Students of Sri Lanka Malay agree that the language has been heavily influenced by the local languages, Sinhala and Tamil. Differences arise over not only the degree and timing of such influence from each language, but also the extent to which the language developed through untutored second language acquisition (on the part of Tamil &/or Sinhala speakers) &/or intense bilingualism (on the part of Malay speakers). Nordhoff’s arguments for Sinhala influence are examined in the context of Thomason’s (2001) framework for establishing contact-induced change and found to be convincing for some features, but weaker or unconvincing in others. The argument for early Sinhala phonological influence is based on an unsurprising distribution and the mechanism of substrate influence (Siegel, 1998, 2008) which has not been shown to operate in the context of intense bilingualism. The linguistic differing consequences of untutored second language acquisition and intense bilingualism have not been thoroughly investigated, except on lexicon (Thomason and Kaufman, 1988). The Sinhalese component of Sri Lanka Malay lexicon stands at less than 1% (Paauw, 2004), a figure inconsistent with the claim of heavy Sinhala influence through intense bilingualism.
There are two interrelated issues addressed by Nordhoff ’s and my papers: first, in what ways did local languages (Sinhala and Tamil) influence the development of Sri Lanka Malay, and when did this influence occur; second, what were the social circumstances that allowed such influence to take place. I have argued for the early influence of Tamil, Nordhoff for that of Sinhala. I hypothesize adstratal influence involving language shift (Thomason and Kaufman, 1988)/untutored second language acquisition/imperfect learning (Thomason, 2001), i.e. the influence of Tamil speakers’ L1 on their L2 variety of Malay; Nordhoff hypothesizes adstratal influence involving borrowing in language maintenance (Thomason and Kaufman, 1988)/intensive and extensive bilingualism (Nadkarni, 1975), i.e. the influence of Malays’ L2 knowledge of Sinhala on their L1.2 We have both proposed our respective socio-historical scenarios which make these hypotheses possible. Our common ground is that we both accept adstratal Sinhala influence on Sri Lanka Malay, but we differ on the timing of this influence; we both accept Tamil influence on Sri Lanka Malay, but we differ on the mechanism that brought this about (L1Tamil influence on L2Malay or L2Tamil/Sinhala influence on L1Malay). While we argue the details of our respective analyses, two larger issues need to be addressed: first, what counts as evidence for the influence of one language on another; second, do the linguistic consequences of L1 on L2 and L2 on L1 influence differ. In this commentary I will reexamine Nordhoff ’s arguments for Sinhala influence in the light of the first question and his socio-historical scenario in light of the second. I will also comment briefly on Nordhoff ’s argument for the dating of Sinhala influence on Sri Lanka Malay.
2. Establishing Contact-induced Change
For the first question, we can turn to Thomason (2001: 93-4) who sets out the following requirements of a case for contact-induced change:
“look at the language as a whole, not just at one bit of it;”
“identify a source language” that had contact with the recipient language “intimate enough to make structural interference possible;”
identify “shared structural features,” which “need not be identical in all respects”;
“prove that the shared features….were not present in the receiving language before it came into close contact with the source language” (emphasis original);
“prove that the shared features were present in the proposed source language before it came into close contact with the receiving language” (emphasis original).
Thomason further argues that “[f]or the sake of completeness” we should also show that the change in the recipient language was not motivated internally or by universal structural tendencies (2001: 94).
How might we apply Thomason’s requirements in order to demonstrate that one language, rather than another, has influenced a third language? With respect to the first and third requirements, both Nordhoff and I have adopted an approach that focuses on grammatical features that differ in Sinhala and Tamil. This is reasonable since we are trying to identify which of two contact languages influenced the development of which features in Sri Lanka Malay. It is also reasonable given that other parts of “the whole language” are covered in the growing literature on Sri Lanka Malay (e.g. Smith, Paauw and Hussainmiya, 2004; Paauw, 2004; Bakker, 2006; Slomanson, 2006, 2009; Ansaldo, 2008, 2009b; Nordhoff, 2009). One of the strengths of Nordhoff ’s paper is that he treats a wider range of phenomena than I do.
With respect to Thomason’s second requirement (“identify a source language”) Nordhoff and I agree, as noted above, that both Tamil and Sinhala have influenced the development of Sri Lanka Malay, but we differ on the time periods ascribed to the influence of each language. I view (Muslim) Tamil as the dominant early influence on the grounds that a common religion caused greater social contacts between Malays and Moors. I see Sinhala influence playing a role later, particularly after independence when nationalist-inspired policy made it the socially dominant language. Nordhoff, on the other hand, sees Sinhala as having played an important role from the very start.3
In contexts where more than one candidate source language is available Thomason’s fourth and fifth requirements do not suffice to prove influence from a specific language. They need to be augmented by a seventh requirement, namely:
prove that the shared features were not present in the other contact languages.
I will argue below that Nordhoff fails to do this adequately for a number of the features he proposes as Sinhala-inspired.
2.1 Indefiniteness Marking
Nordhoff claims that the marking of indefiniteness is “one of the best examples for exclusive Sinhala influence one can find in Sri Lanka Malay grammar.” I would agree that some of the data are convincing, but that they point to recent Sinhala influence on some speakers. Nordhoff adduces two pieces of evidence: the requirement that indefinite nouns be marked by (h)atu ‘one; a’ and the use of double-marked nouns (with (h)atu both before and after), which is reminiscent of the use of the Sinhala numeral eka ‘one’ before the noun in addition to the definite/indefinite suffix –eka/ekak used on many loans. However, there are issues with some of Nordhoff ’s data. First, the requirement that indefinite nouns be marked with (h)atu does not apply to all speakers. For example in (1) based on Nordhoff ’s (13) (h)atu is obligatory for some speakers and optional for others.
Kumar hatthu prentha.oorang [SLM]
Kumar one lawyer
‘Kumar is a lawyer.’ (Nordhoff, this issue (13))
The following example, taken from Hussainmiya’s Kirinda data4 also shows no indefiniteness marking:
kitam-pe baapa pətaama ara.daataŋ jaaga, soolt dipaatman-ka [SLM: conversational]
1pl-gen father firstly top watchman salt department -loc
‘My father firstly was a watchman in Salt Department.’
Since Tamil, as Nordhoff correctly notes, does not use an article here, the fact that Sri Lanka Malay allows its use (and given the fact that Malay satu ‘one’ is not used as an article) indicates Sinhala influence, but the variability points to a change that it coming in rather than an established one. Secondly, there is a model for the Sri Lanka Malay use of both pre and postnominal (h)atu in Tamil, where a prenominal numeral oru ‘one; a’ and a postnominal numeral onru ‘one; a’ may be used to indicate indefiniteness. Nordhoff cites an authority who indicates that double-marking of nouns is ungrammatical in Tamil (fn. 7), but my investigations indicate double-marking is indeed grammatical in colloquial Tamil: example (3) was presented to five native speakers of Sri Lanka Tamil; all found it commonplace. When the double marking was pointed out, the typical reaction was something like “Well, that’s how we speak; of course you wouldn’t write that.”
en e oru three-tonner onru iru-nd-adu, munnaale [Tamil]
1pl.gen one three-tonner one be.pst.3sg.n, front.
‘There was a three-tonour of ours in front.’ (cf. Nordhoff, this issue, ex. (17))
Thus it has not been established that Tamil could not be the source of this construction in Sri Lanka Malay.
Finally, the evidence for (h)atu being used as a loanword integrator (like Sinhala eka(k)) is not convincing. Loanwords are found without a following (h)atu, even when they are indefinite; e.g.,
Heart attack asà-peegang baapa=le su-niin˘˘g˘gal. [SLM]
heart attack cp-catch father=addit pst-die
‘My father got a heart attack and died as well.’ (Nordhoff, 2009: 177)
In Hussainmiya’s Kirinda data, I find the phrases atu riisan ‘one reason’, atu skuul-na ‘one school-DAT’ atu rejiman ‘a regiment’, atu kæn ‘one can’, atu ɖiiyaaroo ‘a D.R.O.’, all without a following atu. Double marking is found in three examples: one with a loan from English, atu petiʃan atu ‘a petition’; one with a loan from Tamil, atu salluge atu ‘a concession’; one with a native noun, atu tumpat atu ‘a place’. It may be that the double usage is more common with English loans, but this has not been demonstrated.
As additional evidence, Nordhoff cites an example (repeated below as (5)) in which hatu is used along with a deictic. However, my sources indicate that the deictic here is used as a discourse marker, not as a demonstrative applying to the noun; i.e. the translation should be ‘Here, I have to do a mock wedding.’ Thus hatthu is simply an indefinite marker, not a loanword-integrator. Interestingly, the Sinhala equivalent given by Nordhoff uses the definite marker eka; so it can, in fact, mean ‘I have to do this mock wedding,’ but as soon as indefinite ekak is substituted for eka¸ the deictic in Sinhala must also be read as a discourse marker. The Tamil equivalent becomes grammatical when the postnominal form onru is used.
inni mock wedding =hatthu mas- gijja [SLM]
mee mock wedding eka karanɖa oonæ [Sinhala]
*inda mock wedding oru ceyya veeɳum [Tamil]
prox mock wedding=indef must- make must
‘I have to do this mock wedding.’ (Nordhoff, this issue (18))
So atthu in both pre- and post-nominal position appears to be a marker of indefinite (like Tamil oru/onru).
2.2 Dental Onsets
Smith, Paauw and Hussainmiya (2004) note that Vehicular Malay alveolar /d/ is generally reflected in Sri Lanka Malay as retroflex [ɖ] intervocalically and dental [d̪] (occasionally [ɖ] or [ɖ]~[d̪]) initially. They suggested that the distribution can be traced to the Tamil phonotactic constraint against initial retroflexes. (Sinhala has no such constraint.) Nordhoff objects to this suggestion on the grounds that the Tamil constraint against initial voiced stops is not also transferred. The principle invoked here seems to be that phonological influence is an all-or-nothing affair. But such is not the case. Thomason, in connection with her admonition that common structural features “need not be identical in all respects” mentions the fact that under Dravidian influence Indo-Aryan developed retroflexes in initial as well as medial environments, despite the Dravidian constraint against initial retroflexes. (2001: 93). More to the point, the lack of aspirated or breathy-voiced stop series in Dravidian did not prevent these series from surviving in Indo-Aryan. The facts are similar in Sri Lanka Malay: Vehicular Malay had both voiced and voiced stop series, but it had no dental/retroflex contrast. Tamil has a single series of stops with voiceless allophones initially or in geminates and voiced allophones elsewhere; it also has a dental/retroflex contrast, which is neutralized in favour of dental in initial position. It should be possible to claim that Tamil-speaking learners of Malay reinterpreted non-initial alveolar /d/ as retroflex medially but dental initially without assuming that they failed to notice a contrast between voiced and voiceless stops in their target. It is certainly true that untutored (even tutored) learners may fail to master a contrast not found in their L1, but this is a tendency, not a law. Indeed, Tamil learners of Sinhala are able to learn the voiced/voiceless stop contrast in that language.
2.3 The Instrumental Case Marker
Nordhoff points out that in Sri Lanka Malay one of the ways of marking ablative is with the bare instrumental marker and that this option is also available in Sinhala. The other option is with locative marker followed by a grammaticalized past-participle of the locational/existential verb, and this is the only option in Tamil. Thus the case for Sinhala influence here is good. Interestingly, in Hussainmiya’s Kirinda data, which date from the 1970s, although ablative expressions occur several times, the only construction found is locative + participle. This suggests that Sinhala influence is recent, though the possibility remains that Hussainmiya’s data (just under 500 utterances in all) are unrepresentative.
2.4 Number of Stop Series and the Distribution of Stops and Nasals
Nordhoff argues that Sri Lanka Malay has three series of stops (voiceless, voiced and prenasalized) like Sinhala and unlike Tamil, which has a single series. The retention of the voiced/voiceless contrast from Vehicular Malay is not an issue here – there is no change; so Sinhala influence cannot be adduced (cf. Thomason’s fourth requirement, above). The development of a typologically marked prenasalized stop series, however, would make a very good case for Sinhala influence. It is a case that rests on a particular phonological analysis, however, (Tapovanaye, 1995; Nordhoff, 2009) that sets up a contrast between nasal + stop clusters and prenasalized stops and between geminate and single consonants in order to predict vowel length. An alternative analysis (Bichsel-Stettler, 1989; Smith, Paauw and Hussainmiya, 2004) sets up contrastive vowel length and treats long (geminate) consonants and long nasals5 in clusters as subphonemic, though the latter are not specifically mentioned. The merits of these two mirror-image analyses can be debated, but this is not the appropriate place. A more satisfactory analysis would treat vowel length and consonant length as part of the same phenomenon of syllable weight. For now, the relevant distributional facts are that nasals in nasal+stop clusters are shorter following a long vowel and longer following a short vowel. Stated this way, the distribution seems unexceptional rather than typologically marked, and Sinhala influence is not required to explain it (cf. Thomason’s sixth requirement).
As noted above, a singe vs. geminate consonant contrast is a product of a particular phonological analysis, but the difference is subphonemic in another analysis. Again the distributional facts are unremarkable: short consonants follow long vowels and long (geminate) consonants follow short vowels (with some modification for syllable position, etc.). But Nordhoff ’s argument does not go through even if Sri Lanka Malay had a robust contrast between single and geminate consonants. Geminate and single consonants have developed from (phonemically) single consonants in Vehicular Malay. Given the fact that Vehicular Malay had a voicing contrast in the stops, it is unsurprising that Sri Lanka Malay has both voiced and voiceless geminate stops. The voicing contrast is a retention and the fact that Sinhala has a similar contrast but Tamil does not is irrelevant.
Nordhoff argues that the presence in Sri Lanka Malay of two existentials duuduk (< Vehicular Malay duduk ‘sit’), for animate subjects, and aada (from Vehicular Malay ada ‘be’) for, inanimate subjects, is due to Sinhala rather than Tamil influence, since Sinhala also distinguishes between animate and inanimate existentials, while Tamil has no such distinction. Moreover the animate existential in Sinhala innawaa evolved from a word meaning ‘sit’, providing a model for the development in Sri Lanka Malay. The analysis is certainly suggestive, but not without problems. First Sinhala innawaa no longer has the meaning ‘sit’, (expressed by in˘da gannawaa). Unless it can be shown that the earlier meaning was lost after the contact between Sinhala and Sri Lanka Malay, the Sinhala derivation could not have influenced the development in Sri Lanka Malay (cf. Thomason’s fifth requirement). The Tamil existential, iru, however, also has the dual meaning ‘sit; be’, reconstructible to early Tamil (Burrow and Emeneau, 1984: 480). While it is true that iru is used for both animates and inanimates, the dual meaning provided a ready model for the extension of the meaning of duuduk. Moreover, because of its original meaning, duuduk would have been restricted to animate subjects. Thus the continuation of this restriction in Sri Lanka Malay is a retention, and Sinhala’s existing distinction between animate and inanimate existentials is irrelevant. Thus what looks at first glance like a clear case of Sinhala influence turns out after closer analysis to be more likely a case of Tamil influence. Interestingly, the parallel development in Sinhala may also be due to (earlier) Tamil influence.
2.7 Non-nominative Subjects
Nordhoff claims that the distribution of non-nominative subjects in Sri Lanka Malay resembles that found in Sinhala and not that of Tamil. He provides several examples of the parallelism, but the analysis is not detailed and no parallel Tamil examples are given. It is not clear whether the non-nominative subjects are requirements of the verb or other semantic factors. The use of the instrumental, for example, seems to be limited to institutional subjects, such as the police, army or fire brigade. Other subjects take nominative, as seen in (5):
baapa (*-dering) su-daatang [SLM, elicited]
father –instr pst-come
I haven’t been able to replicate an accusative subject for tingalam ‘sink’ in Nordhoff ’s example (27); so this is more likely a case of lexical conditioning that is not shared by all speakers. Overall, I would say there is probably a case to be made for Sinhala influence here, but a more thorough analysis is needed.
2.8 Semantics of the Non-past Form
Nordhoff notes that the present tense in subordinate clauses in both Sinhala and Sri Lanka Malay can have past meaning when the matrix verb is past. He notes particularly that this ‘participial’ use of the present is found only in subordinate and not matrix clauses. We cannot therefore say that the usage derives from the fact that time reference is derived from context in Malay varieties, which lack tense-marking (the Sri Lanka Malay present marker, ara-, is a redeployed Vehicular Malay progressive marker). Nordhoff argues further that the fact that Sinhala present form derives from an earlier non-finite form6 provides a model for the extension of meaning of the Sri Lanka Malay present to a participial usage in subordinate clauses. In order for this last argument to be valid, it would have to be shown that both nominal and tensed verb readings were available in Sinhala during the time it was in contact with Sri Lanka Malay (Thomason’s fifth requirement), which Nordhoff does not do. Although he points to “vestigial participial uses” of the Sinhala present, these are clearly present tense forms in a synchronic analysis. Nor does Nordhoff supply evidence for his claim that Tamil influence is not at play here. Certainly, the Tamil present tense form used in main clauses is not used in the equivalent of Nordhoff ’s example (32). The form used, a tensed verbal noun, does have present marking, however:
avar kaɳ-ɖ-aar [kurangu-gaɭ viɭayaaɖ-r-att]-e [Tamil, elicited]
3sg.hon see-pst-3sg.hon monkey-pl play-prs-nmlz-acc
‘He saw the monkeys playing.’
Thus in this type of subordinate clause, the tense context is inherited from the matrix, as in Sinhala and Sri Lanka Malay. If the past verbal noun is used, the reading is past-before-past, as seen in Nordhoff ’s example (33). Moreover, when we look at a broader selection of subordinate clauses, we find that the Tamil main-clause present can be used with past reference in the ubiquitous quotative construction:
avar [vaa-r-an] eɳɖu con-n-aar [Tamil, elicited]
3sg.hon come-prs-1sg quot say-pst-3sg.hon
‘He said he was coming./He said ‘I’m coming.’’
It is possible that the use of present tense marker in past contexts began in Sri Lanka Malay in quotative constructions and then spread to other types of subordinate clause. I do think there is a case to be made for Sinhala influence here, but the case is not quite as clear-cut as it seems.
2.9 Zero Adclausal Nominalization
Nordhoff points out that Sri Lanka Malay parallels Sinhala in allowing unmarked clausal nominalization, while in Tamil nominalized verbs are overtly marked. He concedes that the ‘seeds’ for this constrution were present in Vehicular Malay in the form of unmarked clausal complements of prepositions, thus it may not be possible to satisfy Thomason’s fourth requirement. Moreover, from a general language contact point of view, structural parallelism may develop without overt marking in the affected language. For example, Sri Lanka Portuguese has developed an unmarked conjunctive participle, despite the fact that in both Tamil and Sinhala the conjunctive participle is overtly marked, as seen in (9)
noos taam jaa.foy jaa.paraa tiña. [Sri Lanka Portuguese, conversational]
naangaɭ =um poo –yi kuuɖ-i iru.nd.oom [Tamil]
api =t gihil-la (in˘da-la) hiʈiyaa [Sinhala]
1pl cnj go.pst–cp stay.pst–cp be.pst(.agr)
‘We also went and stayed [there].’ (Smith, field notes, 1974-5)
Therefore, the fact that Tamil has overtly marked nominalized clauses does not disqualify it as a possible model for the development of unmarked nominalized clauses in Sri Lanka Malay (new seventh requirement).
2.10 Summary of Structural Features
Table 1 is a revision of Nordhoff ’s summary chart in light of the discussion above. I prefer to give a textual description rather than plusses and minuses. I have separated the issue of a voicing contrast in the stops and presented it separately from the development of retroflexion and gemination and the number of stop series. Instrumental functioning as ablative has been added. Nordhoff ’s claim of Sinhala influence is certainly substantiated in a few instances, but overall the case for Sinhala influence is considerably weaker than he claims.
3. L1 on L2 vs. L2 on L1 Influence
Before addressing the dating of Sinhala influence, I would like to return to the question of different mechanisms of adstratal influence subsumed by Thomason and Kaufmann’s opposition between language shift and language maintenance. In language shift, aspects of the learner’s L1 are transferred to their L2 target language during the course of untutored second language learning. In this scenario, features of Tamil would be transferred to Sri Lanka Malay by Tamil speakers, particularly women from the Moor community who married Malay-speaking soldiers. In language maintenance, however, a minority group’s L1 is influenced by their knowledge, and frequent use, of the broader community language, their L2. In this scenario, features of Tamil and Sinhala would be adopted by speakers of Sri Lanka Malay because of their frequent use of these languages. No one disputes the fact that indegenous influence on Sri Lanka Malay has been strong. Therefore, in Thomason and Kaufman’s terminlogoy we are in the range between ‘heavy ..adstratum..interference’ and abrupt creolization in the language-shift scenario and in the range between ‘heavy structural borrowing’ to ‘massive grammatical replacement’ in the language maintenance scenario. Neither Thomason and Kaufman (1988) nor Thomason (2001) distinguish between the structural consequences of ‘heavy adstratum interference’ and ‘heavy structural borrowing’, and this is clearly an area where more research is needed. My own work indicates that L1 on L2 influence lends itself more readily to the production of what I term ‘hijacked’ gramatical elements, whose original grammatical and semantic status in the L2/superstrate/adstrate is very different from the status imposed on them by untutored learners (Smith 2001, 2010). A Sri Lanka Malay example would be the ‘hijacking’ of a relative clause marker to express de-verbal nominalization (Smith 2010). The validity of this hypothesis needs to be tested further. A non-structural difference that Thomason and Kaufman discern between ‘language shift’ and ‘language maintenance’ involves borrowed lexicon. Contact-induced developments in language shift (L1 on L2 influence) typically begin with morphosyntax and phonology rather than with lexicon, while contact-induced developments in language maintenance (L2 on L1 influence) typically begin with lexicon (1988: 50). If knowledge of Sinhala has structurally influenced Sri Lanka Malay speakers’ L1, it is surprising that so little Sinhala vocabulary has been borrowed (Paauw 2004).
In sum, despite the plausibility of the ‘metatypy’ scenario painted by Nordhoff, it is not supported by the linguistic evidence.
4. Dating Sinhala Influence on Sri Lanka Malay
Nordhoff argues on phonological grounds that Sinhala influence on Sri Lanka Malay occurred before the language stabilized, which he takes to be before 1800. The features on which the argument is based are long vs. short nasals before stops and long (geminate) vs. short consonants, particularly geminate voiced stops. According to Nordhoff ’s scenario, sub-phonemic variation in Malay dialects in the syllabification of nasals and in the length of intervocalic consonants attained contrastive status through the mechanism of ‘substrate reinforcement’ (Seigel 1998, 2008). I pointed out above that the distribution of these segments in Sri Lanka Malay must be considered together with that of long and short vowels, and that the cooccurrence of short vowels with long consonants on the one hand and long vowels with short consonants on the other is unmarked and does not require an explanation involving language contact. Moreover a voicing contrast in geminate stops was inherited from a voicing contrast in Vehicular Malay and is thus a retention. The case for Sinhala influence here is extremely weak. But even if the case were stronger, the substrate reinforcement argument contains some imponderables. First, we may note that Siegel (1998, 2008) proposes this mechanism in the context of L1 on L2 influence; can it be shown to also operate in the context of L2 on L1 influence? If the knowledge of Sinhala as a second language had such a strong influence on the phonology of Sri Lanka Malays’ mother tongue, why did it not have an equally strong influence in morphosyntax and lexicon?
Nordhoff ’s contribution to the study of Sri Lanka Malay is significant and valuable. His arguments for Sinhala influence on the language have broken new ground. Despite the fact that many of the proposed cases do not stand up to scrutiny, there is a residue of good evidence for Sinhala influence: indefinite marking of equational complements, the use of the instrumental case as ablative, the distribution of dative and accusative subjects, and possibly the time reference of present tense forms in subordinate clauses. These are not core features of the grammar and could have developed fairly recently. For some of them, variation may also point to a recent development. The case Nordhoff makes for early Sinhala influence rests on shaky phonological grounds and an unproven mechanism.
The linguistic consequences of L1 on L2 influence vs. L2 on L1 influence have not been widely investigated. The evidence of loans, however, does not favour heavy L2 influence of Sinhala on L1 Sri Lanka Malay.
AnsaldoUmberto. 2008. Sri Lanka Malay revisited: Genesis and classification. In David HarrisonK.RoodDavidDwyerArianne (eds.) Lessons from Documented Endangered Languages (Typological Studies in Language 78)13–42. Amsterdam: John Benjamins.
BakkerPeter. 2006. The Sri Lanka Sprachbund: the newcomers Portuguese and Malay. In MatrasYvonMcMahonAprilVincentNigel (eds.) Linguistic Areas – Convergence in Historical and Typological Perspective135–159. Houndmills, Basingstoke; New York: Palgrave Macmillan.
BurrowT[homas]M[urray B.]Emeneau. 1984. A Dravidian Etymological Dicctionary2nd ed. Oxford: Clarendon Press; Chicago: Digital South Asia Library http://dsal.uchicago.edu/dictionaries/burrow/ (Accessed November 11 2011. Numerical references are to entry numbers.)
LadefogedPeterMaddiesonIan. 1996. The Sounds of the World’s Languages. Oxford: Blackwell.
NordhoffSebastian. 2009. A Grammar of Upcountry Sri Lanka Malay. Ph.D. thesis University of Amsterdam. Utrecht: LOT. Downloadable at http://www.lotpublications.nl/publish/issues/Nordhoff/index.html
PaauwScott. 2004. A Historical Analysis of the Lexical Sources of Sri Lanka Malay. M.A. thesis York University [Toronto].
SlomansonPeter. 2006. Sri Lankan Malay morphosyntax: Lankan or Malay? In DeumertAnaDurlemannStephanie (eds.) Structure and Variation in Language Contact (Creole Language Library 29)134–158. Amsterdam: John Benjamins.
SmithIan R. 2001. Creolization and convergence in morphosyntax: Sri Lanka Portuguese and Sourashtra nominal marking typology. The Yearbook of South Asian Languages and Linguistics 2001 [issue edited by Peri Bhaskararao and Karumuri Venkata Subbarao]: 391–409.
SmithIan R. 2010. Hijacked constructions in second language acquisition: Implications for Sri Lanka Malay. Paper presented at the Workshop on Sri Lanka Malay Leipzig Nov. 10.
1 I am grateful to Romola Rassool, Mohamed Jaffar, Vani Manoharan and four anonymous Tamil speakers for their assistance. All errors are mine.
2 Adstratal influence typically involves languages of equal prestige, and the borrowing may proceed in both directions. Apart from a few lexical borrowings, however, Sinhala and Tamil have taken little from Sri Lanka Malay. The demographic dominance of Sinhala and Tamil speakers accounts for the unidirectional borrowing.
3 Ansaldo, by contrast, projects the present situation on the past, arguing that Sinhala, as the socially and economically dominant language, was more likely to be the source of features in Sri Lanka Malay; Tamil “cannot be totally excluded” but was “less likely to be a candidate” (2009a: 281). The documented social links between the Malay and Moor communities (Hussainmiya, 1990, and now Slomanson, 2010) are ignored.
4 See Smith (this issue, p. 12)
5 The nasal portion of “prenasalized” stops in Sinhala are actually of comparable length to (single) nasals before stops in other languages; thus phonetically the contrast is between long nasal + stop and normal nasal + stop (Ladefoged and Maddieson, 1996: 121, citing Lisker, 1984)
6 Geiger actually specifies a “nominal origin” (1938: 163)