Sri Lanka Malay and its Lankan adstrates* Umberto Ansaldo ‘The University of Hong Kong Sri Lanka Malay is the vernacular language of the descendants of the Malay-Javanese diaspora of Sri Lanka, Itis a restructured variety of Malay, which emerged from the prolonged contact between speakers of Malay varieties and speakers of Sinhala and ‘Tamil varieties. The grammar shows a typological shift from the Austronesian to the Lankan type, a shift that can be explained by the typological pressure that the adstrates Sinhala and ‘Tamil - which are highly congru- ent — exercise in the trilingual environment. ‘This paper discusses the prevalent grammatical pat- lerns of Sri Lanka Malay in terms of frequency, typological congruence and trilingual admixture. 1 show that, for a complete understanding of Sri Lanka Malay grammar, we must approach it by taking into fall consideration the typological matrix in which it has developed, which includes a Malay-based lexifier and two adstrates, namely Sinhala and ‘Tamil. ‘This argues against a view of Sri Lanka Malay as the product of a bilingual admixture. Keywords: Sri Lanka Malay, typology, Sinhala, Tamil, Case, Tense, Aspect, frequency, tri 1. Background Sri Lanka Malay (SLM) is the restructured vernacular spoken by the Malays of Sri Lanka: the de- scendants of soldiers, political prisoners and slaves brought to Sri Lanka under Dutch and British colonial rule (in particular during the seventeenth to nineteenth centuries). Note that “Malay” is a misnomer here; the term was used during British rule to classify people who came from Java and other parts of the Indonesian archipelago. Indeed the clearest identification is usually with Java for two reasons: (1) the fact that many soldiers, prisoners and convicts were stationed at the Dutch fort in Batavia (Jakarta) before being deported, and (2) the fact that Java was a generic identification for people originating in the Indonesian archipelago (Haan 1922). The Sri Lanka Malays, like other members of the Malay diaspora such as the Cocos Malays, typically refer to their vernaculars as Java (Ansaldo 2008, 2009a). Se ‘This research was supported by the Volkswagen Stiftung’ initiative for the documentation of endangered languages (DoBes), “The core argument of this paper is already put forward in Ansaldo (2005). 1 would like to thank Christine Jourdan, Alain Kihm and Sebastian Nordhoff for their useful comments on this version,368 Umberto Ansaldo From the late 1800s, census data show that the Malays consistently comprised approximately 0.33 percent of the population of Sri Lanka. The largest group of people assigned a Malay origin came as soldiers from such disparate places as Bali, Java, Riau, Ambon, the Maluku Islands and peninsular Malaysia; it was a common practice for Malay and Indonesian women to accompany the soldiers (Ansaldo 2008, 2009a; Hussainmiya 1987; Schweitzer 1931). Overall, three different communities could be distinguished in the past: (1) a rather sophisticated diaspora of noblemen and their families; (2) a “Malay” garrison, which would become the Ceylon Rifle Regiment under the British (Ricklefs 1974); and (3) a group of convicts of whom little seems to be known. As far as the first two groups are concerned, contacts between them were quite frequent (Ansaldo 2008), due among other reasons to the practice of employing noblemen as officers of the troops, as well as to intermarriages (Hussainmiya 1987, 1990). ‘Ihe Malay diaspora was characterised as a close- knit community during colonial rule; the contacts between the different Indonesian ethnicities and different social extractions were maintained in particular through the ranks of the army as well as through common religious practices (Ricklefs 1974). ‘Though linguistically diverse, the Malay diaspora used a Malay lingua franca as a language of interethnic communications this lan- guage has been identified as a Bazaar Malay variety (Hussainmiya 1987), a type of Batavian Malay (Haan 1922), a variety of Moluccan Malay (Adelaar 1991), or a type of Trade Malay (Paauw 2003; Smith et al. 2004). In the ecology of Sri Lanka, this Pidgin Derived Malay (PDM) variety came into contact with two adstrate! languages: colloquial Sinhala and Lankan ‘Tamil. ‘The former was, and is to this day, the dominant language of the majority of the population, the Sinhalese, who see themselves as the legitimate inhabitants of the island. ‘The latter was the language of the largest minority group, as well as a widespread language of trade throughout Monsoon Asia (Ansaldo 2008, 2009a). ‘The Sri Lanka Malay diaspora, 50,000 strong, is at present distributed among various com- munities located around the island, which vary in socioeconomic and educational status and in their linguistic repertoire and communicative practices. Although SLM appears to comprise perhaps five different varieties (Lim & Ansaldo 2006), and there is also a sense of separate identity for each of the different communities (SLM community p.c. 2003-2006), the speakers nonethe- less identify themselves as the same ethnic group and have been predominantly endogamous. The various communities show different degrees and types of multilingualism, depending on their locality and network. ‘The population is tri/quadrilingual to varying degrees of competence, or at least semi-trilingual in SLM, Sinhala or Tamil and English. Although some SLM communities are geographically quite remote, there is no linguistic isolation from the adstrates (Lim & Ansaldo 2006, 2007). ‘This paper argues that the evolution of SLM can only be properly explained by considering the linguistic interactions that occur within a trilingual ecology, in which Malay, Sinhala and Tamil features come into contact (Ansaldo 2008, 200%a, 2009b; Ansaldo & Nordhoff 2009). The argument is structured as follows: Section 2 discusses previous scholarship and methodology: 1. The term “adstrate” is preferred to the dichotomy implied by the terms “substrate/superstrate” considering that, until independence from the British, Malay, Sinhala and ‘Tamil existed on a par as the languages spoken by colonial subjects (Ansaldo 2008),Sri Lanka Malay and its Lankan adstrates Section 3 offers an analysis of SLM grammar. Section 4 evaluates the data and their significance for theories of SLM genesis. 2. Previous scholarship and methodology ‘The evolution of SLM has been captured in terms of (1) bilingual admixture or intertwining (Bakker 2000), (2) creolisation as well as monosubstratal genesis (Smith & Pauuw 2006; Smith et al. 2004), and (3) convergence under trilingual admixture (Ansaldo 2008). Views (1) and (2) have in common the assumption that a Malay-Tamil Moor bilingual admixture due to inter- marriage was the origin of SLM grammar, based on the claim, particularly strong in (2), that no Sinhala influence is found in SLM. Both claims are seriously questioned by the analysis pre- sented below, in which the presence of Sinhala features is obvious. In addition, claims (1) and (2) are criticised in Ansaldo (2008) for being ahistorical and methodologically weak. In relation to history, Tamil-Malay intermarriage rarely features in historical records or oral history and appears to be neither more nor less common than Sinhala-Malay intermarriage.” ‘This position can be more properly evaluated after the discussion presented below, and I therefore return to previous scholarship in the conclusions (see Section 3). ‘To be sure, it is impossible to know with certainty which Malay variety was used in the for- mative stages of SLM since variation was very high, documentation of contact-Malay varieties is still incomplete, and there is no accurate reconstruction available at this time. What is clear is that a PDM variety was spoken, in the sense of Adelaar and Prentice (1996) and Adelaar (2005). SLM. has been shown to derive from PDM varieties (Adelaar 1991, 2005; Adelaar & Prentice 1996), and it is certain prototypical PDM features that can be most reliably used in discussing the early lingua franca of the Malay diaspora in Sri Lanka (Ansaldo 2008, 2009a). As argued in Ansaldo and Nordhoff (2009), it is also important to appreciate that, while descriptions of colloquial Sinhala are available in the literature (e.g., Gair & Paolillo 1997; Karunatillake 2004), very limited information on Sri Lankan (Muslim) ‘Tamil is available at present. We do not have detailed information of these languages and we must therefore rely on descriptions of contemporary varieties, despite the great diversity of the latter (Ansaldo 2008, 2009a). Fortunately, for the sake of typological comparison, there are descriptions of Indian Tamil available, which offer just as precise — if not more accurate - comparison data as contem- porary Sri Lankan Muslim Tamil; after all, the Muslim Tamil variety the Malays encountered was a 30-year-old version of the language currently spoken on the island and indeed origi- nated in South India. Moreover, the extent and importance of dialectal variation in Sri Lankan ‘Tamil is not known, and a reconstruction of a historic form of Sri Lankan Muslim Tamil is not available at this stage. In this paper, the focus is on the case system of SLM, which is obviously of Lankan origin, and on the Tense and Aspect categories, which emerged partly under the influence of Lankan Tense systems but retain some aspectual categories of Malay origin. ee 2 The Tamil Moors were in particularly close contact with the Malays due to their shared Muslim belief. 369370 Umberto Ansaldo a SLM grammar 34 Case Case systems are common and robust features of South Asian grammars and appear to spread relatively easily in intense contact situations, as shown by the development of case markers in vernacular varieties such as Diu Indo-Portuguese (Cardoso 2009), Sri Lanka Creole Portuguese (Smith 1979) and SLM. Indeed, the most striking feature of the SLM NP isa full set of post-nomi- nal case markers; this must be considered a unique feature among restructured Malay varieties, since Malay languages are typically isolating to mildly agglutinative and do not feature nominal case morphology. We do of course find case systems in Indo-Aryan grammars, to which Sinhala belongs, as well as in Dravidian languages, to which Sri Lankan Tamil belongs. Both languages show agglutinative morphology with fusional tendencies. Note also that, because of more than a millennium of intense contact, Sinhala and Tamil grammars have converged typologically and show substantial similarities (Emeneau 1980; Masica 1976). As the rest of this section shows, SLM has developed agglutinative morphology with incipient fusional tendencies in the nominal system, which indicates a typological shift away from the isolating type, as briefly illustrated in (1) (Ansaldo 2009a; Ansaldo & Nordhoff 2009): (1) ni aanak-nay back buku-yay attu aada This student-par good book-acc one exist “This student has a good book? As we can see in the example, Dative and Accusative are marked by suffixes attached to the noun. Note also the verb-final order, which follows Sinhala and Tamil (or “Lankan’) typology, and not Malay/Indonesian, which is strongly verb-medial. Morpheme suurces Itis typical of PDM varieties (Adelaar & Prentice 1996) to derive the lexicon primarily from ma- terial of generic Malay origin, Table 1 lists the SLM case markers and their probable etymology:? As can be seen, the lexical origin of these grammatical markers is clearly Malay (see Nordhoff 2009 for alternative etymological origins) ‘Table 1. Etymology of case markers in SLM Marker Etymology Dative nang Malay nang ‘towards “Accusative” or definite object marker (D0) “yang Malay -nya [+def marker] Possessive pe PDM pe [from possessive punya] Locative ka Malay directional ke Instrumental/Ablative -ring Jakarta bikin ‘to make’ Comitative le (Malay de < dengan ‘with’) Nominative o 3+ Exact etymologies are difficult to obtain considering the diversity of Malay/Indonesian features involved in Bazaar Malay varietiesSri Lanka Malay and its Lankan adstrates, ‘Table 2. Functions of case in Sinhala, Tamil and SLM* Case Sinhala Tamil SIM Nominative ‘Agent ‘Agent ‘Agent Dative Experiencer Experiencer Experiencer Goal, Goal Goal Beneficiary Beneficiary Beneficiary Possessor Possessor Possessor Accusative Patient Patient Patient Genitive-Locative Location Location Possession ‘Temporary possession _Path/Temporary poss. Location Instrumental-Ablative _Instrument/Source Source Instrument/Source Comitative Association Association/Instrument Association * See Ansaldo (2009a) for details. 3.1.2 Structural features of case in SLM, Sinhala and Tamil As this section shows, the syntactic and semantic features of SLM are predominantly of the Lank- an type. Table 2 illustrates the functions of case in SLM and its adstrates, Sinhala and Tamil. In comparing thematic roles in the three languages, we derive the following observations: 1. Prototypical Agents in SLM, Sinhala and Tamil are unmarked. Experiencers and Goals in SLM, Sinhala and ‘Tamil cover the same functional range: They correspond to Dative case in the adstrates and have lack of Volition or Control as the key semantic feature. 3. Accusative marking is really definite object marking in SLM and shares the feature [+defi- niteness] with Tamil and its optionality with Sinhala (Nordhoff 2009). 4, SLM shares the Instrumental-Ablative syncretism with Sinhala, 5. In SLM, Genitive and Locative receive different marking; SLM therefore has a specialised Possessive marker, an innovation in relation to the adstrates.’ It is clear that the SLM case system is heavily influenced by Sinhala and ‘Tamil grammatical cat- egories; the Sinhala influence is particularly obvious in (4), and the combined influence of both adstrates is argued for below. 3.3. Frequency and function While syntactically, case is realised post-nominally in SLM, congruent with Lankan grammar, semantically it has aspects that can be considered innovative. In what follaws, I focus on the core cases of SLM and the grammatical relations they mark. Arguably the most prominent case in Sinhala and Tamil is Dative, which is highly versatile Grammatically, as Table 2 shows. A particular feature of Dative in these languages is that it codes Experiencers, namely subjects that are not in complete control of the event predicated and there- fore are not prototypical Agents. How this works is illustrated in Figure 1 below, ‘This alignment is clearly present in the grammar of SLM: Dative case is assigned in the er. way as in the adstrates Sinhala and Tamil, as shown in Table 3 and illustrated in examples 2) and (3),372 Umberto Ansaldo NOM-DAT alignment NOM < >DAT Agent [+Control] Experiencer [Control] Psych Vs; modal predication Figure 1. ‘The Control feature Table 3. Assignment of Dative in SLM and adstrates Sinhala and Tamil ‘Theta-role SLM Sinhala ‘Tamil Experiencer v v v Goal/Benefactive v v v Possession v v v (2) go-day Mr: Jalaldeen-yay kutumun I-pat Mr. J-acc see ‘Isee Mr. Jalaldeen’ (3) Sir aanak-pada-yay ruma-nay_ e-luppa teacher child-pi-acc house-par pasr-send “The teacher sent the children home? Note that the conflation of Experiencer, Benefactive, Goal and Possession is not peculiar or unique to South Asian languages, but can be seen as a universal tendency of Dative case marking, as clearly shown by Blake (1994: 145), who describes Dative as the main non-core case used to mark complements. ‘When we turn to the functions of Accusative, we find that, unlike Dative, Accusative is not very versatile. In Sinhala, Accusative is only used with animate objects. Moreover, even with animate nouns, this case is very likely to be optional in Sinhala (Gair & Paolillo 1997). In Tamil, Accusative case is only obligatory for a human direct object. With non-human referents, this case is only used to mark definite objects. In SLM, Animacy appears not to be widely marked, though the pronominal system shows variable case marking for Dative first and second person singular, a manifestation of the Animacy distinctions captured in Silverstein’s (1976) hierarchy: While first- and second-person pronouns are marked with the pa suffix dang, the rest of the paradigm receives the suffix nang: go-dang, lu-dang ‘I, ‘you' vs. dia-nang ‘s/he, kitang-nang ‘we, lorang- nang ‘you-pt and derang-nang ‘they’ (see also Ansaldo 2005, 2008, 2009a; Ansaldo & Lim 20043 Nordhoff 2009 presents a diverging analysis). As in Tamil, Accusative tends to mark definiteness in SLM; however, as in Sinhala, it is quite optional. Moreover, Accusative marking is also used sporadically for what appears to be emphasis, an innovative trait of SLM when compared to its adstrates: (4) ini kendera-yay bapi ‘this chair-acc take.go “Take this chair away? Interestingly, in Colloquial Malay varieties, an emphatic marker -nya is found that closely re- sembles the SLM Accusative marker -yang (or -ya)..Sri Lanka Malay and its Lankan adstrates 37; ‘The contrast between Dative and Accusative in SLM is particularly revealing of the forces behind contact-induced change and can be conceptualised as follows. In both Sinhala and Tamil (the Lankan adstrates), Dative covers identical functions, leading to high type frequency in the contact situation; due to its grammatical versatility, Dative markers are also very prominent in discourse, leading to high token frequency. On the contrary, the functions of Accusative do not overlap completely in the Lankan adstrates, so its type frequency is lower. Moreover, Accusative marking can be optional and is functionally restricted, so that its discourse frequency in the con- tact environment is lower. ‘The picture is thus quite clear; High token and type frequency causes SLM ~ which is based on an ancestral language that lacks nominal morphology ~ to adopt identi- cal Dative functions as in the adstrates. Low type and token frequency, on the other hand, results in the partial adoption of adstrate features, as well as innovation in SLM. Moving our analysis outside the core cases, we can make two more interesting observations, In SLM, we find Instrument/Ablative syncretism, meaning that the Instrumental case marker -ring may indicate Source: (5) market-riy ais-tra’ baru ikkay villi bawa market-abi ice-NEG new fish buy bring ‘Get me some fresh (lit. ‘not frozen, new’) fish from the market? ‘This parallels the functions of Instrument in Sinhala, but not in ‘Tamil, in which Instrument and Ablative are clearly distinct (Ansaldo 2005, 2008; Silva 2003); clearly, in this divergence between Sinhala and Tamil, SLM seems to follow the former. However, this ig not always the case: In the realisation of Possessive, a different outcome is found. As Table 2 shows, SLM has two distinct cases for Locative and Possessive; the SLM possessive case sulix is -pe, a derived form of the Malay punya ‘to possess’ (6) goppe tumman go-yay e-tolak Loss friend I-ace_ past-push “My friend pushed me? Reduced variants of punya are well attested as distinguishing features of contact-Malay varieties (PDM) such as Bazaar Malay, Baba Malay, etc. This Possessive marker is one of the most robust features of Bazaar Malay varieties (Ansaldo 2009a), which leaves Tittle doubt that its retention in SLM must be a manifestation of the Founder Principle invoked by Mufwene (1996), whereby the influence of early settlers and features oftheir language varieties can havea long-lasting influence on the new language. As an archaic feature ofa Malay lingua franca, it is most likely that this fea- {ure was maintained from the original vernaculars of the SLM community and its adaptation led to the development of a new case that distinguishes SLM from its adstratés, 3-44 Evaluation ‘The case system of SLM indicates that overall frequency determines selection in the typological latrix (TM) (Ansaldo 2009a). ‘The general congruence of Sinhala and Tamil leads to the high type frequency of the Lankan grammatical features that are selected, as seen in morphology, Constituent order and case marking.374 Umberto Ansaldo In the case of Dative, Tamil and Sinhala essentially show total congruence in the contact ma- trix of a typical South Asian Dative; moreover, Dative has very high discourse-frequency, as itis obligatory and employed in a wide range of functions. ‘The result can be captured as follows: High type/token frequency features are likely to be replicated in the new grammar, In the case of Accusative, however, there is less typological congruence between Sinhala and Tamil, since the former is more sensitive to Animacy than Definiteness, Therefore type frequency is undermined, Moreover, Accusative appears less prominent in terms of discourse semantics, be- ing more limited in its functional range and optional in most cases. This results in low discourse frequency, as is also revealed by its absence from other analyses of SLM case (Smith et al. 2004: see also Slomanson 2006). In this case, Ue SLM. Accusative emerges as an adaptive innovation, which only partially combines the dominant features of the adstrates and also exhibits a novel function, We can therefore generalise as follows: Low type/token frequency features are likely to be recombined innovatively in the new grammar. Finally, in the non-core cases, we see two interesting patterns emerging. First, we note that, where there is a significant lack of typological congruence between the adstrates, as with the Instru- ment/Ablative case, SLM patterns with the dominant language, Sinhala. This can be attributed to token frequency, confirming the general pattern in which frequent features are more robust in the evolution of grammar (Ansaldo 2009b; Bybee 2006), in this case justified by prestige and numerical majority. However, this is not always the case: A different pattern can be observed in the possessive. Here an early possessive morpheme retained from the ancestral language - one of the most robust PDM features across contact varieties of Malay — leads to the formation of a new case marking, which I have attributed to the Founder Principle. These different outcomes show that typological analysis based on frequency can explain much of what happens in the restructur- ing process, but not necessarily everything. 3.2 Aspect and Tense While Malay varieties do not mark semantic roles through case, they do exhibit 'Tense/Mood/ Aspect (TMA) distinctions; the result is a more complex interaction between Lankan and Malay features than is observed in the case system. In both Sinhala and ‘Tamil, ‘Tense is a more salient semantic domain than Aspect. On the other hand, Malay varieties in general do not mark Tense, but have elaborate aspectual systems. For these reasons, I focus on temporal distinctions in the adstrates rather than aspectual vues; the latter domain is optional and variable. For example, ‘Tamil Aspect is optional and is usually marked through aspectual verbs, rather than grammati- cal markers; therefore, it is not fully grammaticalised (Schiffman 1999). When we look into how ‘Tense is realised in the adstrates, we see that while Sinhala has two tense categories, Past and Non-past, ‘Tamil has three, namely Past, Present and Future. As shown in Table 4, SLM has overSri Lanka Malay and its Lankan adstrates 375 time adopted tense distinctions that follow the Sinhala pattern:‘ also shown in Table 4 is the fact that aspectual distinctions still play a significant role in the marking of events in SLM. Table 4 illustrates the predominantly Malayic origin of Aspectual (and Tense) markers in SLM (Nordhoff 2009). As can be seen, all TMA markers can be etymologically identified in vari- cties of Malay, particularly PDM (see Adelaar 1991 for a detailed discussior Ambonese Malay for the perfective and past markers, Moluccan, Baba and Bazaar Malay for the present/progres- sive, and Bazaar and Moluccan Malay for the negation marker, The only non-evident marker is the Continuous form ambe())dudu, illustrated in (7): () Go market-nay — (e)pi abe ar-dudu_si-jato 1 mnarket-towards (PAsr).go go pRoG-sit PrV-fall “While I was going to the market I fell? ‘The form ambil exists in Indonesian Malay as a pre-verbal particle with identical functions (Slomanson 2006). As Table 5 shows, the resemblance between SLM and other contact Malay varieties is quite strong in terms of their aspectual systems and indicates the retention of ances- tral language features, While SLM has predominantly aspectual marking, it does mark at least Past vs, Non-Past, a fact that implies a closer functional alignment with Sinhala. At the same time, Tense in general isnot obligatory in SLM and Aspect marking is clearly more frequent in discourse. Considering little doubt that Malay aspectual functions are reflected in SLM. Still, the fact that, morphologi- cally, TMA in this creole is more agglutinative than isolating reveals the depth of congruence with the Lankan type in the overall grammar of SLM. Note also that TMA morphology in SLM is pre-verbal, while it is post-verbal in Lankan typology; this yields a certain variation in the Table 4. Morpheme sources for TMA markers in SLM (Ansaldo 2009a) Function Morpheme Etymology Perfective/past ca PDM su(dah), (perfective) Past e si>se>e? Perfect ada (post-verbal) PDM ada ‘to have’ (emphatic)> Progressive/habitual/present/non-past —ar(e)- PDM ada (progressive) Future/irrealis ati-fai- M. nanti ‘soon! Completive bis (post-verbal) 1M. habis finish’ Continuous ambe.(er)dudu.(wattu) (post-verbal) _ambel + duduk ‘tale sit? Negation tada PDM tar/tra ‘no(t)’ (M. tidak ada ‘not have’) —_ 4. ‘There appears to be some variation in SLM varieties here: Ansaldo and Nordhoff (2009) find evidence of a hybrid system in the Kandy variety, where a third tense seems to have emerged. However, the argument in favour of a clear future marker is weak: It is very infrequent and more likely to be a matter of irrealis modality (Ansaldo SNordhoff 2009; Smith & Pauuw 2006: 168-169). % Similar functions are found in Ambonese Malay (though there itis preverbal; van Minde 1997: 191),376 Umberto Ansaldo ‘Table 5. Aspect in PDM varieties (Ansaldo 2009a) Function Baba Malay (BM) Bazaar Malay (BZM) Cocos Malay (CM) Perfective suda suda/habis (su)dah Incompletive belum belum b¢eSlum Recentive baru baru Future mo/nanti nanti nanti Irrealis mo/nanti mahu mau/mu Progressive lagi (lagi) lagifada lagi Past/habitual dulw dulu constituent-order of the SLM VP which, though predominantly V-final, also allows for V-me- dial structures (Slomanson 2006).§ 3.2.1 Discussion ‘We have observed the following in the evolution of the verbal domain in SLM: 1. Sinhala and ‘Tamil do not show overall congruence in Aspect; moreover, there are PDM as- pectual features in the matrix. SLM maintains Malay aspectual features due to (a) low type and token frequency of Lankan aspectual features; this is the result of the lack of typological congruence between the adstrates and the optionality of Aspect marking; and (b) vitality of the ancestral language, which can be seen as another manifestation of the Founder Prin- ciple. 2. Sinhala and Tamil Tense categories have a relative degree of functional congruence and are salient semantic categories in discourse. This results in Tense development in SLM along Lankan lines. 3. TMA isa less tightly organised system than case. In relation to the third point, it must be noted that there is an important systemic difference between Case and TMA. The systemic organisation of case is tightly structured as a declension system around interdependencies of Theta-role assignments (e.g, Haspelmath 2006), while TMA really involves different semantic domains. As aptly observed by Matras (2000, 2003), among oth- ers, in contact-induced change a whole system can be transferred from one language to the other, especially when the system has high semantic/pragmatic visibility. This type of “categorical fu- sion” has been documented in other contact environments, such as discourse particles in Singlish (Ansaldo & Lim 2004; Lim 2007; Lim & Ansaldo 2006). ‘The contrast between the evolution of Case and TMA in SLM also shows that tightly organised systems are likely to be selected in toto in the restructuring process. Without such clear systemic organisation, there is less typological pressure from the Lankan adstrates on the PDM features; this pressure is also diminished by the well-developed aspectual system of PDM varieties that are rich in aspectual categories. 6. ‘This could be either a case of retention of Malayic V-medial order or a result of the fact that spoken Sinhala al- Jows substantial word-order variation in the VP (Silva 2003), which may trigger selection of VO-OV patterns from the matrix. ‘This is a common observation in contact language formation (see Chapter 7), and perhaps a general reflection of the fact that constituent order is not always as fixed as grammatical theories like to portray it.Sri Lanka Malay and its Lankan adstrates 377 3.2.2 VP and NP compared In terms of constituent order, the VP shows less congruence with Lankan grammar than the NP (Slomanson 2006, 2007).’ The only way to explain the contrast between the N and V domains lies in the notion of typological “gang-up” (Ansaldo 2005, 2008): It is the congruence of the two adstrates, combined with the lack of PDM marking of grammatical relations, that leads to an overall restructuring of the NP towards the Lankan type in SLM. Likewise, it is the lack of ad- stratal congruence, together with the presence of PDM aspectual material, that leads to limited restructuring of the VP in the SLM TM. his shows that any plausible explanation of the evolu- tion of SLM must be derived from the observation and analysis of the trilingual situation, in which all the three languages present in the TM are given serious consideration. As we will see, this corroborates the claim that previous accounts that rely on bilingualism alone (Slomanson 2006: 156) and the overwhelming influence of Tamil (Smith & Pauuw 2006) need to be revised (see also Ansaldo 2008). 4 Genesis of SLM and the role of the adstrates In what follows, I retrace and critically evaluate the claims put forward to account for the genesis of SLM. Apart from the work of Slomanson (2006), who recognises the influence of Lankan lan- guages in the evolution of SLM grammar, other accounts have taken the view that SLM resulted from bilingual contact between ‘Tamil Moors and Malays. ‘Ihis view was born out of historical speculations found in the work of Hussainmiya (1987, 1990), as well as the idea that SLM might be the product of language intertwining (Bakker 2003); it has led to a problematic methodologi- cal approach in which Sinhala data are discounted from the analysis (Smith & Pauuw 2006; Smith et al, 2004). ‘Ihis state of affairs has been referred to as the “Tamil bias’; it is summarised below (Ansaldo 2008, 20092): 1. Hussainmiya (1987), the first extensive account of SLM history, notes that the Dutch Thom- bas (records of marriage kept under the Dutch) list some mixed marriages, including Malay- ‘Tamil Moor, Ambonese-Malabarese and Malay-Sinhalese; he also suggests that SLM may be influenced by Sinhala, Tamil or both.* 2. Hussainmiya (1990) notes frequent episodes of religious and cultural exchange between Ma- lays and Moors that could have taken place in the mosque because of their shared Muslim faith. 3. Based on the above facts, Bakker (2000) argues for rapid typological convergence and admix- ture with Sinhala and ‘Tamil as a recent development; he assumes that “creolisation” may ac- count for the earlier stages of SLM’s history and postulates an evolutionary path from a Ma- lay-Tamil jargon to contemporary SLM. ‘This argument rests on the assumption that SLM’s genesis may have followed the same route as that of Sri Lankan Creole Portuguese (SLCP). _——— EE 2. This is corroborated by the generative analysis of the SLM VP (Slomanson 2006). 8. Hussainmiya humbly adds that his “exer ig not undertaken with any formal training in linguistic tech- Miques...” (1987: 157-158).378 Umberto Ansaldo 4. Based on (2) and (3), Smith (2003) investigates Accusative urarkers in SLM in order to ‘prove the substantial influence of Tamil. He concludes that one can only argue for a lack of Sinhala evidence, and not for a positive Tamil influence. 5. Smith et al. (2004) present a slightly revised position, claiming a general influence of ‘Tamil and a general Lankan influence in the evolution of SLM. 6. Based on (3), Heine and Kuteva (2005) simply present SLM as a Tamil-Malay creole. Smith and Pauuw (2006) argue for a monosubstratal (i, Tamil) genesis of SLM. 8. Slomanson (2006) argues for convergence between SLM, Sinhala and Tamil (though Malay- ‘Tamil bilingualism is invoked to explain some aspects of tense marking). x In order to assess the original historical claims found in Hussainmiya’s work, Ansaldo (2008) investigates the same marriage records referred to in the historical sources. The results show that many of the records for the period up to 1796 were damaged by water, making significant parts ofthe entries impossible to assess. While the most revealing information for identification here is the parties’ signatures, there is hardly any information on ethnic groups, which makes it difficult to identify Malays/Indonesians and Moors, considering that both groups share the practice of adopting Arabic names. In a particularly interesting section in the Thombas dedicated to mixed marriages (Hussainmiya 1987), only five of 238 entries clearly refer to individuals of Javanese origin: Of these, two records refer tw a Javanese-Moor marriage, one to a Javanese-Javanese mar- riage, and the remaining two are unclear. From 1796 until 1919, mixed marriages were still very rare, with a majority involving Western parties, not Moors; there are two clear entries involving Malays, one married to a Eurasian (between 1867 and 1897), and one to a Burgher (1885-1 897). From 1897 onwards, race is clearly specified; of 196 entries, only one is Malay. Clearly, this evi- dence is not adequate to support a claim of bilingual admixture or of privileged contacts with the Moors on the part of the Malays. On the contrary, it indicates a very low degree of intermarriage, a fact that is fully corroborated by the oral history collected from 50 Malay families of Sri Lanka (Ansaldo 2008), from which it clearly emerges that endogamy was and still is widely practised in the Malay diaspora. To be fair, in steps (1) and (2) there was only a weak suggestion of a possible ‘Tamil influ- ence in the evolution of SLM; however, subsequent linguistic literature has stubbornly developed this claim and turned it into a truism, notwithstanding the fact that direct attempts at proving it have basically failed. The parallel with the Portuguese Creole of Batticaloa expressed in (3) is likewise historically inaccurate. SLCP is the variety of Indo-Portuguese developed in Batticaloa, a traditionally Tamil-inhabited territory on the Eastern coast of Sri Lanka, This variety, like SLM, displays a strong influence of Sinhala/Sri Lankan Tamil typology, but may well show a stronger influence of ‘Tamil than of Sinhala.? However, it cannot be assumed that what may hold true of the genesis of SLCP can be extended to the history of SLM, as the environments in which the two varieties evolved are markedly different (in fact, no two ecologies of contact are ever the same; Mufwene 2001). SLCP may have developed in a small, focused community in contact with pre- dominantly Tamil-speaking communities, where predominantly Portuguese males would have 9. Though Bakker does not present any evidence in favour of this latter point, a historical explanation might be found if we consider that the Batticaloa area has typically been Tamil-dominated.Sri Lanka Malay and its Lankan adstrates come into contact with Tamil speakers, though Sinhala influence has not been ruled out. SLM, on the other hand, evolved in Malay-speaking communities, including women, in more diffuse settings in Sinhala- and Tamil-speaking areas of Sri Lanka. ‘Therefore, while SLCP may be a re- sult of Portuguese-Tamil admixture through intermarriage, in the case of SLM, intermarriage and admixture with specifically Tamil speakers is neither attested nor historically plausible. It is crucial to consider that, although intermarriage was a common, quasi-institutionalised prac- lice in Portuguese domains overseas (Bethencourt 2005; Chapter 7 this volume), it was clearly discouraged in SLM communities, and especially with Tamil Moors, who had low social status. Postulating development from an early jargon is likewise historically unfounded and theoreti, «ally problematic: Why would the communities of Malays, who were socially fully functional as soldiers, guards, planters and fishermen, speak a “jargon”? ‘The most serious attempt to date to argue for Tamil as a primary “substrate” is claim (4) which, crucially, failed to find any evidence. ‘The fact that absence of Sinhala influence is found, rather than presence of Tamil influence, is not surprising at alll considering that the focus was on finding proof of Tamil, not Sinhala, influence. The same methodological problem affects a more recent attempt underlying the claim of a Tamil-based TMA system (8), as only Tamil was inves- tigated and Sinhala data are absent. As this paper shows, the emergence of case in SLM is largely due to a process of typological congruence of the Lankan adstrates that imposes a new grammatical system in the emerging grammar. ‘This is most vividly visible in the emergence of Dative, where high type and token frequency leads the SLM to develop identical functional properties to Sinhala and Tamil, While typological congruence also causes the emergence of other cases, a lower frequency of types and tokens leads to only partial adoption of Accusative functions. As shown above, there sre clear functional-typological reasons why the SLM Accusative, one of whose functions is to mark defi- niteness, looks the way it does. It is the result of lower type/token frequency and behaves cross. linguistically like other typical Accusative cases. A direct influence of Sinhala is also found in the case system, in the Ablative-Instrumental syncretism. Likewise, I have shown that the aspectual system of SLM is mostly replicated from PDM, a conclusion also reached following a different, Benerative approach by Slomanson (2006). Finally, the analysis of ‘Tense presented above argues for congruence with Sinhala, not Tamil, in SLM’ basic Tense categories, as confirmed for the Kandy variety of SLM by Ansaldo and Nordhoff (2009). Considering the overwhelming linguistic and historical evidence in favour of a trilingual admixture, it seems fair to Suggest that the only way to explain the genesis of SLM is through trilingual admixture and typological pressure from the Lankan adstrates that caused a major restructuring in which a typological shift from Austronesian to Lankan grammar took place. ‘The same thing happened to word order: Sinhala and Tamil are verb-final, while PDM vari- eties are V-medial; in the contact matrix, the following shift occurs: SOV + SOV + SVO = SOV (Ansaldo 2009b). Heavy restructuring followed by typological shift is a documented development in situations where a community faces strong cultural pressure from the outside (Thomason & Kaufmann 1988). This is often the cause of language death, unless the community in question shows strong ethnolinguistic vitality. ‘This picture accounts strikingly well for the past and present situation of the Malays in Sri Lanka: tiny minority, not officially acknowledged, has to negotiate sway between two powerful ethnicities. In fact, the ‘Malays have traditionally been by far one of 379380 Umberto Ansaldo the most multilingual groups of the island, often trilingual if not quadrilingual (speaking Dutch or English as well), a skill that has allowed them to Prosper as middlemen in different historical periods (Ansaldo 2008, 2009a; Lim & Ansaldo 2006, 2007). At the same time, the Malay diaspora has always been culturally prominent and keen to Preserve its unique identity; this can be seen in recent attempts to revitalise SLM through the introduction of (Standard) Malay classes as well as revitalisation attempts initiated by the community. SLM is thus the result of a process of ac- culturation in which a minority group on the one hand accommodates to the linguistic practices of the majority while on the other hand tries to preserve its ancestral identity. List of abbreviations ABL ablative PL plural BM Baba Malay Pev perfective BZM Bazaar Malay POSS possessive CM Cocos Malay PROG progressive par dative SLCP Sri Lankan Creole Portuguese Dom definite object marker SLM Sri Lanka Malay NEG negation TM typological matrix PDM Pidgin Derived Malay TMA Tense/Mood/Aspect References Adelaar; K. A. 1991. Some notes on the origin of Sri Lankan Malay. In Papers in Austronesian Linguistics, no. 1, H, Steinhauer (ed.), 1-22. Canberra: The Australian National University. ‘Adelaar, K. A.2005, Structural diversity in the Malayic subgroup. In The Austronesian Languages of South East Asia sand Madagascar, K. A. Adelaar & N. P. Himmelmann (eds), 202-226. London: Routledge Curzon, Adelaat, K. A. & Prentice, D. J. 1996, Malay: Its history, role and spread. In Atlas of Languages of Intercultural Com munication in the Pacific, Asia and the Americas, 8. A. Wurm, P. Mithlhiuser & D. T. Tryon (eds), 673-693. Berlin: Mouton de Gruyter, Ansaldo, U. 2005. Typological admixture in Sri Lanka Mal: dam. Ansaldo, U, 2008, Revisiting Sti Lanka Malay: Genesis and classification. In A World of Many Voices: Lessons from Documenting Endangered Languages ('Iypological Studies in Language 78), K. D. Harrison, D. §, Rood & A. Dwyer (eds), 13-42. Amsterdam: John Benjamins, Ansaldo, U.2009a. Contact Languages, Ecology and Evolution in Asia, Cambridge: CUP. Ansaldo, U. 2009, Contact language formation in evolutionary terms. In Complex Processes in New Languages [Creole Language Library 35], EO. Aboh & N. Smith (eds). Amsterdam: John Benjamins. Ansaldo, U. 8 Lim, L. 2004. Kirinda Java and the Malay creoles of Sti Lanka, Curacao Creole Conference, Curaga0, Netherlands Antilles. 11-15 August 2004, Ansaldo, U. & Nordhoff, S. 2009. Complexity and the age of languages. In Complex Processes in New Languages {Creole Language Library 35], E. O. Aboh & N. Smith (eds). Amsterdam: John Benjamins Bakker, P2000. Rapid language change: Creolization, intertwining, convergence. In Time Depth in Historical Lin- uistics, C. Renfrew, A. McMahon & I.. Trask (eds), 585-620. Cambridge: McDonald Institute for Archae0- logical Research. of Amster- the case of Kirinda Java. Ms, UniverSri Lanka Malay and its Lankan adstrates Bakker, P 2003. Mixed languages as autonomous systems. In The Mixed Language Debate: Theoretical and Empiri- cal Advances, Y. Matras & P. Bakker (eds), 107-150, Berlin: Mouton de Gruyter. Bethencourt, F. 2005. Low cost empire: Interaction between Portuguese and local societies in Asia. In Rivalry and Conflict: European Traders and Asian Trading Networks in the 16th and 17th Centuries, E. van Veen & L. Blussé (eds), 108-130. Leiden: CNWS Publications. Blake, B. 1994, Case, Cambridge: CUP. Bybee, J. 2006, From usage to grammar: The mind's response to repetition. Language 82: 711-733. Cardoso, H. 2009. Diu Indo-Portuguese. A Grammatical Description. PhD dissertation, University of Amster- dam, Emeneau, M. B. 1980. Language and Linguistic Area. Palo Alto CA: Stanford University Press. Gair, J. & Paolillo, J. 1997. Sinhala. Munich: Lincom, Haan, Bde, 1922, Oud Batavia. Gedenkboek. Batavia: Kolff Haspelmath, M. 2006. Against markedness (and what to replace it with), Journal of Linguistics 42: 25-70, Heine, B. & Kuteva, T. 2005. Language Contact and Grammatical Change. Cambridge: CUP. Hussainmiya, B. A. 1987. Lost Cousins: The Malays of Sri Lanka, Kuala Lumpur: Universiti Kebangsan Mala Hussainmiya, B. A. 1990. Orang Rejimen: The Malays of the Ceylon Rifle Regiment. Kuala Lumpur: Universiti Ke- bangsan Malaysia, Karunatillake, W. S. 2004. An Introduction to Spoken Sinhala, Colombo: Gunasena, Tim, 1. 2007 Mergers and acquisitions: On the ages and origins uf Singapore English particles. World Lnglishes 26: 446-473, Lim, L. & Ansaldo, U. 2006. Keeping Kirinda vital:'The endangerment- empowerment dilemma in the documenta- tion of Sri Lanka Malay. Amsterdam Center for Language and Communication Working Papers 1: 51-66. Lim, L. & Ansaldo, U. 2007. Identity alignment in the multilingual space: The Malays of Sti Lanka. In Linguistic Mentity in Multilingual Postcolonial Spaces, E. A. Anchimbe (ed.), 218-243, Newcastle upon Tyne: Cambridge Scholars Publishing. Masica, C. 1976, Defining a Linguistic Area: South Asia. Chicago IL: University of Chicago Press, Matras, Y. 2000. Fusion and the cognitive basis (or bilingual discourse markers. International Journal of Bilingual- ism 4; 505-528, Matras, Y. 2003. Mixed languages: Re-examining the structural prototype. In The Mixed Language Debate: Theo- retical and Empirical Advances, Y. Matras & P. Bakker (eds), 151-176. Berlin; Mouton de Gruyter. van Minde, D. 1997. Malayu Ambong: Phonology, Morphology, Syntax, Leiden: CNWS. Mufwene, S. S. 1996. The founder principle in creole genesis. Diachronica 13: 83-134, Maufwene, S. 8. 2001. The Ecology of Language Evolution. Cambridge: CUP. Nordhoff,S. 2009. A Grammar of Sri Lanka Malay. PhD dissertation, University of Amsterdam. Paauw, S. 2003. What is Bazaar Malay? ‘The Sth International Symposium on Malay/Indonesian Linguistics, Nijmegen, June 2003. Rickles, M. C. 1974, Jogiakarta under Sultan Mangkubumi, 1749-1779: A History of the Division of Java. London: our, Schiffman, 11. 1999. A Reference Grammar of Spoken Tamil, Cambridge: CUP. Schweitzer, C. 1931. Reise nach Java und Ceylon 1675-1682. The Hague: Nijhoff, + Silva, A. W. L, 2003. Teach Yourself Tamil: A Complete Course for Beginners. Kandana, Sri Lanka: ers, Slomanson, P. 2006. Sri Lankan Malay morphosyntax: Lankan or Malay? In Structure and Variation in Language Contact (Creole Language Library 29], A. Deumert & S. Durtleman (eds), 135-158. Amsterdam: John Benjamins. Slomanson, P. 2007. the perfect construction and complexity drift in Sri Lankan Malay. Lingua 118: 1640-1655. Smith, 1. 1979, Substrata versus universals in the formation of Sri Lanka Portuguese. Papers in Pidgin and Creole Linguistics 24-57: 183-200. ubudu Print- 381382 Umberto Ansaldo ‘Smith, [. 2003, ‘The provenance and timing of substrate influence in Sri Lanka Malay: Definiteness, animacy and ‘number in accusative case marking. The South Asian Language Analysis Roundtable XXIII, Austin, Texas, 10-12 October 2003. Smith, I. & Paauw, §, 2006. Sri Lanka Malay: Creole or convert? In Structure and Variation in Language Contact {Creole Language Library 29], A. Deumert & S. Durrleman (eds), 159-181. Amsterdam: John Benjamins. ‘Smith, 1., Pauuve $, & Hussainmiya, B. A. 2004. Sri Lanka Malay: The state of the art. In Yearbook of South Asian Languages 2004, R. Singh (ed.), 197-215. Berlin: Mouton de Gruyter. ‘Thomason, S. & Kaufman, T. 1988. Language Contact, Creolization and Genetic Linguistics. Berkeley CA: Univer- sity of California Press.