Language change in a multiple contact setting: The case of Sarnami (Suriname)

The South American nation of Suriname features a situation of multiple language contact in which speakers use various languages in changing constellations, and often simultaneously. Sarnami (Surinamese Hindustani) shows traces of koineization of various Indian languages, and the effects of multilingualism involving Sranan Tongo and Dutch, the two dominant languages of Suriname. Sarnami has undergone substantial contact-induced change in its lexicon and grammar, including the rise of SVO alongside the inherited SOV basic word order. We conclude that the ever growing influence of Sranan Tongo and Dutch may lead to more extensive restructuring with similar outcomes as “creolization”. Traditional labels are therefore not always adequate beyond the realm of the canonical creoles involving European lexifiers and (West) African substrate languages.


Introduction
Suriname is well known among creolists for the three clusters of Creole languages that have emerged there: The coastal Creole Sranan Tongo, and the two clusters of maroon languages, i.e.
Western (including Saramaccan) and Eastern Maroon (including Ndyuka), for details cf. Carlin and Arends (2002). These languages have been the subject of a rich literature, both from a diachronic and synchronic perspective. There is also website with historical materials at www.ru.nl/suca. 1 We are grateful for comments from Bob Borges and Margot van den Berg, as well as from the editors and two anonymous readers. This paper was written with the support of the European Research Council "Traces of Contact" Grant to the LinC group at the Centre for Language Studies, Radboud University Nijmegen (www.ru.nl/linc). We are indebted to Renata De Bies, Lila Gobardhan-Rambocus and Motilal Marhé for their invaluable help in obtaining the Sarnami data on which this paper relies.
No matter how interesting these Creoles are -and they have been at the centre of discussions about Creole genesis ever since Schuchardt's applied his theory of Sprachchemie ("language chemistry", cf. Schuchardt 1914) to them -Suriname harbours a number of other very interesting phenomena of language contact. The country is a unique laboratory of language contact, in which about about twenty languages with very different typological make-ups interact with each other in a whole range of diverse contact scenarios. Most importantly, Suriname features a situation of multiple language contact, in which more than two languages are normally in contact with each other in various constellations. In the case of Suriname, there are two dominant languages, namely Sranan Tongo (henceforth referred to by its short name Sranan) and the colonial language Dutch that function as donor languages to other languages while simultaneously exerting an influence on each other. Trilingualism (and competence in even more languages) among a good proportion of Suriname's population has been leading to interesting contact-induced changes and convergence phenomena in the languages of the country.
In this paper, we illustrate some of these changes by focusing on Sarnami, a language spoken by the Indian-descended population of Suriname. Sarnami is not a creole language in the traditional sense, and certainly not a pidgin. Rather it exemplifies the survival of an Indian diaspora language in the Caribbean. Sarnami, referred to by its speakers as Sarnámi (the accent denotes a long vowel) or by its Dutch name Hindostaans is the result of the koineization in Suriname of several closely related languages spoken in the present-day Indian federal states of Uttar Pradesh, Bihar and Jharkhand. The languages that were to merge and become Sarnami were transplanted to Suriname during the indentured labour trade of the 19 th century, during which tens of thousands of people were shipped from India to the Caribbean by the European colonial powers in order to substitute for African slave labour after the abolition of slavery. Sarnami is the only Indic language in the Caribbean with roots in the colonial indenture trade that still enjoys a stable speaker community.
Related varieties in Guyana and Trinidad have all but disappeared. Sarnami is the result of a number of language contact scenarios: leveling as a koine in the situation of diaspora; superstrate influence from the colonial prestige language Dutch; and adstrate influence from the Surinamese national vernacular Sranan. The combination of these three processes has yielded a unique new language. By examining a less known language of Suriname like Sarnami, we hope to broaden the discussion of language contact in Suriname beyond the scope of the languages traditionally referred to as creoles.
The emergence of Sarnami and its present situation are the result of a complex set of circumstances, both social and linguistic. A considerable part of this paper is therefore dedicated to shedding light on historical and socio-linguistic aspects of multilingualism in Surinamese society in the past and present.
This paper is organized as follows. In section 2 we discuss the theoretical background by reviewing a number of relevant scenarios (including maintenance, shift, and creation). In section 3 we describe our methodology. Section 0 presents the context of multilingualism and language contact in Suriname, both from a historical and a contemporary perspective. In section 0, we move on to Sarnami, the main focus of this paper, from the perspective of koineization, code-switching and borrowing, as well as language change due to adstratal influence. Section 6 contains points for further discussion and some conclusions.

Theoretical background
Language contact and mutual borrowing of lexical items and structures in the languages of Suriname is a consequence of widespread multilingualism. We will therefore first review some of the concepts related to language contact and multilingualism that we will be referring to. The typology of language contact that Thomason and Kaufman (1988) propose provides for three principal contact scenarios. We define scenario as the organized fashion in which multilingual speakers, in certain social settings, deal with the various languages in their repertoire.
In a maintenance scenario the language that borrows (henceforth the recipient language), from another language (henceforth the donor language) continues to be spoken by its speaker community, i.e. it is maintained. The literature shows that there is a large range of variation in maintenance scenarios. In some cases of maintenance, the recipient language may undergo more moderate lexical and structural transfer from a donor language (e.g. Heinold 2009 on transfer of English morphology to French and German as a case of rather moderate transfer; Bubenik 1990 on the influence of Persian and Arabic on western Indic languages as instances of extensive lexical borrowing and more moderate structural transfer). Other cases of maintenance show extensive transfers of phonological features, lexical material and structural patterns (e.g. Hainan Cham, whose Austronesian typological profile has been significantly altered due to contact with Sinitic, cf. Thurgood & Li 2003). The classification of a scenario as involving maintenance may also be theorydependent. For example, a strong position on relexification -i.e. the mapping of one language's semantic and morphosyntactic properties onto another's phonological shapes -may in fact be seen as a case of maintenance of the language providing the semantic and morphosyntactic content.
Such a position is implicit in Lefebvre's (1998) interpretation of the rise of Haitian Creole.
In the second scenario suggested by Thomason and Kaufman (1988), a community leaves behind its traditional language and shifts to another language, usually due to the socio-economic and/or political dominance of the community speaking the language shifted to. Contact effects in shift scenarios may be very similar to those encountered in maintenance scenarios. Studies have shown that intermediary stages of language shift and obsolescence (cf. e.g. the case studies in Aikhenvald 2012) show the same kind of heavy structural and lexical borrowing that may characterize maintenance scenarios in which a recipient language is not threatened by language loss (for an illustrative example, cf. Gómez-Rendón 2007).
A principal difference between shift and maintenance is pointed out by van Coetsem (2000): Language shift involves a change in directionality of borrowing (termed "agentivity" by van Coetsem) between a recipient language and a source language. Hence during a shift, contact effects chiefly manifest themselves through structural rather than lexical influence from a shifting community's traditional language which is usually still spoken alongside the dominant language by some proportion of the shifting community. In a maintenance scenario, however, the traditional language of the community remains the dominant language and lexical borrowing is usually far more common, or at least as common as structural borrowing from the donor language. The distinction is also relevant for Suriname, which features a range of maintenance scenarios of varying depth or extensiveness of contact.
The third major scenario proposed by Thomason and Kaufman (1988) involves the creation of new linguistic systems composed of elements of contributing languages. Creolization as one type of language creation is particularly important in the linguistic trajectory of Suriname. In the Surinamese creolization scenarios, European superstrate languages (English and Portuguese) provided most of the lexicon while several African substrate languages provided some lexicon and substantial parts of the grammatical and phonological systems. Next to genetic inheritance from contributing languages, creolization in Suriname also seems to have involved various degrees of restructuring of the input languages driven by linguistic-cognitive factors -the respective contribution ascribed to either of the two factors being subject to theoretical leaning (Alleyne 1980;Lefebvre 1998;McWhorter 2005;Bickerton 2009).
Language creation in Suriname concerns not only creolization but also koineization as diachronic and synchronic processes. We understand koineization as a less pervasive type of language creation in that there is less restructuring of the input languages involved in the creation of the koine, as has been amply observed in cases of dialect contact (cf. e.g. Auer 1998 and the classic study of the rise of the Indic koine of Fiji, cf. Siegel 1985). The literature suggests that typological proximity and mutual intelligibility are the chief linguistic reasons responsible for the more modest restructuring of an interlanguage or koine with respect to its input languages (cf. e.g.the studies in Braunmüller 2009;

Kühl and Petersen 2009).
A stable multilingualism over some generations, as in Suriname, can lead to structural convergence between the various languages spoken in the same geographical space (Winford 2003). In the process, the languages in contact may become more similar by mutual accommodation, i.e. bidirectional change, for example by adopting a compromise on the basis of already existing common structures. In this paper, we employ "convergence" in a broader sense, however, as a cover term for the multiple contact scenario characteristic for Suriname. Here, borrowed structures may stem from the two dominant donor languages Dutch and Sranan simultaneously, and these two languages may interact in their influence on a recipient language. Due to this circumstance, it is often difficult to attribute instances of contact-induced change in a language like Sarnami to a single source.
In our classification of contact phenomena in the Surinamese languages we rely on models that differentiate between the borrowing of forms or matter (morphemes and their phonological shapes) and structures or patterns (morphosyntactic and semantic structures without the corresponding forms). The latter phenomenon has been also been referred to in the literature (with varying degrees of overlap in meaning) by terms like "calquing" (Haugen 1972), "metatypy" (Ross 1996), "grammatical replication" (Heine and Kuteva 2003;Heine and Kuteva 2010), "pattern replication" (Matras and Sakel 2007;Sakel 2007), "rule borrowing" (Boretzky 1985) and last but not least "relexification (e.g. Muysken 1981;Lefebvre 1993).
Such an approach not only allows operationalizing these two fundamental types of borrowing, it also allows yet finer distinctions of structural borrowing. Borrowing of patterns allows us to differentiate for example, between the replication of lexical versus grammatical structures. It may also encompass cases of partial replication in which a donor language pattern undergoes adaptation, i.e. is grammaticalized to fulfill functions in the recipient language that differ to some degree from those attested in the donor language (Heine & Kuteva 2003;Meyerhoff 2009). The differentiation between the borrowings of forms (matter) and structure (pattern) also leaves room for identifying combinations of matter and pattern borrowing, in which a form and its morphosyntactic and semantic specifications are carried over into another language. As we move on, we will see that both types of borrowing and combinations between them can be found in our Surinamese corpus.
Before moving on to the next section, we wish to point out that we share the general understanding that the outcomes of multilingualism and language contact are of course not solely determined by linguistic factors. Socio-economic, political, cultural and demographic factors, the time-depth of cultural and linguistic contact between communities and so forth, are at least as important in fashioning the processes and outcomes of contact between languages (Myers-Scotton 1993;Roberts 2005;Gómez-Rendón 2008). Our focus in this paper is on the linguistic as much as the extra-linguistic factors of contact-induced language change in Suriname.

Methods and data
This study relies for the largest part on field data collected in Suriname in 2011-12 as part of the ERC Mauritius (Mauritian Bhojpuri) make up about thirty hours. All unreferenced examples in this paper stem from our own field data.
The data was collected according to a unified methodology in order to allow comparison across varieties and languages. Data collection methods involved the use of broad (story-based) and narrow (video clip-based) visual stimuli on the one hand and (semi-)structured interviews on specific topics on the other. Elicitation was complemented by recordings of natural discourse. In Suriname, we also conducted about fifty sociolinguistic interviews in Sranan on the backgrounds of speakers and their attitudes vis-à-vis the languages they speak.
We are much in favour of approaches employing quantitative analyses based on large diachronic and synchronic corpora in order to differentiate between codeswitching and borrowing, as well as between "normal" variability and contact-induced change (e.g. ( Van Hout & Muysken 1994;Poplack, Zentz & Dion 2012). However, when working with less documented languages, as in the case of Suriname, one is in a less fortunate position. There is a lack of sizeable corpora of diachronic data for all languages but Sranan (cf. http://suca.ruhosting.nl), and the collection and handling of even modest corpora of synchronic data involves considerable efforts. It seems then that only a mixed strategy is feasible. This involves quantitative investigations based on smaller corpora and extrapolation based on in-depth morpho-syntactic investigations of particular structural areas.

Multilingualism and language contact in Suriname
Suriname has been the scene of complex and overlapping population movements throughout its history. In this section, we give a brief overview of how these movements have driven the development of multilingualism in the recent history of Suriname and in present times. Patterns of community-wide multilingualism have probably characterized the societies of Suriname from well before colonial conquest. Linguistic diversity in Suriname has increased significantly since the beginning of the colonial period, reaching a peak in contemporary Suriname and ushering in the type of extensive language contact that characterizes the country today (for detailed overviews of multilingual Suriname, cf. Charry, Koefoed and Muysken 1983;Carlin and Arends 2002).

Historical overview
Key events in the history of Suriname and their sociolinguistic significance that we refer to in the following are presented in

Precolonial contact and creolization
Taking pre-and early colonial times as a starting point, there were originally three indigenous language families represented on the territory of present-day Suriname, namely Warao, Carib, and Arawak (cf. Hoff 1995). Particularly striking is the partly convergent development within the Arawak and Cariban languages, including the creation of a 16th century Carib Coastal Pidgin (Taylor and Hoff 1980). Convergence must have been the consequence of multilingualism in a situation of maintenance as defined in section 2, probably with both Carib and Arawak enjoying similar degrees of prestige.
With the beginning of European colonization of Suriname in the 17 th century, the linguistic situation becomes more complex. The Netherlands ends up being the sole colonial power in 1667. The establishment of a plantation economy leads to the deportation from the western seaboard of Africa of an estimated total of approximately 350'000 Africans between 1675 and 1803 (Postma 1990). Arends (1995)  The Creole languages of Suriname, however, thrived and have differentiated into the three distinct clusters of Sranan, Western and Eastern Maroon Creole (Smith 1987;Smith 2002). Amongst these, Sranan has spread beyond the coastal belt into the interior to become the most-widely spoken Creole of the country. We should also mention that concomitant with creolization were other types of language creation. In the emergence of Ndyuka-Tiriyo pidgin, for example, pidginization was the dominant process (Huttar and Velantie 1997).
The indigenous languages of Suriname have undergone quite fundamental contact-induced changes since colonization as well, both through contact with each other (Carlin 2006) as well as with Sranan and Dutch (Rybka 2009). After independence and with the increase of regional and global economic and migratory flows, significant communities of speakers of Guyanese Creole, Haitian Creole, Brazilian Portuguese and Mandarin have further increased the linguistic diversity of an already highly heterogeneous society.

The Asian languages of Suriname
The full abolition of slavery in 1873 after a transitional period of ten years of forced labor prompted the Dutch colonial regime to "import" indentured labourers from Asia, as in other plantation economies throughout the Caribbean and elsewhere in the colonial world in order to substitute for slave labour (Saunders 1984;Kale 1998). Through these arrangements, a total of about 30000 (male and female) labourers were transshipped to Suriname from northern India between 1873 and 1916 (Damsteegt 1988: 95). A total of about 30000 labourers arrived from Java (Indonesia) between 1890 and 1939 (Bersselaar, Ketelaars and Dalhuisen 1991). Surinamese population of half a million self-identifies as "Hindostaans" (Indian-descended) and 15% as "Javaans" (Javanese-descended) while the category "others" of 6% subsumes amongst others the Chinese-descended population and the Indigenous peoples of Suriname. Self-identified "Kreolen" and "Marrons" (both African-descended) Surinamese make up 18% and 15% respectively of the population. The substantial number of Surinamese who self-classify themselves as "mixed" (12%) or leave their ethnicity unreported (6%) might be indicative of a growing proportion of Surinamese either claiming a mixed heritage of various constellations or rejecting ethnic labeling altogether.
The migratory mass movements of the indenture period have been equally transformative for the linguistic situation in Suriname as they have been for the demography of the country. Various northern Indian language varieties merged to form the koine Sarnami, the community language of the Indian-descended population of Suriname (Damsteegt 1988). Here follows a brief summary of language contact tendencies involving Asian and non-Asian languages of Suriname apart from Sarnami, which is covered in more detail in section 5. Besides change due to contact with Sranan and Dutch, some degree of koineization also affected the Javanese language since it arrived in Suriname (cf. Vruggink 1987). This is probably also due to the fact that a small but not insignificant part of the "Javanese" population of Suriname had its origins elsewhere in the Indonesian Archipelago than Java (Gobardhan-Rambocus and Sarmo 1993). One change that has been documented is the erosion of the formal 'high' registers of speech and the corresponding abandonment of a complex system of honorificity expressed through personal pronouns in Surinamese Javanese (Wolfowitz 2002) as well as extensive borrowing from Sranan in particular, but also Dutch (Gobardhan-Rambocus and Sarmo 1993). Contrary to Sarnami, there are indications that Javanese is not as vital anymore as it still was in the second half of the 20 th century and that there is an ongoing language shift, particularly by speakers below twenty to Sranan and Dutch. Our recordings of Surinamese Javanese show effects of language attrition with younger speakers, i.e. frequent hesitation and the use of repair strategies, retrieval difficulties, and morphological simplification.
The language of the Chinese community was, for a long time, chiefly Hakka (also called "Kejia").
But was not yet as widespread within these communities as it now is. Nowadays Sranan plays the role of a donor language together with Dutch. Sranan is the only language of Suriname that virtually every Surinamese has at least some knowledge, however in growing competition with Dutch. It should be pointed out that the expansion of Sranan is solely a consequence of an incremental growth because the language has not benefited from state support of any kind whatsoever since it was abolished as a language of instruction in 1876 (cf . Table ).
This stands in stark contrast to the development of Dutch, which has also witnessed a considerable growth in speaker numbers throughout the 20 th century due to sustained institutional and elite support. Since colonial times, Dutch has been the sole language of government business and parliamentary affairs, and the de facto language of education at the primary, secondary and tertiary levels. It has remained the language of upward social mobility and high prestige and is extensively used by officialdom and by coastal Surinamese in a variety of registers. One of the consequences of this disposition is that Dutch has witnessed a fundamental transformation within the last hundred years or so. From being a language of the colonial administration and a relatively small Dutch- Suriname is reflected in often negative and self-denigrating attitudes of speakers towards the non-European languages they speak and in the corresponding language practices. Our field data shows the existence of a range of attitudinal and communicative patterns with respect to Sranan as a concomitant of its social subordination to Dutch, compounded by negative attitudes related to its historical origins in slavery. However, the social and functional division of labour between Dutch and Sranan outlined above has also led to Sranan enjoying a large amount of covert prestige. In many contexts, using Sranan is an act of identity assertion, defiance and resistance against norms transmitted through Dutch, with all its problematic associations with elitism, the colonial past and a post-colonial present.

Data on multilingualism in Suriname
Determining the size of speaker communities in present-day Suriname is not easy in the absence of a comprehensive linguistic survey. We therefore have to rely on self-reported language knowledge in official census data or extrapolate from existing micro-surveys. The 2004 official census is the only one so far to provide figures for selfreported language use in all districts. Households were asked to name the "language spoken most often" and the "second language spoken". The figures are given in  Table 2 confirms the observation made further above that Sranan and Dutch constitute the two main axes of multilingualism. The two languages show the highest total percentages of "most often" and "second language" uses. At the same time they manifest the largest differences between "most often" and "second language" uses. The differences in social function between these two most widely spoken languages of Suriname transpire in the significant differences in percentage of "most spoken". The percentage of 9% for Sranan for "language spoken most often" is surprisingly low, particularly in comparison to an equally surprisingly high score of 46,6% for Dutch. We attribute these percentages to prevailing language attitudes in Suriname that result from the functional and prestige differences between these languages referred to in the preceding section. Hence the high prestige of Dutch leads to overreporting of use as "language spoken most often", while the, the low prestige of Sranan leads to underreporting of use as a primary language.
As for the other languages listed in Table 2, the lower percentages in the "second language" column seem to point to these languages largely functioning as in-group "ethnic" languages. For Sarnami for example, the relation of "most spoken" (about 75% of the total) and "second language" (about 25% of the total) may well be indicative of a partial language shift to Dutch and Sranan, or at least a certain decline in use. The same holds for Javanese.
We have seen that language creation has been of primordial historical importance for the rise of linguistic diversity in the country. In the present, we find the maintenance of community languages alongside language shift to the two dominant languages, Sranan and Dutch. In section 5, we will show the types and extent of interaction that these two languages can have with another Surinamese language. We focus on Sarnami, the language of the Indian-descended community of Suriname in order to explore aspects of language contact in more detail.

Sarnami: koineization, contact and maintenance
In this section, we turn to Sarnami and explore in more detail how the situation of widespread multilingualism and multiple language contact has affected a particular language since its implantation in Suriname. In section 0, we discuss koineization, the process which gave birth to Sarnami as an independent language. In 0, we address contemporary contact-induced change in the grammar of Sarnami. We conclude that Sarnami appears to be converging towards Dutch and Sranan in the domain of clausal word order. We assume that the main cause of this development is wide-spread multilingualism in Sranan and Dutch in the Sarnami-speaking community.
Structural and lexical features seem to indicate that Sarnami is the result of the mixing of at least three languages of the Bihari group of Indic, namely Bhojpuri, Magahi and Maithili. These three languages are classified as languages in their own right, separate from Hindi (cf. Masica 1993: 12ff.). The grammar and lexicon of Sarnami also reflect the influence of varieties of the eastern reaches of the Hindi-Urdu continuum, namely Braj and Kannauji, generally classified as varieties of Hindi, (ibid.) and Awadhi, generally seen as a variety more distinct from Hindi albeit closely related.
As previously mentioned, Sarnami is the only Indic language of the Caribbean with a sizeable speaker community. Trinidad Bhojpuri and Guyanese Bhojpuri have only very few elderly speakers left with younger speakers having completed the shift to Creolese (Guyanese Creole, Gambhir 1981 and Trinidad Creole, Mohan 1990, cf. Boer (1998 for the complex mix of circumstances that has favoured the survival of Sarnami as opposed to its Caribbean sister languages).
We will henceforth collectively refer to the Indic languages that constituted themselves during the indentured labour trade as overseas Indic (rather than overseas "Hindi", cf. Barz and Siegel 1988).
Our data and sociolinguistic interviews seem to indicate that Sarnami is actively used by all generations within the Indo-Surinamese community in a pattern of trilingualism involving Sranan and Dutch as additional languages. Our observations point to a functional specialization with Sarnami serving largely as a language for the informal and family domain. A significant aspect of this pattern of trilingualism is that the Sranan and Dutch are nonetheless extensively used within the Indo-Surinamese community and not simply reserved for out-group communication. The following two sections describe how two types of language contact have affected Sarnami during and since its emergence in Suriname. Through koineization the speakers of various closely-related Indian languages forged Sarnami into a common community language. Language contact with Sranan and Dutch, two typologically more distant languages, has affected Sarnami in yet other ways.

Sarnami as a koine
In the following, we provide a brief overview of the koineization of Sarnami. We concentrate on the verbal system to show the effects of this process on the shape of Sarnami (a more general overview including other aspects of koineization in Sarnami can be found in Damsteegt 1988  A complicating factor is that Hindi-Urdu has served as a prestige supra-lect and written register to the languages of north-eastern India at least since the departure of indentured labourers in the 19 th century and has continued to do so in the Indic languages of the Caribbean and Indian Ocean (Bhatia 1982;Damsteegt 1990 28 for South Africa). A preponderance of Bhojpuri-speaking immigrants has also been proposed for Suriname (Damsteegt 1988: 28). The linguistic evidence from these sources seems to suggest an important role of Bhojpuri in the koineization process that the north-Indian languages underwent in their new homelands.
We will show in the following, however, that in the structural area singled out in this section, there is quite a degree of inter-variety mixing in Sarnami. Although Sarnami has a strong Bhojpuri import it is difficult to establish with certainty that the Bhojpuri element is dominant. One of the principal reasons for this is the existence of intra-and interlectal continua within and between the contributing languages. Centuries of multilingualism would have led to convergence of the languages spoken in the areas where indentured labourers came from (cf. e.g. Abbi 1997 for convergence in present-day Jharkhand).
Sarnami shows the characteristic effects of koineization that have been widely documented in the literature, namely mixing, leveling, simplification and reallocation (Kerswill 2002). Table 3 below features verbal suffixes that serve to express perfective/past aspect/tense in Sarnami, Bhojpuri, Awadhi, Maithili and Magahi. We provide two varieties of Bhojpuri in order to exemplify the degree of intralectal variation in one of these languages (and the possibility of mutual borrowing by the contributing languages): The southern standard variety, described by Tiwari (1960) and Sadani Bhojpuri, spoken in southern Jharkhand (Horstmann 1969). For now, the table only contains nonhonorific suffixes. Honorificity will be covered below. Variants are separated by a comma, feminine gender forms are provided in parentheses where they exist:  -The /-lẽ/ variant of 3PL is found in southern Bhojpuri alone.
It is noteworthy that Southern Bhojpuri features gendered suffixes in the 2 nd and 3 rd persons and separate suffixes for plural number. Some varieties of Awadhi also show gendered verb suffixation in intransitive clauses (Saksena 1971: 249). But the lack of a gender distinction in the verbal morphology of Sarnami may not be symptomatic of simplification. The use of feminine gender seems to have been optional in Bhojpuri itself for a long period of time (Gambhir 1981: 249)  Firstly, the four Indian contributing languages feature a larger range of honorificity distinctions than Sarnami, reflected both in the use of specific personal pronouns as well as in separate inflectional suffixes on verbs. Table 4 provides the corresponding personal pronouns and person endings for perfective/past in Sarnami, southern Bhojpuri and Maithili. For clarity of exposition, Table 4 features the past/perfective person suffixes alone; it does not include the past/perfective tense-aspect marker /-l/:   Table 4 shows that Maithili represents the most complex type. It features a four-fold honorificity distinction in the 2 nd person (non-honorific, mid-honorific, honorific and high-honorific) and a twofold distinction in the 1 st and 3 rd persons (non-honorific vs. honorific). Honorificity is expressed both by the use of distinct personal pronouns and/or dedicated verbal suffixes. Whenever the personal pronoun remains the same, verbal suffixation expresses honorificity and vice versa. This is, for example, the case with the distinction between 1SG.FAM and 1SG.HON as well as 2SG.FAM and 2SG.MHON.
Southern (standard) Bhojpuri is second in line with one layer less in the 1 st and 2 nd person paradigm.
However, Bhojpuri features the usual additional gender differentiation (in parentheses) that we already encountered in Table 3 above. With only two levels of honorificity in the 2 nd person and no differentiation according to gender, Sarnami shows the lowest number of distinctions.
A second feature of interest in Table 4 is that Sarnami also shows signs of mixing. The 2SG honorific pronoun āp is also found as a 2SG high honorific pronoun in southern Bhojpuri. Tiwari

Codeswitching and borrowing
In the following, we discuss some of the effects that multilingualism and language contact have had on Sarnami. We first take a look at codeswitching as a conventionalized linguistic practice within the Sarnami-speaking community. We then turn to structural borrowing into Sarnami in the next section.
Codeswitching is seen as a principal cause of contact-induced change and convergence (e.g.

Muysken 2000; Backus 2004
). Codeswitching is also hypothesized to have accompanied the rise of mixed languages like Gurindji Kreol (Meakins 2011) and Media Lengua (Muysken 1981;. The data presented in the following paragraphs highlights some of the strategies employed during codeswitching in order to accommodate non-native material from Sranan and Dutch, two languages that are typologically quite distinct from Sarnami. Multilingual competence coupled with frequent codeswitching also appears to lie at the heart of lexical and structural borrowing in Sarnami. The following sentence exemplifies the kind of codeswitching that characterizes our Sarnami In spite of its brevity, the text reveals in an exemplary fashion some of the tendencies of Sarnami as a language in contact: (1) a substantial percentage of tokens, namely a quarter, consists of nonnative items of either Dutch or Sranan origin; 3 All words in the respective languages and word classes were counted and we did not exclude changes in the base language, i.e. entire (chunks of) sentences in Dutch and Sranan. We do not attempt to make a distinction between established loans and nonce borrowings, hence the column entitled "Sarnami" simply lists words with an Indic etymology. "Determiners" includes bound forms in Sarnami (hence ego 'a' as well as -wa/-ya 'DEF'). "Auxiliaries" lists unbound function words (all of which are verbal in this text) expressing tense, aspect and modal functions. "Adpositions" includes core casemarking postpositions like ke 'ACC/DAT' in Sarnami, and the preposition aan 'DAT' in Dutch, as well as elements expressing more peripheral semantic functions, e.g. meṉ 'in' (Sarnami) and naar 'to'. "Clause linkers" includes coordinating (aur 'and' (Sarnami) and subordinating linkers (ki 'that'(Sarnami); omdat 'because' (Dutch)) as well as linking adverbs like tab 'then' (Sarnami) and dan 'then' (Dutch).
(2) the Sranan percentage of the total of tokens (4%) is far lower than the Dutch percentage (22%); (3) although the sample contains non-native items from all word classes, nouns (33% non-native), adverbs (46% non-native) and clause linkers (36%) score particularly high in the text sample. It is well reported that nouns, free (non-selected) constituents and pragmatic elements including clause linkers are particularly prone to switching. The results of the Type count do not differ greatly from those of the Token count due to the small sample size. Nevertheless, one observation can be made with respect to the Type count: Only Sarnami shows a lower Type than Token percentage. The higher Type than Token percentage for Dutch and Sranan indicates that a higher proportion of words from these two languages are "one-shots", i.e. they only occur once in the text. This seems to indicate that the narrative backbone of the story consists of native (Sarnami) lexical items.
The amount of codeswitching recorded in Table 5 contrasts starkly with the situation encountered in the frog stories in the Indian sister languages of Sarnami. In a randomly picked frog story in Maithili of 790 words, we found not a single English item and only three tokens of Hindi origin. In the same vein, a Bhojpuri frog story of 753 words chosen at random contained a total of 14 Hindi tokens and a single English token (namely 'time').
A word is in order on the low proportion of Sranan items in the text analyzed in Table 5 above.
For one part, the correspondingly high proportion of Dutch items is a consequence of the observer's paradox. As previously mentioned, the social functions of Dutch and Sranan differ widely. Dutch is the language of first contact in many communicative exchanges, and is often associated with a larger social distance between interactants. The recording setting with its air of formality, the potential for a hierarchical relationship due to the age-gap between interviewer and interviewee, and the association of Dutch with story-telling tasks inherited from the school-setting create a bias towards Dutch as the primary switching language. We have observed the tendency that the more egalitarian, in-group, male-only, informal and emotionally charged the context is, the more likely it is that Sranan becomes the primary language that speakers switch to. We have already mentioned that the use of Sranan vs. Dutch (codeswitching) is subject to complex socio-pragmatic norms conditioned by the respective status of Dutch and Sranan in Surinamese society. It is therefore well possible that Sranan exerts its influence on Sarnami more covertly, hence primarily through pattern borrowing (whether lexical or grammatical) while Dutch does so covertly and overtly, hence via a combination of matter and pattern borrowing.
There appears to be a diachronic dimension to the relative importance of Sranan vs. Dutch as donor languages to Sarnami. In Suriname, the use of (spoken) Dutch, including the entire range from native-like to diverse types of L2 varieties, seems to have grown exponentially in the past halfcentury or so.
During the colonial period, however, Sranan appears to have played a far more important role than Dutch as a donor of lexical material to Sarnami. The most comprehensive dictionary of Sarnami (Santokhi and Nienhuis 2004) contains hundreds of items of Sranan origin from diverse semantic fields such as the natural habitat, culture, and technology, and far fewer from Dutch. There appears to be a consensus amongst Sarnami native-speaker linguists that such items have entered the Sarnami lexicon for good (e.g. Kishna 1979;Marhé 1985;Santokhi and Nienhuis 2004). Their widespread acceptance points to them having been borrowed into Sarnami sufficiently long ago, maybe during or in the aftermath of the indenture period. Some examples are given in (2)  Society-wide, generalized bi-or multilingualism in a donor language is not a prerequisite for lexical borrowing into a recipient language of the type shown in (1) above, not even in cases of extensive lexical borrowing. It is usually sufficient for a small but influential proportion of (passive) bilinguals to introduce non-native lexical items (cf. e.g. Sakel 2007: 25). In fact, we must assume that widespread multilingualism involving Sranan (and even less so Dutch) was not the norm in the Indo-Surinamese community before the deep social transformations that began in the mid-twentieth century and have been referred to above. Until then, spatial segregation, a predominantly rural settlement pattern, economic specialization, endogamy, lack of access to education and institutional discrimination by the colonial state (cf. Hira 1998) would have limited the possibilities of large-scale social interaction and hence the acquisition of Sranan, particularly for children and women. The latter were particularly restricted in their ability to network outside of the community due to economic restrictions imposed by the colonial state and patriarchal social structures (cf. de Koning 1998).
The distinction between nonce borrowings and loans is far more difficult to make synchronically, in view of the characteristic pattern of Sarnami-Sranan-Dutch trilingualism and the resulting omnipresence of codeswitching. An exhaustive treatment of the question of nonce borrowings versus loans would seem to require statistical analyses of the kind mentioned in section 3, with the known limitations of small data sets in this respect (cf. Poplack 2012). Nonetheless, a more intuitive approach may also serve to distinguish 'one-shot' nonce borrowings from established loans.

Numerous items of Sranan and Dutch origin in our corpus are not contained in Santokhi and
Nienhaus's 2004 dictionary of Sarnami, for example, but are overwhelmingly (at least 90% of tokens) used instead of a Sarnami equivalent throughout our corpus. Apart from items that stem from material or technological culture (e.g. trap> Dutch 'stairs', blik> Dutch 'tin'), there are also numerous items like the ones listed in (3). The list contains only predicates borrowed from Sranan and Dutch into Sarnami, loosely grouped according to semantic field. We address the bipartite nature of Sarnami predicates in the next paragraph below: ( We have seen that (3) above exclusively contains bipartite expressions featuring a Sarnami element on the one hand, and a Sranan or Dutch element on the other. These complex verbs may serve to shed light on the fuzzy boundary between codeswitching and lexical borrowing in contemporary Sarnami. Complex verbs in Sarnami come in two guises. One type is a "compound verb" involving a two-verb string in which a non-finite lexical verb is followed by an inflected member of a circumscribed group of light verbs. In (4) the Sranan verb senwe 'be nervous' combines with the Sarnami light verb máre, which literally means 'hit', but indicates rashness of action when used as an aspectual verb. Also note the presence of the inserted Dutch compound noun schooltoets.
very be.nervous 'hit'-INF be.PRS school-exam come-INF '(They) are suddenly very nervous (that) the school exam is coming.' The other type of complex verb is often referred to as a "conjunct verb" in Indic linguistics (e.g. McGregor 1995: 63). It involves a nominal element in place of the lexical verb, followed by one of the two generic verbs kare 'do' and howe 'be(come)'. Mixed conjunct verbs are a particularly productive means of integrating Sranan and Dutch lexical material into Sarnami structures, and in this way, constitute the principal means of deriving new verbal meanings. The phenomenon of mixed conjunct verbs is also well known from contact situations involving other Indic languages (e.g. Borowiak 2007). The following sentence is an example of a mixed conjunct verb: the Dutch noun aanval 'attack' appears as the object of the Sarnami generic verb kare 'do' to render the translation equivalent of 'to attack'. Once again, note the presence of additional Dutch items, namely the noun uil and the pragmatic element toch: (5) auro bai-wá ke uil aanval kar-il, toch?
and boy-DEF ACC/DAT owl attack 'do'-PFVP right 'The owl attacked the boy too, right?' The use of mixed conjunct verbs gives Sarnami speakers quite some leeway with respect to phrasal constituency. The following example illustrates the great liberty with which speakers may manipulate morphosyntactic relations in code-mixed utterances: In (6), the Dutch compound verb druk maken is integrated into a conjunct verb structure featuring the Sarnami generic verb kare 'do'. The Dutch verb is carried over into the code-mixed utterance along with its argument structure, however only partially. In Dutch, druk maken selects a prepositional phrase introduced by over 'over, about' containing the stimulus NP. This feature is retained in the Sarnami sentence above. At the same time, the additional use of the Sarnami postposition pe 'on' after the stimulus NP replicates the semantics of the Dutch prepositional phrase, while retaining Sarnami (postpositional) constituent order -a more Sarnami-like structure would have involved the use of the ablative postposition se instead of pe. What is not transferred into Sarnami from Dutch are argument structure features relating to the experiencer role. In Dutch, druk maken is a reflexive verb, like many experiencer verbs. But in (6) there is no reflexive construction and the experiencer is marked for accusative-dative case by the postposition ke, hence in the conventional Sarnami way (Kishna 1981).
Similarly complex patterns of contact and strategies of integration of non-native material into Sarnami structures are found with the way the language treats Dutch verb-particle combinations.
Here too, we find the replication of lexical as well as morphosyntactic patterns of Sranan and Dutch in Sarnami. Consider the following Sarnami sentence, elicited through the frog story: (7) ego kaháni hai, u já haigá ego laṟká ke over.
a story be.PRS DIST go be.PRS a boy ACC/DAT over '(It) is a story, it is about a boy.' (Sarnami) Example (7) above features a verb-particle combination, a type of bipartite complex predicate commonly found in Dutch and other Germanic languages. The verb já 'go', while clad in a Sarnami phonological shape, reproduces argument structure features of its Dutch equivalent gaan 'go': Firstly, an inanimate noun like kaháni 'story' is not normally employed in a Theme semantic role in Sarnami. Secondly, the complex predicate additionally features the Dutch clause-final adverbial/preposition over. The two elements together form an idiom, a calque from Dutch gaan over '(to) be about', which is common as an introductory line of narratives in Dutch (i.e. 'this story is about', lit. 'goes over'). Constructions like the one in the Dutch near-equivalent sentence in (8) below, are the source of the Sarnami structure in (7): (8) dit verhaal gaat over een jongen en zijn kikker.
this story goes over a boy and his frog.
'This story is about a boy and his frog.' (Dutch) Yet Sranan might just as well provide the model for structures like já -over in (7) above. Compare (9), taken from the frog story rendered by a speaker of Sranan. The example involves the Sranan verb-particle sequence go -abra 'go over'. This structure, though ultimately of Dutch origin, has been nativized in Sranan and is not considered a Dutch calque by our informants: (9) a tori disi e go abra wan boi nanga en dagu.
DEF.SG story PROX IPFV go over one boy and 3SG dog 'This story is about a boy and his dog.' (Sranan) The presence of the Dutch item over in (7) above, rather than a Sarnami equivalent indicates that we are dealing with a combination of pattern and matter borrowing. There is nevertheless an important difference in the syntactic behaviour of over in (7) above and its Dutch and Sranan equivalents in (8) and (9). The Dutch element over is not an adverbial/particle in the Sarnami Sarnami and Dutch. The former language has no complex predicates, in which an adverb/particle with spatial semantics collocates with verbs to render all sorts of spatial and metaphoric meanings as in the Germanic languages (cf. e.g. Müller 2002). The semantic overlap between Sarnami and Dutch is, in this respect, the spatial meaning that the Dutch particle has.
The examples in this section seem to suggest that Dutch/Sranan contact in Sarnami approaches a situation referred to by Van Coetsem (2000) as the "extended mode of borrowing" or by Auer (1999) as the mode of "language mixing". In this situation, bi-or multilingual utterances become a default speech form, a phenomenon that has been observed in African high contact settings as well (Blommaert 1992;Auer 1999). Rather than involving a single base language into which non-native material is inserted or as in conventional codeswitching, speakers navigate back and forth with considerable ease between the grammatical systems of the languages in contact and adapt grammatical structures of the languages in contact to each other in flexible and innovative ways.
Such contact strategies appear to be possible due to widespread multilingualism in the Sarnamispeaking community, and it also seems to provide the backdrop to the kind of structural borrowing that we now turn to.

Structural borrowing: from SOV to SVO
We now turn to clausal word order, a grammatical domain in which language contact is very likely to be responsible for an ongoing change in Sarnami. Pragmatically unmarked word order of basic clauses is generally Subject -Object -Verb (SOV) in Sarnami (Marhé 1985: 26 SVO word order is also attested in our Sarnami field data. Compare the following sentence (15): (15) tab u dekh-il ego hol jamin meṉ.
then DIST see-PFVP one hole ground in 'Then he saw a hole in the ground.' (Sarnami) Our data indicates, however, that Sarnami enjoys a much higher frequency of SVO than that observed in our Indian control group. The high frequency of SVO in Sarnami is subject to little variation between speakers and higher than could plausibly be attributed to the pragmatic function of focus alone. Instead, SVO seems to be competing with SOV as an unmarked basic word order. We suggest this development is induced by contact with Sranan and Dutch.
For the purposes of this study, we looked at the distribution of SVO word order in a sub-corpus of  Table 6 below presents the absolute (provided as a number) and relative (expressed as a percentage over the total number of transitive clauses featuring overt objects) frequencies of SVO in both samples. Total absolute and relative frequencies are given in the last row. Only prototypically transitive clauses are considered in Table 6, excluding clauses involving Goal objects of the verb já 'go', inherent complements of conjunct verbs as well as object interrogative pronouns. Both nominal and pronominal objects are counted. We also excluded clauses that were formally SVO, but in which the transitive object was separated from the rest of the clause by a prosodic break, since such occurrences of SVO may be instances of afterthought topicalization of O.
At first glance, the relative frequencies in parentheses already point to a significant difference in occurrences of SVO between Sarnami and the two Indian samples. The lowest individual relative frequency of SVO in Sarnami (SP7 and SP8, 28%) is still more than twice as high then than that of the highest score in the Indian sample (SP5, 12%). The relative frequency of the total in Sarnami (36%) is nine times higher that of Bhojpuri/Maithili (4%). Furthermore, there is no Sarnami frog story with no occurrence of SVO at all, while the Indian sample contains three stories with an SVO relative frequency of 0%. The statistical evidence from our corpus is corroborated by other types of evidence: South African Bhojpuri, an overseas Indic variety closely related to Sarnami, is also reported to have had a higherthan-usual presence of SVO clauses due to extensive contact with, and shift of the speaker community to English (Meshtrie 1991: 183ff.). Further evidence that the higher occurrence of SVO in Sarnami is likely to be contact-induced comes from word order in Sranan and Surinamese Dutch. In the former language, SVO is the only acceptable word order in basic and complex clauses: Further analyses of our corpora should be able to establish whether this development is parallelled by corresponding changes in constituent order in other syntactic units, e.g. verb-object order in subordinate clauses, the order of head noun and relativizer in relative constructions, the order of verb and adverbial phrase as well as main and auxiliary verb. But with respect to word order in main clauses alone, the development in Sarnami is already a good example of the kind of convergence that typifies Suriname as a linguistic area, in which pressure for change on a third language is particularly strong when the two dominant languages Sranan and Dutch already converge in their typological make-up.
6. Discussion and conclusion In section 5, we described some of the many changes that Sarnami has undergone, as illustrative of the Asian immigrant languages of Suriname. We are currently investigating the same or very similar phenomena in Surinamese Javanese and Surinamese Hakka. One question may be asked with respect to Sarnami and its status as a product of language contact. Do the effects of koineization in Sarnami have something in common with other language creation scenarios and creolization in particular?
To begin with, Sarnami has principally retained characteristics of its (Eastern) Indic typological profile, which we only refer to in passing, e.g.: the use of retroflex consonants and stress; a TMA system that relies on inflection and the use of auxiliaries and generic verbs a nominal system featuring long and short forms of nouns, prenominal adjectives as well as the use of a classifier and postnominal determiners; last but not least, we have shown that the language still features SOV word order, albeit in competition from SVO.
In contrast, the Afro-Caribbean English lexifier Creoles spoken in Suriname and elsewhere in the Atlantic basin show far more ingenious mixing of typological features from their European and African source languages, as well as varying degrees of restructuring that appear to be the result of the work of general cognitive principles of language acquisition. Some of these features are: the existence of tone systems in some languages (e.g. Nigerian Pidgin, Faraclas 1985; Saramaccan, Good 2004) next to English-like stress in others (e.g. Jamaican Creole, Gooden

2007);
-TMA markers that are functionally reminiscent of the systems of the West African littoral (Ameka and Kropp Dakubu 2008), grafted on to English-derived forms like 'go' and 'done' (e.g. Yakpo 2009: 191ff.); the structure of English determiner phrases combined with the semantics of substrate determiner systems (e.g. Aboh and Ansaldo 2007).
Nevertheless, there are also similarities. Both koineization and creolization involve a systemic restructuring to varying degrees, often at the expense of inflectional and derivational morphology of the input languages (cf. Plag 2003 for an overview). In Sarnami, we have seen the partial demise of a multi-layer honorificity distinction expressed in verbal inflection and dedicated personal pronouns. But such changes do not wholly call into question the genealogical link of Sarnami with its contributing languages in the way that many Afro-Caribbean English-lexifier Creoles do, at least with respect to their lexifier English.
A main cause of the less radical departure of a koine like Sarnami from its contributing languages appears to be the structural and lexical similarity of these languages among each other. The high degree of typological proximity, structural equivalence and mutual intelligibility would have made it possible for the speakers of the different Indian varieties to codeswitch without significant structural constraints, and swap lexical items and function words with each other (as captured by the mechanism of 'congruent lexicalization ', cf. Muysken 2000;Deuchar, Muysken and Wang 2007).
In contrast, the emergence of the Afro-Caribbean English-lexifier Creoles of Suriname with their English lexifier-superstrate and their Kwa and Bantu substrates involved typologically far more disparate languages.
At the same time the socio-economic circumstances of the rise of the koine Sarnami were quite different from those under which the Surinamese creole languages came into being. Despite the difficult working conditions on the sugar estates, indentured labourers enjoyed a fair degree of social and economic self-determination and could maintain their cultural and religious institutions in Suriname including the use of their traditional languages. There was therefore far less of the cultural, social, political and economic alienation and upheaval that Africans were subjected to when they were deported to Suriname, and which the creation of creole languages with largely European lexicons vividly attest to.
A characteristic that nevertheless sets Sarnami apart from its Indian input languages is the amount of contact with Dutch and Sranan, both in the extent of codeswitching as well as in the more lasting changes we have described.
There are two possibilities how Sarnami will develop in the Surinamese context. One is that Sarnami shares the fate of its Caribbean relatives in Guyana and Trinidad and falls into disuse with speakers eventually shifting to Sranan and Dutch. As pointed out in section 2, the initial stage of language shift may at first not be discernible in its symptoms from extensive language contact in a maintenance situation. Only time will tell whether some of the contact phenomena observed in Sarnami are actually symptoms of a beginning language shift. Assessing the vitality of Sarnami will crucially depend on whether we are able to identify significant age-related differences in the language practices of Sarnami speakers.
The second possibility is that Sarnami is maintained as one of the languages of the Indiandescended population of Suriname. If this is the case, it is, however, also very likely that the influence of Dutch and Sranan on Sarnami will continue to grow. As a consequence, larger parts of the Indic lexicon might eventually be replaced and (certain subsystems of) the grammar may end up being restructured more fundamentally. The end result may well be a language that will have diverged more radically from its Indian relatives and may thus be more similar in nature to the Surinamese creoles in the extent of restructuring involved. This shows that traditional labels are not always adequate when we go beyond the realm of the canonical creoles involving European lexifiers and (West) African substrate languages.
The present chapter illustrates the rich possibilities that Suriname offers with respect to the study of creolization, koineization and other forms of language creation and contact-induced change. In Yakpo and Muysken (in prep.), we bring together some of the strands of research in this domain, interpreting them in a single comprehensive framework. To mention just a few, it is necessary to compare the dynamics of Sarnami with that of the other languages of Asian origin, i.e.
Javanese, as well as Hakka. Furthermore, the Amerindian languages in Suriname (from the Arawak, Cariban, and Warao families) have interacted with each other, with the Creole languages, and with the colonial language Dutch.
Third, the various maroon Creole languages amongst them Ndyuka and Saramaccan, are undergoing complex processes of koineization among each other, as well as convergence towards the dominant coastal Creole language Sranan. Sranan has become a very diverse and possibly "diffuse" language (Le Page and Tabouret-Keller 1985). Finally, Dutch itself has diversified in the Suriname context, interacting with the other languages of the country, notably Sranan.