs, and in an under-representation of non-standard semi-clausal expressions.
This could be accounted for by the author's defective knowledge of both standard Dutch and Cape Malay idiomaticity.

Morphological characteristics of the corpus
Standard Dutch and Std.Af. morphology differ considerably.However, not all morphological variation in the corpus can be directly attributed to the effects of competition between Std.
Dutch and Std.Af. morphological rules.Some of the morphological variation may be ascribed to the free variation which was characteristic of early Std.Af.Besides this, some of the nonstandard features in the corpus may be of Malay or Classical Arabic origin.
The most unmistakable traces of Dutch influence, i.e. of the kind that was condemned by early editions of the AWS, occur before Hanif Edwards.Among them are a few bound morphemes, such as the plural marker -en in dagen ("days", Std.Af. dae), the ending -en in the infinitive form wassen ("to wash", Std.Af. was), and in the past participle vergelyken (Std.Dutch vergeleken, Std.Af. vergelyk/vergeleke).There are also traces in the early Roman writings of verb agreement, in the form of the 3 rd person singular-t in begint ("begins", Std.
Af. begin), and the 1 st person form heb of the auxiliary/main verb hebben ("to have", Std.Af.

hê).
Reflecting the hesitation in feature selection that characterized early written Afrikaans, one finds many of the "archaizing" features on which the AWS did not originally take a clear stance.Among these features are the preserved etymologic fricatives and gutturals in stem- There is a range of features, non-standard in both Dutch and Afrikaans terms, which are not quoted by Steyn (1931) and Deumert (2003).At the level of lexical derivation, there are regular cases of irregular generalization of prefixes/suffixes (cf.section 3.1), and a cumulation of suffixes, as in wetlikigheid ("legality", Std.Dutch/Af.wet(te)li(j)kheid; cf.Waher 1994).
Cases of irregular plural formation with the suffix forms -heide or -geite occurring instead of Arabic lexical items may sometimes remain subject to Arabic morphological rules.In this line, we find Arabic nouns displaying the Arabic plural morpheme -in, the "sound plural" or "broken plural" marker, as in the substantives muslim, whose plural form is found as muslimin, sifa ("attribute") that becomes sifāt and rukn ("pillar") that becomes 'arkān.The plural morpheme -s may nonetheless occur for the plural marking of Arabic lexical stems, as in 'ālims ("learned ones", Arabic 'ulamā, singular 'ālim).Less frequent is the unequivocallly Afrikaans desinence -e, as in kitābe ("books").A frequent phenomenon is the juxtaposition of the Arabic plural ending with the -s suffix as in sifāts ("attributes", Arabic sifāt, singular sifa).
The only Malay lexical item occurring in a plural form, lebarang, is found with an -s, while in Malay no plural suffix existed. 16Malay verbs are fit into the Afrikaans verbal derivation pattern, as shown in gabatjah ("read", from Dutch/Af.ge-+ Mal.batjah).
English loanwords mostly carry the suffix -s.It cannot be said with certainty whether this suffix is the normal English plural suffix or an overgeneralized Afrikaans -s plural suffix.
Most non-standard morphological patterns in the corpus can also be found in the CCDC.
Certain meso-/acrolectal features from the CCDC discussed by Deumert (2003) may be represented in the present texts, albeit sometimes to a negligible extent.These features comprise personal and relative pronouns whose basilectal forms are firmly established throughout the corpus, and only have to compete occasionally with mesolectal/Dutch forms (respectively zulle, die/dat without gender-agreement property) during the Dutch period (Deumert 2003: 185-187, 189-193).There is no visible competition between the Dutch and the Afrikaans patterns of adjectival inflection, in the sense that suffixing with -e is generalized, save in the case of a few polysyllabic Latinate adjectives (Deumert 2003: 169-175).Variation at the level of verbal person agreement is very limited, except in the early Roman writings and in the case of atematic verbs (Deumert 2003: 143-145).A striking similarity between the CCDC and our corpus is the sudden appearance of Std.Af. hier-/daardie.These demonstratives remain unknown up until the Afrikaans period, at which stage they totally obliterate the previous mesolectal form deze. Finally, some of the meso-/acrolectal forms identified in the CCDC do not occur at all in the present texts, particularly the preterite (the only preterite form occurring in the corpus is that of the copula wees) and irregular past participle formation.Some doublets discussed in Deumert (2003: 187-188), such as my and the formerly widespread possessive and object pronoun myn, unfortunately occur too sporadically in the present corpus to allow any meaningful cross-comparison.
There are further similarities with other corpora.These similarities involve the instances of irregular plural-marking, which are represented in Di Patriot (save in the case of the irregular plural formation of -heid into g/heite (Steyn 1931: 49)).Attested in Straatpraatjes and/or now recognized as characteristic of South Western Afrikaans, are instances of possessive forms involving an expletive enclitic syn/se (Pheiffer 1996: 150;Ponelis 1996: 229-230), irregular ge-prefixation of atonic verbal prefixes (Pheiffer 1996: 151-152), and the use of het as infinitival (Pheiffer 1996: 151).

Syntactic characteristics of the corpus
The main areas of syntactic variation found in the texts involve the position of the verbal element and the order of succession of its components, and negation patterns.Alongside SOV order, characteristic of both Std.Dutch and Std.Af., a cursory glance at the corpus reveals a strong presence of non-standard VX in the form of SVO, as in ( 6).On the other hand, evidence of influence from the typically Classical Arabic VSO pattern can be found in the early stages of the corpus -especially in the Bayanūddīn -as in ( 8).

(8) [gaat]V in die istignai [elf sifaats]SUBJ
the divine absolute is comprised of eleven attributes If we leave out of consideration these sporadic instances of VSO, the corpus presents a continuum of compromise forms between SXV and SVX, which do not seem to reflect any syntactic logic other than that of an interlingual stage.In its full extent, that range of variants is best observable independent clauses featuring an infinitival, and in dependent clauses.In such SVO clauses, there may be traces of a satellite-nucleus order (Ponelis 1979: 527-533).
Adjectival predicates/adverbials are mostly found preceding the pivot verb, as in ( 9).The rightward movement of the auxiliaries wees, word and het does still allow argument (object or subject) extraposition, as in ( 12) and ( 13).Af. pattern of double negation, there are occurrences of a single operator nie(t), especially in the period 1881-1930, as illustrated in (14).Multiple negation may also occur, especially in the period 1856-1881, as in ( 15). ( 14) is niet nodig manne spuit men need not spit (15) …en om nie kwaai vriende te bly nie meer as drie dage nie …and not to remain enemies for longer than three days As regards other syntactic features, some patterns quoted by Ponelis (1981: 76) from the Bayanūddīn can be found, such as elision of the 3 rd person singular subject, which occurs to a significant extent until the end of the 19 th century.Some of the other features characteristic of the Bayanūddīn occur only exceptionally or not at all in the other parts of the corpus, namely article-deletion, raising of interrogative pronouns, lack of inversion, positioning of nie-1 before the pivot verb in independent clauses, expression of past tense by means of past participles unaccompanied by a tense auxiliary.The construction of dependent clauses as independent clauses is likewise exceptional or absent, save in the case of conjunctive clauses where Std.Af. (as opposed to Dutch) allows dat-deletion, as in ( 16).
(16) bayang van mai vriende het gevraag vir my ø ek moet vertaal die riesalaat many of my friends have asked me to translate the risalaat The diachronic relation between SVX and SXV throughout the corpus in illustrated in Figure 4, where (A) reflects the frequency of clauses in which the adverbial particle or predicative adjective incorrectly occurs after their governing verb, and (B) reflects the frequency of clauses in which the verb incorrectly occurs before the direct object.which is attested in Straatpraatjes (Pheiffer 1996: 153), and is nowadays observable in spoken Cape Malay Afrikaans, Klopper (1983: 289) notes that "(w)hereas Standard Afrikaans in subordinate clauses as well as in main clauses featuring an auxiliary/modal is characterized by an (S)OV order, we find evidence of (S)VO, again under influence of English, in the lects of mainly working class Malays and Coloureds who have grown up in the Cape Town area." Since SVO is found consistently from the earliest stages of the present corpus, it is not certain that English played a role in establishing that pattern. 19A more likely source could in fact be Malay, which is of the SVO type. 20The prevalence of SVX order with X=direct object may initially have been inspired by Malay.The decline of that language at a time when English had still not taken a firm hold among the Cape Malay community may have left the field open for standardization towards Std.Dutch/Af.That process of standardization may eventually have lost its momentum as a result of growing familiarity with English.Another hypothesis is that SVX represents a vestige of a Creole SVO stage, in which substrate and superstrate did not have a direct influence (Muysken 1988: 290).English, again, would have revived that remnant structure.problematic.Deumert (2003) makes no mention of it in the CCDC, whereas it is attested in Straatpraatjes (Pheiffer 1996: 152-153).The diffusion of multiple negation in our corpus as well as its common occurrence in Creoles suggest that it may, as much as double negation, have been a basilectal feature inherited from an interlectal stage (Bickerton 1977: 60-61).The dominance of nie-2 in our corpus (except in the period 1881-1898) contrasts sharply with its relatively low frequency in the CCDC, where only 21% of all negative clauses contain the Afrikaans brace negation (Deumert 2003: 202).This diffusion of nie-2 may confirm the "strong association with persons of color and with the lowest socioeconomic strata of Europeans" that has been seen in that feature (Roberge 2000: 145).

Not represented in
Certain meso-/acrolectal syntactic features attested in Deumert (2003) are not represented in the corpus.These features are the use of te where Std.Af. prescribes the sequence om te (Deumert 2003: 204-207).Some paradigms of syntactic use described in other corpora are unfortunately underrepresented or not represented at all in the present corpus.As a result, the Cape Malay positioning of the particle te within the infinitive clause, at a time when a large amount of cross-lectal variation was noted in this regard (Steyn 1931: 32;Deumert 2003: 205), cannot be documented.

Overview
In reaction to the traditional view that Cape Dutch had by the late 18 th century assumed a stable form essentially identical with that of today's Afrikaans, attempts have been made to demonstrate that, instead, Cape Dutch constituted a converging linguistic continuum in which basilectal and acrolectal features were potentially cohabiting in individual repertoires (Den Besten 1989: 234;Roberge 1994Roberge : 156, 1995: 81-83): 81-83).Partial to complete knowledge of both basilect and acrolect in actual spoken usage is attested by observations such as those by Swaving (1830in Scholtz 1951: 19), according to whom the "Bastard Dutch" from the Cape "is not totally foreign to even the most civilized members of the Christian upperclass", whereas on formal occasions trouble may be taken "to shed the most salient features of Cape speech".Also, despite evidence of an enduring ethnic differentiation of Cape Dutch varieties, it seems that that "Bastard Dutch" may have been spoken across the colour barrier as a nativized variety (Von Wielligh 1925: 94;Scholtz 1964: 170;Hahn 1982: 84).The valorization of the Afrikaans standard, and eventually its officialization in 1925, entailed the "legalization" of an array of basilectal features which in the process became stylistically neutral, or sometimes acknowledged colloquial forms.
This gradual evolution is reflected in written sources from the 19 th and early 20 th centuries, such as journalistic prose, private correspondence, and literature, which present us with evidence of cohabiting basilectal and acrolectal features.In the period preceding the establishment of the new Afrikaans norm, the graphic occurrence of basilectal features could be an involuntary indicator of a lower-class social origin, in a society where conformity with metropolitan Dutch was still regarded as a worthy pursuit.In its conscious written use, however, the use of the basilect could also obey jocular motives, or, in White society, convey an attachment to the rising Afrikaner ideology, as heralded by the nationalist press-organ Di Patriot .In the latter case, the idealized quest for a distinct (White) ethnic identity provided the framework for a "standard ideology" (cf.Milroy 2001), through which basilectal features were fed into a composite written norm, the ancestor of Std.Af., representing a standardizing compromise between local speech and Std.Dutch (Steyn 1931;Pienaar 1931: 163-171, 184;Deumert 2003: 236-240).
Despite some attempts at political rapprochement with Afrikaner nationalism, the political and cultural elite of the Coloured community did not overtly commit to the Afrikaans language ideal in the pre-recognition period (Davids 1987: 53-58).Therefore, the press-organ of the APO (1909APO ( -1922)), i.e. the privileged political platform of Capetonian Coloureds, allowed the self-conscious use of the basilect only in its parodial Straatpraatjes columns, whereas meso-/acrolectal forms and English remained the natural medium of formality (Adhikari 1996: 6-7;Ponelis 1996: 132-135).
In The use of vir as an object marker is the only feature whose frequency remains relatively stable across the corpus (cf.section 3.3).The reason for this is probably that vir was legalized in its object-marking function in the new Afrikaans norm, as a result of which no more convergence towards the Dutch model was needed.
Linguistic trends in the corpus do not always follow a linear curve towards the standard target.Falling in that category, the diachronic relation between SOV and SVO could be described as a case of "destandardization".The Dutch/Afrikaans SOV order seems to make headway only during the period 1881-1930, before eventually losing ground to SVO.On the other hand, certain non-standard features seem to stand their ground in the face of their standard counterparts.An example is objective vir as used with transitive verbs, where Std.
Last but not least, the present texts exhibit influence from an extraneous linguistic reference, i.e. that of Classical Arabic.This influence mostly comes to bear at the lexical level, where it does not seem to assimilate completely into the Dutch/Afrikaans matrix.This lack of assimilation finds expression in the increasing use of Arabic plural markers (cf. sections 3.2 and 3.3).As indicated earlier, possible traces of the Arabic VSO order are found, which do however decrease in frequency over the period covered.
Most features quantified here are relatively overrepresented in the Bayanûddin, which suggests that Abu Bakr Effendi's idiolect may not have been representative of his linguistic environment.Some of the un-Dutch/Afrikaans features quoted by Ponelis (1981: 75-78) as characteristic of the Bayanûddin, particularly elision of the past auxiliary, lack of inversion, anteverbal positioning of first negation, and VSO, are marginally or not at all represented in the other parts of the corpus.
Although perhaps not justified on the grounds of stylistic incompatibility, a comparison between the data from the present corpus and those from the (mainly White) CCDC can yield interesting insights.Some acrolectal/mesolectal features represented in the CCDC are remarkably absent in the present data (Deumert 2003: 221-222 Af. rather than Std.Dutch when the former achieved recognition.

Conclusion
If placed within the perspective of the standardization of Afrikaans, the data extracted from the present Cape Malay texts present a basilectal variety evolving in the direction of Std.
Dutch and afterwards Std.Af.Interesting, though, is that standardization in the direction of the superstrate is seconded, or sometimes even thwarted, by an autonomous process of feature selection and language engineering.Kitaab-Hollandsch has thus functioned as an "elevated basilect" in the shadow of Dutch and emerging Std.Af., deriving its in-group High status from its extraneous Qu'ranic and Classical Arabic references.Placing the present data within the perspective of the formation of Afrikaans, one finds in the Dutch period evidence of unfamiliarity with meso-/acrolectal varieties.This evidence provides insight into the limits of bidialectalism among the Cape Malay community in the period preceding the recognition of Afrikaans.One also finds suggestions of differences between the Cape Malays' spoken varieties and ORA, which may provide meaningful insights into the possible differentiation between early Slave Dutch and Hottentot Dutch.For these reasons, further research on the early history of Afrikaans can no longer afford to dispense with the Cape Malay writing tradition.

1.
Reflecting the prestige of the "Malay imagery" among non-Asian Cape Muslims, a British observer noted in 1848 of a group of Muslim Mozambicans that they "loved to be considered as Malays" (in Shell 1997: 276).
2. Klopper (1983) talks of "Kaapse Maleierafrikaans" as a variety proper to Cape Muslim Coloureds and distinct from that of Cape Christian Coloureds.

3.
The Nusantaran connection of the Cape Muslims has provided the materials for a "Malayist" ethnic imagery, purposefully exploited during the 20 th century to justify a specific political treatment from South Africa's White authorities (Davids 1987: 56-57;Haron 2000;Hoosain Ebrahim 2004: 47).The overstatement of "Malayism" in a racial sense also served the Apartheid government to justify a separate "Cape Malay" subcategory of the population group it was referring to as "Coloureds" (Kähler 1971: 4).

4.
The main regions of origin of the early Cape Indians suggest that they were comprised of an Indo-European-speaking group and a Dravidian-speaking group (Bradlow 1978: 103).

5.
In the court cases reviewed by Den Besten (2000: 958) (up to 1772), about one fifth of slaves knew Dutch and over two thirds knew either Creole Portuguese or Malay 6.
An early 19 th century British observer could still make out ethnic contrasts among one Mosque congregation in Cape Town, noting that the service he was witnessing was attended by "chiefly slaves, Malays and Madagascars" (in Tayob 1999: 29).

7.
The expression was first coined by Van Selms (1951), in specific reference to the version of the Arabic alphabet devised for vernacular writings.The sense in which Van Selms used the term "Afrikaans" included "Cape Dutch".

8.
The majority of Coloured pupils were attending Christian mission schools by the turn of the 19 th century, while government schools had "to all intents and purposes become reserved for white children only" (Horrell 1970: 14, 33).Lack of accessibility and fear of Christian indoctrination were adressed by the network of predominantly Malay, and afterwards Dutch/Afrikaans-medium, madaris (Davids 1992: 150;Hoosain Ebrahim 2004: 56-57).The first secular learning institution earmarked for Muslim children was the Rahmaneyeh Institute, founded in 1913 (Davids 1992: 150-151).Education for Coloureds was made compulsory only in 1945 (Horrell 1970: 37).

9.
Afrikaans spelling became effectively official as of 1925.11.Interestingly, the forms seg and leg are described by Malherbe (1917: 55) as not very common and confined to the north as mainly past participle stems, whereas the forms sê and lê would be more characteristically used in the Boland.The question arises as to whether Cape Town usage may in this regard have differed from that in the Boland, or whether the authors attempted to approximate Dutch when using these forms. 12.
Although some early grammarians mention the possibility of inflecting atematic verbs (Bouman and Pienaar 1924: 113), the AWS has only tolerated free variation between gaan and its 3 rd person form gaat without grammatical specification until its 1991 edition, where the fossilized conjugated form was scrapped.The past participle gegaat has never earned recognition.Variation between inflectional -t and -n in that verbal paradigm is common in the CCDC (Deumert 2003: 144-145) and Straatpraatjes (Adhikari 1996: 150-151).
13. Van Oostrum and Kritzinger (1923: 97) mention that "some verbs use is instead of het: die trein is om agtuur aangekom.This usage does however sound contrived and is therefore in the process of disappearing".
16. Modern Indonesian and Malaysian, which constitute modern continuations of Malay, form their plurals by duplication, or do not mark it when it is inferable from the context.
21. Kloppers (1980Kloppers ( , 1983) ) has found that the spoken Afrikaans variety of Cape Muslims ("Kaapmaleierafrikaans") has not yet come as close to the Std.Af. target as has the variety of the Western Cape Christian Coloureds.In Klopper's (1983: 98) terms, the linguistic gap between the two groups must be explained in terms of the Muslims' "strong identification with their religion", which makes them "less receptive to social pressure", whereas "Christian Coloureds, who belong to a wide range of denominations and generally aspire to the same values as do the whites, are more sensitive to social pressure" (cf.also Kotzé 1989: 253).
variation in the transcription of long vowels in open syllables: <oo> and <ee> cohabit with <e> and <o>, respectively.Certain words are also found in the form which they later acquired in Std.Af. spelling (e.g.het, "have", instead of Std.Dutch heb).Hanif Edwards obviously discarded Dutch conventions in favour of Std.Af. ones.Deviations from Dutch and Afrikaans orthographic conventions, potentially indicating local phonological characteristics, include the following:

Figure 2
Figure 2 illutrates the evolving shares of Arabic, English, Malay, Dutch/Afrikaans-based and "mixed" lemmata (excluding function-words, proper nouns and formulae) in the non-standard lexicon found in the corpus.

Figure 3 Figure 3 .
Figure 3 reflects the frequencies of prepositions used in incorrect association with prepositional verbs (A), of vir occuring as [+human]/[+animate] object marker in the dative case (B), and of non-standard semi-clausal expressions (C).
Std. Dutch/Af.-hede(n), or with stems ending in <d> (to be realized[t]  in stem-final position and[d]  in the flectional paradigm), as in afgotte (Std.Dutch/Af.afgode(n), "idols", from the singular Std.Dutch/Af.form afgod), 14 are also found in each period.At the pronominal level, especially in the Dutch period, are the periphrasitic possessives hulle syn/se and onse (possibly a derivate of ons se), which may have been inherited from an interlectal/Creole developmental stage.15In all periods, adverbial pronouns may occasionally be sequenced along a nucleus-satellite order without featuring the R-element, as in a(ch)gter dit, op dit, instead of Std.Dutch/Af.daara(ch)gter, daarop.At the verbal level, we find the form het (main verb "to have") fulfilling an infinitive function instead of the Std.Af. infinitive hê or its Dutch counterpart hebben.
het laat afkom]V [die kitape]OBJ truly Allah sent down the kitaabs Such examples suggest that the elements which in Std.Dutch/Af.should form the central boundary (i.e. the whole verbal element in dependent order, the infinitival, the verbal particles) are placed in a position contiguous to the core (i.e.mainly the subject, the pivot verb, free adverbials), thus preceding the midfield (i.e.mainly the direct or indirect object, free adverbials). 17The adverbial component of what should be the central boundary may likewise be contiguous to the core, along a nucleus-satellite order, as in (7).It therefore seems that the notion of a central boundary in Std.Dutch/Af.terms does not universally apply to the syntactic varieties that we are presented with.(7) dan moet [gooi]V uit van die put… then you must throw out of the well… (9) die moeslim (…) moet staanbaar [maak]V [die pilare (van sy geloof)]OBJ the Muslim has to maintain the pillars (of his faith) (Hanif Edwards)In accordance with Std.Dutch/Af., stranded prepositions are also mostly found before the verbal element, but do sometimes occur after the verbal element in apparent reflection of the English pattern, as in (10).(10)een skuld [wat]REL jy nie bewysen [het]V voor nie a guilt that you have no evidence forAs regards the ordering of elements within verbal strings, consistent conformity with aspects of the Std.Dutch/Af."relict" satellite-nucleus order may be found in the positioning of auxiliaries vs. past participles in dependent clauses.Wees ("to be") and word ("to be/become") generally move away as passive auxiliaries from the V 2 position when they function as pivot, infinitival, or past participle, and occur to the right of the main verb's past participle.The tense auxiliary het (Std.Af. hê, "to have") is also -except in a few cases in the Bayanûddinsystematically found moving away as tense auxiliary from the V 2 position in dependent clauses.It occurs to the right of the past participle, as in the examples in (11).(11) wat weens [gemaak]PP is (…) moet [bygesit]PP word the profits made must be put aside die Nabie (…) het gesé toe hy [gevra]PP [gewees]AUXPP [het]AUX… the prophet (…) said when he was asked… Gerald Stell 112

( 12 )
ook as hy vir die aarme [gegee]PP [het]AUX [iets]OBJ also if he has given something to the poor (13) in die tweede blaas sal lewendig word [al die mahluqatas]SUBJ at the second blow all mahluqatas will come into being 18 (1856) With regard to negation, there are three competing patterns in the corpus.Alongside the Std.

Figure 4 .
Figure 4.The diachronic relation between SVX and SXV

Figure 4 ,Figure 5 .
Figure 5. Distribution of negating patterns contradiction to the values governing contemporary secular writing, the high concentration of basilectal features in the religious texts of the present corpus suggests that a Cape Dutch basilectal base could lend itself to formal functions in an in-group context.As those yardsticks for language dignity provided to Europeans by the Statenbijbel could not have an immediate value in the perspective of Islam, Muslim writers were inclined to carve a new High register out of their Cape Dutch variety and Classical Arabic.Dignification of the vernacular was ensured originally by the use of the Arabic alphabet (Davids 1991: 97-98), and more generally by its amalgamation with a sacralized Classical Arabic lexicon, or its semantic reflection of Islamic concepts.A process of stylistic engineering along in-group references is most conspicuous in the growing amount of Dutch/Afrikaans morphological derivations of Arabic lexical roots (e.g.wajibheid from Arabic wājib + Dutch/Afr.heid), and of new coinages replacing Arabic religious terms (e.g.manierte instead of sifāt) (cf.section 3.4).Arbitrary stylistic diversification is further illustrated by the occasional use of Classical Arabic VSO order (cf.section 3.5).In-group language engineering did not stop Cape Malay Dutch/Afrikaans from converging towards its superstrates.Convergence towards Std.Dutch/Af.is illustrated at the phonological level by the phenomenon of desillabization of consonant clusters (e.g.pelek > plek, cf.section 3.1).At the morphological level, we find in the Dutch period the generalized mesolectal gender-neutral deze, the use of which probably reflects a wish to approximate Std.Dutch norms (cf.section 3.4).At the syntactic level, we find multiple negation, stereotypical of Coloured usage, fading into a Dutch-modelled single negation pattern.Both multiple and single negation patterns subsequently disappear in favour of a Std.Af. double negation pattern (cf.section 3.5).Finally, we find increasing conformity to Std.Dutch/Af.in the positioning of separable verb particles and the positioning of verbs (the latter only during the period 1856-1930, cf.section 3.4).Because Std.Af. substituted itself for Dutch as an official target norm in the early 20 th century, traces can also be found of what one could term a process of "restandardization", affecting a range of Dutch-like features which are firmly established in the early parts of the corpus.An illustration of this "norm re-setting" at the lexical level is the sudden replacement of deze with Std.Af. hier-/daardie in Hanif Edwards's texts (cf.section 3.1).At the morphological level, intervocalic [g/x], a historical Dutch feature, eventually competes with the pattern of [g/x]-deletion in intervocalic position favoured by the AWS (e.g.hoge > hoë, cf.3.1).Also, the widespread deflected verbal pattern which occasionally cohabits with (often irregularly assigned) Dutch person agreement markers in the early Roman writings eventually achieves complete hegemony in Hanif Edwards (cf.section 3.4).The occasional relative pronoun die eventually gives way to Std.Af. wat (cf.section 3.5).At the syntactic level, the occasional single negation nie(t) disappears in favour of the Std.Af. double negation nie…nie from Hanif Edwards onwards.
from texts in the Arabic alphabet are rendered by using the DIN 31635 convention.In the subsequent sections of this article which are not relevant to phonology, example words or sentences from Arabic texts are rendered by using approximate Std.Af.Spelling, except in the case of Arabic lemmata or formulae.
2002) built a case for the possibility of an interlingual identification having been operated by Khoikhoi speakers between the Dutch and the Khoikhoi SXV orders.The author identified Khoikhoi post-verbal particles serving as past or passive markers on the basis of which the Afrikaans VP-final position of het and word in dependent word order could have been established in the Cape interlectal continuum (cf.also Ponelis 1993Malaysian and Indonesian, the purest forms of Malay must have been of the SVO type.Cape Malay may not have been different, but its Ceylonese and Indian secondary speakers may have been familiar with the SOV type (Den Besten finds that the literary varieties used by Cape Malay authors have been moving ever closer to a Std.Afrikaans (Std.Af.) target.This is the claim to be substantiated here.The texts singled out for the present study span roughly one century, from Ishmuni's Betroubare Woord (1856) to Shaykh Hanif Edwards' Die sterke fondament (written in the 1950s).The corpus comprises 22 texts.Of these texts, 11 are written in the Arabic alphabet, and 11 in the Roman alphabet.To facilitate a diachronic comparison, the texts were ordered into five successive "periods".In fact, the first two periods may rather be called "points in time", as they are each comprised of only one text (as no other contemporary text is available), i.e. die Betroubare Woord (1856) and the Bayanûddin (1869).The point of giving special treatment to die Betroubare Woord is to illustrate Davids' observation of "pre-Bayanûddin tendencies".Since it is perceived as a turning point in Arabic-Afrikaans literature, the Bayanûddin is treated separately.First among the "post-Bayanûddin" periods, the third period is comprised of five texts written exclusively in the Arabic alphabet, and covers the years 1881 to 1894.The Other innovations involve lexical stems foreign to Dutch/Afrikaans, to which is appended an Afrikaans derivative morpheme.This goes for the Arabic-based wājibkhayt ("necessity", from ", Std.Dutch/Af.gerechts-/geregshof),andthe verb translate ("to translate", Std.Dutch vertalen, Std.Af.vertaal).English function words are rare, but may occur in the form of, for example, the subordinate conjunctions āftir ("after") or wānever ("whenever") instead of Std.Dutch/Af.nadatand wanneer ook al. Interesting is that English lexical items may undergo some degree of morphonological adaptation, such as trawelaar ("traveller", Std.Af.reisiger) in the Roman texts, or ūder ("order", Std.Af.orde) in the Arabic texts.Forming the bulk of the non-standard Dutch/Afrikaans lexicon, lexical innovations occur regularly.Those innovations may be derivatives of Dutch/Afrikaans stems, such as verleiery ("temptation", Std.Dutch/Af.verleiding), tevredeskap ("satisfaction", Std.Dutch/Af.Gerald Stell 102 tevredenheid), gevoelte ("feeling", Std.Dutch/Af.gevoel), sterkheid ("strength", Std.Dutch/Af.sterkte), getuiging ("testimony", Std.Dutch/Af.getui(g)enis), and gewetenskap ("science, knowledge", Std.Dutch/Af.wetens(ch/k)ap, kennis).Arabic wājib + heid, Std.Dutch/Af.nodigheid), the Malay-based baiangheid ("multiplicity", Malay baiang + heid, Std.Dutch/Af.veelvuldigheid), and the English-based miksel ("mixture", from English mix + sel, Std.Dutch/ Af. mengsel).
Std. Dutch/Af.verwijzen/verwys).Sometimes, such semantically modified or recomposed separable verbs seem inspired from English phrasal verbs, such as afhou, which occurs with the sense of "to keep off" (Std.Dutch/Af.wegblijven/bly van).From Kitaab-Hollandsch to Kitaab-Afrikaans 105 Regarded as a remnant from a previous interlectal stage, possibly under Creole Portuguese or Malay influence, vir is regularly found functioning as a [+human]/[+animate] object marker verbs can be followed by an unpredicted preposition, as in geloof van ("to believe in", Std.Dutch/Af.geloven/gloaan).That preposition may semantically reflect the English preposition associated with the English synonym, as in twyfel in ("to doubt in", Std.Dutch/Af.twijfelen/twyfelaan),or the Classical Arabic synonym, as in praat op ("to talk about", Std.Dutch/Af.praten/praatoor,Arabic takallama 'an).A number of normally transitive Dutch/Afrikaans verbs occur with non-transitive properties, for example volvoer met (Std.Af.volvoer(en)), as in (1).Conversely, a number of normally intransitive verbs appear to be transitive, such as voldoen (Std.Dutch/Af.voldoenaan, "to meet/live up to/satisfy") in (2).(3) …om te maak in syn tyd al die ietse wat (…) is wajib op hom …in order to perform in its time all those things (…) which he must do final position, as in seg ("to say", Std.Dutch zeggen, from AWS 1991 onwards only sê), 11 or in intervocalic position, as in tyd/tyde ("time/times", Std.Dutch tijd/tijden, Std.Af. tye).Further, we find the pronominal form myn with either a possessive or an object function.Despite being Dutch in form, the determiner deze occurs, mainly in the Dutch period of the corpus, stripped of its gender-agreement property, i.e. before theoretically neutral nouns in Dutch.This also goes for the relative pronouns die or dat, likewise Dutch in form.Both occur in a few cases instead of wat, irrespective of the historical gender of their antecedents, before completely disappearing in the Afrikaans period.Irregular cases of past participle prefixation occur throughout the corpus in the form of the particle ge-affixed to atonic verbal prefixes, as in verdeel/geverdeel ("divided", Std.Dutch/Af.verdeel(d)) or geondersoek ("researched, investigated", Std.Dutch onderzocht, Std.Af. ondersoek).
hegemony in the inflectional paradigm (hoog > hoge) until Hanif Edwards.From Hanif Edwards onwards it finds itself increasingly challenged by the Std.Af.pattern of [x/g]deletion in similar contexts (hoog > hoë).At the pronominal level, we find the long possessive forms myn and syn (Std.Af.my and sy from AWS 1921).At the level of verbal morphology, mostly during the Dutch period, we find evidence of irregular assignment of the 3 rd person singular ending -t to atematic verbs(gaan, staan, slaan, sien, doen).That morpheme may be used indiscriminately in its historically correct function (e.g.hy gaat "he goes", from 1991 AWS onwards only hy gaan), as a plural person marker (e.g.hulle gaat "they go", from 1991 AWS onwards only hulle gaan), or as a past participle suffix (e.g.hulle het gegaat "they went", Std.Af.hulle het gegaan). 12ere are some instances of verbs indicating movement or change of state which, instead of taking het as a conjugated tense auxiliary, take the non-concord form, reminiscent of the Dutch use of zijn.13Thecorpus contains a number of un-Dutch morphological features which have never been recorded in the AWS or other prescriptive grammars.Among these features are unexpected plural morphemes, such as -s in daghs ("days", Std.Dutch dagen, Std.Af. da(g)e), and the juxtaposed plural endings e + s in plekkes ("places", Std.Dutch/Af.plekke(n).spesiaal ietse ("the special matters", Std.Af. would here expect spesiale).As regards pronominal forms, there are a few occurrences of the colloquial Dutch zulle (appearing in only one Roman text from 1914), while hulle is firmly established in both subject and object positions.
).The absent features include among other things gender distinction, instead of which is found the generalized Afrikaans article die or demonstrative deze (cf.section 3.2).Neither is the preterite tense form represented, except where it is nowadays allowed in Std.Af.(cf.section3.5).Other features found in the CCDC occur only exceptionally, such as the verbal infinitive ending -en (only four occurrences), and, predominantly in the early Roman texts, verbal markers of person question in the eyes of Cape Malay writers.A state of linguistic insulation could be cultivated to some extent, as still illustrated by today's linguistic tendencies among the Cape Malay community.21Thatthis state of insulation should not be overstated is suggested by the visible concern on the part of Cape Malay writers for identification with Std.
agreement (cf.section 3.4).Conversely, some features shared by both corpora are more represented in our corpus.Among these are features which have become part of Std.Af., such as the subject/object pronoun hulle, dative object marking vir (cf.section 3.3), and the use of been a secondary