Atlas of Pidgin and Creole Language Structures Online

1. Introduction ⇫ ¶

Sri Lankan Malay (SLM) is spoken on the Indian Ocean island of Sri Lanka, known as Ceylon until 1972. It is also spoken overseas, by individuals, in families, and within ethnic social networks in émigré communities, primarily in the English-speaking world (Australia, Canada, Britain, and the United States), as well as in the Gulf States, where Sri Lankans have pursued employment opportunities. There are no statistics on the number of speakers living outside Sri Lanka, and the statistics that are available for Sri Lanka are ethnic statistics rather than linguistic ones. Currently there are approximately 40,000 ethnic Malays in Sri Lanka; however, since not all ethnic Malays are fluent speakers of SLM and not all fluent SLM speakers categorize themselves as “Malay”, the actual number of speakers is not known. There are fluent speakers, mostly Muslims, from other ethnic backgrounds. In Kirinda, in the southeastern Hambantota district, Sri Lankan Malay speakers form an absolute majority of the population, as well as a large proportion of the population in the much larger town of Hambantota, after which the district is named. The largest number of speakers is found in the greater Colombo area, in which traditional patterns of vernacular transmission have shifted since language policies favouring the use of Sinhala were introduced in the nineteen fifties. As a result of those policies, the generation of Sri Lankan Malay speakers under fifty is less fluent in the language than previous generations, due to the loss of sociolinguistic domains and (in many families) a shift to Sinhala. There has been less shift in smaller Sinhala-majority towns such as Kandy, Kurunegala, other highland areas, and in a small number of Colombo-area enclaves.

2. Sociohistorical background ⇫ ¶

Malay speakers, many or most of whom also spoke other Austronesian languages, were brought from Indonesia by the Dutch beginning in the mid-seventeenth century, although there would later be subsequent migration from Malaysia as well. Historically, the most significant second language for native speakers appears to have been Shonam (Sri Lankan Muslim Tamil), based in part on the existence and content of bilingual texts published in modified Arabic orthography in the nineteenth century (Hussainmiya & Slomanson 2013+). Additional support for this socio-historical background comes from communal marriage records demonstrating a strikingly high rate of intermarriage between Malays and (Shonam-speaking) Moors. Further documentary evidence, part of which is discussed in Hussainmiya (1987, 1990, 2008), and in Hussainmiya & Slomanson (2013+) points to very close cultural, communal, and familial ties between Shonam-speaking and SLM-speaking communities. The degree of interaction with Sinhala communities and individuals was less consistent and there is less evidence for network multiplexity in Malay-Sinhala contacts. The maintenance of ethno-religious boundaries has historically permitted very small linguistic communities to persist in South Asian societies, even in urban areas. The language spoken by the co-territorial majority was not necessarily a grammatical model language for the minority contact language in societies organized in this way. This dynamic only began to break down in Sri Lanka in the twentieth century, and particularly in the post-colonial period. This displacement of the minority language by the majority language (SLM and Shonam by Sinhala) has been quite uneven across municipalities, regions, and social classes. While classical Malay was cultivated by some individuals, particularly in the nineteenth century, there is documentary evidence that Shonam continued to enjoy high functional significance in Malay religious life generally. This evidence includes Shonam religious literature published by and for Malays. This cultural symbiosis only began to break down in the twentieth century, particularly in the late British colonial period, in which urban Malays often successfully adapted to the requirements of English-medium and external aspects of colonial culture.

Although the absolute number of speakers is small relative to the total population of the country, Sri Lankan Malay is spoken in a large number of communities in the cities, suburbs, and small towns of the Sinhala-majority part of the country. There are also two small communities on the outskirts of Trincomalee on the east coast (Romola Rassool, p.c.), historically a Tamil-majority area, as well as isolated speakers in the nearby Muslim island community of Kinniyai. The largest absolute number of speakers is found in contiguous and heavily Malay communities in the northern periphery of Colombo, including Wattala-Hendala, Akbar Town, and Hunupitiya, in a demographically declining Malay community in the Slave Island neighbourhood in Colombo, and by many families in Maradana and other Colombo neighbourhoods with substantial historic Muslim populations. There is some functional attrition and loss of lexical material as well as language shift among the younger generations in these areas, a matter still in need of systematic investigation. In the highland communities, although there is less language shift among the younger generation, the functional range and fluency of the middle and older generations is not being reproduced in a sufficient number of younger speakers to insure language maintenance in future generations.¹

The communities in Hambantota district include the town of Hambantota, the Malay settlements at Bolana and Badagiriya, and the village of Kirinda. Kirinda, a densely-populated rural municipality, is the only SLM-speaking community in which all resident individuals are SLM-dominant, where all young people residing in the community use SLM preferentially, where outsiders tend to acquire SLM and begin to adopt local linguistic norms over time, and where there is no evidence of functional decline or shift. To the extent that this community continues to exist, there is no threat to the survival of the language; however, the status of the language in other communities ranges from somewhat less stable to clearly endangered. The Kirinda community, founded as a Malay settlement in the early nineteenth century, is also one of the few communities in which the historic SLM-Shonam diglossia has remained uninterrupted in recent decades, with all formal religious functions performed in Shonam. This is in spite of the fact that the population of the village is overwhelmingly Malay and there are virtually no resident speakers of Shonam as a first language. In Kirinda, as a traditional community, the language of primary and secondary education is Tamil rather than Sinhala.² The teachers include both native SLM speakers, and native Shonam speakers who reside elsewhere. There is also provision for the secondary teaching of Sinhala and English, in order of importance. (Tamil/Shonam was also the medium of education in the village prior to Sri Lankan independence.) Although the entire population of late primary school age and older is trilingual, children remain monolingual SLM speakers until they begin school. While SLM is ordinarily an unwritten vernacular, an increasing number of younger Kirinda Malays with access to text-messaging and electronic networking do not hesitate to write the language in Latin script, using improvised orthography that sometimes reflects the influence of Tamil (i.e. by using a single grapheme for [t, d], [p, b], and [k, g] respectively, as in Tamil, in which the voicing of stops is conditioned by syllabic context).

Whereas in Kirinda, Sri Lankan Malay is vigorously maintained by dense and multiplex peer networks, its vitality in other communities is more strongly dependent on intergenerational transmission and maintenance in the home. In urban communities, SLM continues to be transmitted frequently in bilingual families in which one of the parents is a native Shonam speaker, although there is as yet no research on the proportion of the modern SLM-speaking population coming from families of this type. The external peer group for children from such families is increasingly likely to be a Sinhala-speaking one, as ethnoreligious boundaries weaken, and as the traditional disassociation of the Sinhala language and Islam begins to weaken. (The first Sinhala translation of the Qur'an did not appear until 1986.) TheThe monolingual Sinhala peer culture of the younger generation of urban Sri Lankans is in the process of replacing the communal linguistic cultures of SLM speakers and Shonam speakers, although the process is still far from complete.

3. Phonology ⇫ ¶

The vowel inventory is shown in Table 1. Close front vowels are lax in closed syllables and close-mid front vowels are frequently lax in open syllables. Doubling of stops and of vowels in the orthography employed in this article represents length rather than segmental information.

The consonants are shown in Table 2. With rare lexical exceptions, the dental/post-alveolar difference is not phonologically contrastive.

4. Noun phrase ⇫ ¶

4.1 Pronouns ⇫ ¶

Sri Lankan Malay does not distinguish between biological genders in its pronominal system, and the language has no grammatical gender. Null subjects occur where a pronoun might be expected (i.e. in complement clauses and adjuncts whose subject shares a referent with the matrix subject), and in matrix clauses in which the subject referent is retrievable from the discourse or external context.

4.2 Case ⇫ ¶

Case markers cliticize to the right edge of noun phrases, including pronouns. The distribution of the indefinite determiner attu, variably pre-nominal and post-nominal, provides evidence that the case markers are not simply nominal suffixes. Attu frequently intervenes between noun and case marker, as in buk=attu=ring [book=det=abl] ‘from a book’. Depending on the properties of individual matrix verbs, complement clauses are also case-marked under certain conditions. Case contrasts include accusative, dative/allative, and ablative (which has some instrumental functions). Unlike the dative/allative and ablative case markers, the accusative case marker (§4.2.1) has not developed from a postposition, but its distribution (as a noun phrase clitic, rather than a nominal suffix) does not differ from the case markers derived from postpositions.

I do not treat possessive forms with =pe (see Table 3) as genitive case, since although morphological case is a contact linguistic development in SLM, the possessive clitic and its distribution are a straightforward retention from the isolating Malay varieties that were brought to Sri Lanka from Southeast Asia.

4.2.1 Accusative case

Accusative case marking is present across Sri Lankan Malay dialects and speech communities. Accusative case markers are obligatory in definite patient noun phrases, regardless of animacy. They may also be used in indefinite contexts where they indicate specificity. There are a number of semantically identical variant forms, including nyang, nya, yang, and ya.

Slomanson (2006) proposed an etymology for these forms in which all the variants are derived from the complementizer nyang. In that analysis, the accusative case marker was introduced as a generalization of a reanalyzed accusative-marked WH-element, apa (Malay ‘what’). The string apa yang in (2) was reanalyzed as the WH-constituent, deriving originally from apa followed by complementizer yang, as we find in lexifier varieties in Southeast Asia. The lexifier construction is a WH-cleft (see (2), from Malay). The reanalyzed status of yang was generalized to include accusative case-marking of non-interrogative noun phrases (3).

Example (3) shows apa-yang in a question:

SLM yang is ungrammatical with WH-subjects in sentences.

Tense-bearing sentential complements are marked with accusative =yang or a variant such as =nya.

4.2.2 Dative/allative case

The Sri Lankan Malay dative/allative marker =nang (variant: =na) is derived from the homophonous allative preposition in Javanese (Slomanson 2005, 2006). As a case marker, it is used with the dative arguments for which individual verbs subcategorize. It also signals the thematic status of nominal arguments as recipient, beneficiary, and experiencer NPs.

4.2.3 Ablative case

The ablative case marker =dəri (variant forms =dəring and =ring) is derived from the Malay preposition dari ‘from’. It is also used instrumentally, in contexts such as ‘with a pen’, which is a functional-semantic extension, given the straightforward etymology of the case marker. This is the case clitic used in all SLM varieties to express the meaning conveyed by ‘in’ in the expression ‘in Malay’, or ‘in’ any other language.

4.3 Relative clauses ⇫ ¶

Relative clauses obligatorily precede their nominal heads (cf. 7).

A relativizer homophonous with any of the variant phonological shapes associated with the post-nominal accusative case morpheme (=nyang) sometimes appears as a prefix on the verbal predicate in a relative clause. This is highly variable, and many speakers do not have it at all. The relativizer is entirely absent from the Hambantota district varieties.

4.4 Adjectives ⇫ ¶

Attributive adjectives precede nouns within noun phrases. Non-adjectival modifiers in compounds also precede the head noun; however, this is subject to dialect variation. There are lexicalized constructions in which an attributive adjective follows a noun, but younger speakers may reverse the order in these as well. Whereas the conservative order found in Colombo is a retention of the Malay lexifier order, the orders are invariably reversed in Kirinda. Table 4 exemplifies terms for types of individuals, with a consistent contrast between the items as listed in Bichsel-Stettler’s (1989) Colombo-area glossary, and contemporary Kirinda usage.

4.5 Categorial flexibility ⇫ ¶

Predicate adjectives are productively tense-marked and negated by most speakers, although for many older speakers, this is restricted to a relatively small class of items.

Speakers for whom this is not completely productive use a generic auxiliary to support non-present tense morphology to indicate change of state, as in kosong si-aɖa [empty PST-AUX] for ‘emptied’, and kosong si-jaaɖi [empty PST-become] with the same meaning.

5. Verb phrase ⇫ ¶

5.1 Linear order, tense and aspect

Tense and aspect morphemes are obligatorily adjacent to associated lexical verbs, with or without phonological reduction. Tense morphemes appear as prefixes (10), and the sequence of tense morpheme and verb cannot be interrupted by adverbs or other material.

5.2 Tense morphology

Tense morphemes are restricted to affirmative verbs (and adjectives) in which there is no modal prefix. For affirmative verbs, there is an explicit three-way tense contrast (past tense su-/si-, present tense a(rə)-, and future tense a(n)ti-), although the present-tense form of the verb may be used to refer to the future, if the future reference is recoverable from context or a temporal adverb. Past tense markers demonstrate considerable dialectal variation in phonological shape, although most forms are variants of su- and nyang-. Although tense markers obligatorily precede most finite verbs, on which they are phonologically dependent, high-frequency stative verbs (for example tau ‘know’) may remain unmarked. The pre-verbal distribution of functional morphemes in the verbal domain is characteristic of Malay varieties generally, in which adverbial particles cognate with SLM functional elements mark aspectual contrasts. Unlike the lexifier Malay varieties and unlike canonical creoles, SLM does not permit the stacking of functional elements at the left edge of a verb. In casual prosody, these elements are never accented, and the vowel is typically reduced, so that in the case of su, the vowel nucleus is often reduced to schwa or deleted. Paratactic sequences of events require repetition of the prefix on each non-matrix verb, except in verb-verb compounds and serial verbs, even where this is redundant. In this respect, the pre-verbal functional elements in SLM behave as straightforward inflections.

5.3 Perfect aspect: abbis

The distribution of full and phonologically reduced variants of the Malay perfect marker abbis (asa, as, ai, e) depends on whether the element is expressed in a finite clause. In a non-finite (participial adjunct) clause, abbis- precedes the verb (cf. 11), and in a finite clause, -abbis follows the verb (cf. 12, where the predicate has a past-tense marker).

A finite perfect construction can also be formed by abbis-V plus the auxiliary aɖa:

5.4 Progressive aspect: ambe(l)

The progressive post-verbal element ambe(l) occurs in the progressive construction (cf. 14), in which the highest ambe clause adjoins to the auxiliary aɖa. In the perfect progressive construction (cf. 15), although abbis appears in pre-verbal position, ambe, also an aspectual element, does not.

The periphrastic perfect and progressive aspectual constructions are summarized in Table 5.

Temporal conjunctions attach to lexical verbs in SLM in general, with kapang (‘when’) appearing in pre-verbal position where we would ordinarily find a tense marker. Ambe(l), based on the right-branching aspectually progressive subordinator sambil (‘while’) in earlier grammars, now occurs post-verbally. This is because it reflects aspect rather than tense. Only elements bearing a semantic relationship to tense (a tense feature) are grammatical in pre-verbal position in finite SLM clauses (Slomanson 2011+). It is a subordinator rather than an auxiliary or an ordinary aspectual suffix. Other subordinators lacking a semantic relationship to tense, such as kulung (‘if’), also appear post-verbally.

5.5 Finiteness

Just as Sri Lankan Malay has pre-verbal tense marking, it also has pre-verbal infinitival marking (mə-). This is obligatory on lexical verbs in the complement clause of a matrix predicate (cf. 16). The post-verbal dative/allative =na(ng) clitic is not obligatory, and is subject to considerable variation in rapid speech.

In (16), a free-standing predicate modal of ability takes a complement whose infinitival status is marked by the mə- prefix and optional dative/allative clitic =nang. As the matrix predicate, the modal is independently negated, and comma intonation can easily separate tərə-boole from the rest of the sentence, a clue to its status as a phonological word. The phonological shape of the negation element indicates its finite status. The categorial status of the modal is ambiguous, because many SLM adjectives can be tense-marked.

The morphological finiteness contrast in the phonological shape of negation elements is also encoded in syntax, as we will see. Finiteness does not make a consistent semantic contribution to the sentence. It does however perform a discourse-pragmatic function, permitting the reordering of clauses in sentences representing sequences of events, in order to focus a participial adjunct clause that is not the most recent event in a sequence (Slomanson 2011+). In common with infinitival complements, participial adjunct clauses are not finite. Both the temporal stacking of verbs and the finiteness contrast are demonstrated in the SLM example in (17), in which the aspect-marked non-finite forms abbis-pi and abbis-blajar precede and contrast with the tense-marked finite form su-tulis. The formal marking of finiteness permits a temporally non-primary clause such as mulbar abbis-blajar to be focused, by dislocation to the right edge of the sentence (cf. 18), while retaining its status as a preceding event, relative to the matrix verb.

5.6 Negation

5.6.1 Negation and finiteness

Negation is affected by the finiteness contrast, yielding a system which is not identical to the negation in vehicular Malay or Shonam, but which is nevertheless clear linguistic evidence for specifically Shonam influence, since pre-verbal negation rigidly blocks tense morphology in negated contexts. Its morphosyntactic instantiation in SLM however – pre-verbal and affixal – differs from what we find in Shonam. The negation element, tər(ə)-, modifies main verbs that are interpreted as having tense, although there can be no tense morpheme when the negation morpheme is present (cf. 19). Tər(ə)- is also used to negate adjectives, including modal elements with predicate adjective distribution (cf. 20).

Non-finite verbs, including imperatives, verbs in non-finite nominalized clauses and verbs in participial adjunct clauses, are negated with jang-, a functional extension of the negative imperative marker in the Malay lexifier varieties. There is also a post-verbal negative imperative marker tussa. Bukang marks constituent negation.

5.6.2 Temporal contrasts in finite negation morphology

The functional contrast between the finite negation element tərə- (and its variants) and the non-finite negation element jang- (and its variants) obscures the three-way tense contrast in SLM, a replication of the constraint blocking the co-occurrence of tense and negation markers in Shonam. The co-occurence of negation and past tense form of verbs, for example, is quite odd for Shonam and Tamil speakers, although this is completely productive in Sinhala. The constraint is partly circumvented with the finite marker tuma- (and its variants) marking non-past negation on lexical verbs, and with the finite marker tərə-marking past negation. These can be treated as portmanteau forms, and are not blocked by the constraint, which applies to the stacking of functional morphemes.

5.7 Modality

A syntactic subclass of modals in SLM appear obligatorily to the immediate left of the lexical verb (henceforth abbreviated MV, based on the linear order modal-verb), just as tense, negation, and infinitival morphology do, although not simultaneously.

We can see in the examples in (21) through (25) that the surface distribution of the modal in the MV construction (21) is identical to that of tense morphemes (22) and negation elements (23). The negation element tər(ə)-, a portmanteau form, nevertheless indicates that the verb to its immediate right is finite.

The modal of obligation məsti, which can be phonologically reduced to məsə or məs, exhibits the same syntactic behavior as bər-, the preverbal phonologically dependent MV variant of boole. The modal məsti (and its variants) cannot occur post-verbally nor as a free-standing predicate. These modal prefixes can only occur with verbs that would have a tense prefix if the modal were not there. Consequently, they are associated with finiteness, just as the negation marker tərə- is.

Finite matrix verbs preceded by a bound functional element such as a modal or a negation marker take infinitival rather than bare verbal complements. We see this in sentences such as Miflal [mə-pi=na] bər-liiyat [Miflal INF-go=dat/all can-try] ‘Miflal can try to go’. The bracketed infinitival complement can also appear to the right of the matrix verb. It is also possible for a bare verb to appear to the right of a matrix MV form such as bər-liiyat. In that case, the verb may be part of a verb-verb compound or a serial verb construction.⁵

modal of ability (phonologically reduced)

past tense

negation (tense contrast unmarked)

modal of obligation (M-V order only)

negation and modality

Modals other than məsti can also occur in constructions in which the modal has the same distribution as a predicate adjective, and like predicate adjectives, can be negated and tense-marked, although not simultaneously (cf. 26). For some speakers, these can appear to the left of a lexical verb (cf. 27), to which they are however not bound (neg-mod v). If these bimorphemic modals (prs-mod or neg-mod) were to cliticize to the left edge of a lexical verb, this would violate the constraint blocking the stacking of functional morphemes at the left edge of a lexical verb.

modal of obligation (predicate adjective)

negation and modality

One frequently-occurring modal predicate, kəmauan, is a noun that can be negated in this syntactic context.

6. Sentences ⇫ ¶

6.1. Simple sentences

Word order is somewhat flexible, with strong preference for object-verb ordering, but frequent verb-object orders, and highly frequent use of null subjects. The extent to which VO order marks object focus is variable depending on speaker, discourse context, and prosody.

6.2. Complex sentences

The quotative marker kata functions in practice as a declarative complementizer, permitting recursive embedding of clauses. Relative clauses (§4.3) precede their heads, and can consist of a simple clause and a gap (i.e. [learned Malay girl] meaning ‘the girl who learned Malay’). Conjunctive participles of the form abbis-verb (‘having learned Malay’) can be stacked, as can adjunct clauses conveying simultaneity with ambe (‘while’). Such adjunct clauses can be used as main clauses by adjoining to an auxiliary such as aɖa or duuduk (variant: ɖuuɖung).

7. Other features ⇫ ¶

7.1 The benefactive construction

In the benefactive construction, consisting of V + kasi (lit. ‘give’), which is productive, the activity performed by the lexical verb is performed for the benefit of someone other than the agent. The beneficiary is frequently implied rather than stated, as in the latter (main) clause in (29). The beneficiary, when it is expressed, is in the dative.

7.2 Passive voice

Passive voice is marked post-verbally with kinna, which is bound to a verb with tense morphology (cf. 30). This is productive across verbs and contexts in Kirinda, but the function and distribution of kinna varies across dialects. In the Kandy-area data and description in Nordhoff (2009), for example, kinna is a tense-marked verb indicating surprise or adversity with respect to the event referred to by a subsequent verb, and Nordhoff claims that a productive passivization strategy is unavailable (2009: 502-503).