1. INTRODUCTION. On the island of Tanna, Vanuatu, many people say that they don’t speak their language the way they used to. This language is Kwamera—or, to give it its local name, Nninfe¹ or Nfe (Lindstrom 1986; Lindstrom and Lynch 1994). Around 3,500 people living along Tanna’s southeastern coasts speak Kwamera. The rest of Tanna’s 33,000 inhabitants speak four other languages (Crowley 2000:69–71) spread across this 550 km² island. People often complain that today’s Kwamera is a ragged, degraded version of the purer, better, more complexly mysterious talk of their forefathers. Moreover, they foresee further linguistic degradation in store; adults not uncommonly accuse their children of using an even more debased style, as they decry the ignorance of  youth in general (see similar comments by Papua New Guineans quoted in Sankoff 1976:307; also Mühlhäusler 1979a:166). Elsewhere in Melanesia, some have predicted accelerating language death and the disappearance of vernaculars as Islanders shift to creolized national Pidgin languages, such as Papua New Guinea’s Tok Pisin (Kulick 1992) or to Bislama, the Pidgin English national language of Vanuatu (Crowley 1990, 2000, 2004; see also, e.g., Meyerhoff 2001, 2003).

Expanding national Pidgins pose obvious dangers for the survival of vernaculars. Use of Bislama in Vanuatu, since that country’s independence in 1980, has come to signal a speaker’s participation and identification with the national community. Particularly in urban settings, it marks a person’s engagement with important modern institutions that range from the nation’s Parliament, to the churches, to popular music and the media, to the recreational consumption of kava in Port Vila’s many nakamal or kava bars. Crowley (2004:7) has estimated that about 10% of Vanuatu’s population speaks a creolized Bislama as a first language—many of these speakers the children of mixed-language marriages born in one of the country’s two towns, Port Vila or Luganville. But Bislama has an impact far beyond these urban areas. It reaches into the most remote island village, spoken “by people in just about every part of the country” (Crowley 2004:1). A prestigious, rapidly spreading Bislama has potential consequence for the future of Vanuatu’s vernaculars in rural as well as urban homes (Lynch and Crowley 2001).

Bislama speakers include most of the population of Tanna. The Tannese, who have lived on their island for 3,000 years, currently maintain five principal vernaculars despite widespread multilingualism and Bislama code-mixing. Code-mixing is an obvious practice that may produce language change in situations of language contact, even though convergence or other sorts of change are not inevitable (Myers-Scollon 2002:298; Thomasan 2001:133; Winford 2003:14). Kukick (1992:261), however, who studied code-mixing of a national Pidgin and a vernacular in a small rural community in Papua New Guinea, identified this as one sociolinguistic factor that is leading to the replacement of the local language with Tok Pisin. 

Mufwene’s (2001) ecological model of language evolution traces language change back to individual speaker choices among variants and alternatives, as constrained by internal and external language ecologies. A language’s internal ecology comprises its speakers’ repertoire of structures and patterns. Internal ecology can give direction to linguistic change. For example, if people begin to borrow features from other languages (say, Bislama), they may be predisposed to choose features that most resemble what they already find familiar. External ecological factors include other languages in the neighborhood together with their prestige values and uses, the ebb and flow of human demography, and also social practices that shape speech events (e.g., a favored language of religious ritual, etc.). Within these ecological conditions, a language’s “vitality” (Mufwene 2001:199) reflects the degree to which speakers choose to employ local as opposed to exogenous words and structures. We might suppose that Bislama code-mixes potentially weaken Kwamera’s vitality, although this sort of weakness can also be a strength if people have reason to maintain the distinctiveness of both languages so as to preserve possibilities of code-mixing itself, along with any communicative functions that mixing might have.

This paper discusses Bislama’s infiltration into contemporary Tannese talk as apparent in Kwamera speakers’ Bislama code-mixes. I assess the level of Bislama use in transcripts of four debates that I tape-recorded in 1982 and 1983.² The paper concludes with some remarks on the phonological, morphological/syntactic, and lexical/semantic consequences of such mixes—on how Islanders’ insertions of Bislama into oratorical and everyday talk may or may not be effecting linguistic change in Kwamera. My conclusion is that Bislama, so far at least, has enriched rather than impoverished Tanna’s linguistic ecology. Islanders’ use of Bislama has not yet seriously undermined indigenous language. 

2. CHANGING KWAMERA. It is difficult to gauge, exactly, just how much language change has actually occurred over the past two hundred years on Tanna. In South Pacific societies, where knowledge is valued but stored principally in living memory, people entertain a degradational view of history. The past is a heroic past. Special powers and forces that were once common no longer exist today. And the elders of a community, who control its most valuable information, through mischance or a perverse venality, manage to die right and left without passing along important secrets to their heirs. Strengthening this pervasive degradational Melanesian Weltanschauung is a second expectation that ancestral talk is properly opaque and mysterious. Traditional songs are larded with meaningless words and phrases that people label the “speech of the ancestors” (Lindstrom 1985:332–333). Only the wisest among the living profess to understand the true meanings of these passages. Given a degradationalist worldview and a presumption that ancestral language is often ordinarily opaque, it is not surprising that many Islanders suppose their language to be changing rapidly for the worse. 

Among linguists, too, there is some debate about the rates of change manifested by Pacific languages. To account for the complex linguistic diversity of Melanesia, in particular, some theorists have posited quicker, more subtle forces for change than the ordinary, slower effects of linguistic isolation over time. Grace (1981), for example, has suggested that “a high rate of replacement through borrowing of core vocabulary” (1981:266) has speeded up the rate of language mutation in Melanesia. Additional Pacific sociolinguistic practices that might effect rapid linguistic change include word tabooing (Keesing and Fifi’i 1969) and the prestige value of showing off one’s multilingualism (Salisbury 1962:63; Sankoff 1976:291–292).³

And yet, returning to Kwamera, there is evidence that this language has not experienced marked change, at least over the past century. This is shown by a translation of the Christian New Testament. First published in 1890, and prepared by Tannese pundits tutored by Presbyterian missionary William Watt, it can serve as a baseline to measure language change.⁴ Copies of later editions of the book still circulate around Tanna. Although today’s readers are sometimes confused by early mission orthography, by a formal style, and by the embarrassments of translating Christian myth into island terms, much of the language itself remains accessible.⁵

There is one major difference, however, between Watt’s New Testament and Kwamera today. This is the degree of penetration of Bislama into contemporary talk. Some nineteenth-century Islanders learned South Seas Jargon, or “sandalwood English” as long ago as the 1830s, and Bislama itself began to stabilize on overseas and local plantations by 1890 (Tryon 1979:73–74). Islanders have been speaking Bislama for over a century and, during the past 50 years or so, nearly everyone born on Tanna has learned the language. Even though Watt began his translation efforts in the 1870s, Bislama is little apparent in his texts. In the New Testament, Watt does rely upon what might be called “Christian jargon” that he either invented or borrowed from translation attempts elsewhere in the Pacific.⁶ He introduced Biblical or Polynesian forms such as angelo (angel), Atua (god), diabolo (devil), nakalasia (church), napostelome (apostles), yaprofeta (prophet), and so on. He also made use of English terms for items beyond the semantic ken of the island, e.g., thron, lamp, kold, bras (brass), Paradais, kovernor, ship (sheep), and vain (wine). Not many of the text’s neologisms or borrowings clearly derive from Bislama. For example, “thank yu,” “work,” and names for numbers above five may come from either Bislama or English. Only a few terms such as kurimatau ‘cow’ are obvious early Pidgin etyma. 

Despite the gross translation difficulties Watt faced, the number of alien forms in his testament is not great.⁷ The book’s vocabulary is almost entirely Kwamera. Comparatively, Bislama mixing in contemporary spoken Kwamera is more far reaching. This comparison—between a nineteenth-century literary translation and twentieth-century oral discourse—can only be suggestive. Watt and his island collaborators purposely may have rejected available Bislama forms for their text, aiming at scriptural formality. Even so, the book provides at least prima facie evidence that confirms local speakers’ suspicion of increasing latter-day reliance on Bislama. But is such mixing reshaping, or even eventually exterminating, languages like Kwamera?

3. CODE-SWITCHING, CODE-MIXING, AND BORROWING. Bislama penetrates Kwamera at several different levels as people mix the two codes in their talk. Initial studies of various sorts of language interference were concerned to discriminate mixing from shifting from borrowing. Typically, code-switching stands at the least intrusive end of a penetration continuum. At the most intrusive stands borrowing. Somewhere in the middle is code-mixing. In code-switching (or “code-changing”), speakers shift back and forth between two codes such as Bislama and Kwamera. Switches take place at constituent boundaries in talk, and often relate to a corresponding change in speech event affect, mode, topic, interlocutors, or some other component of the situation in general (see Pfaff 1979). In Vanuatu, as in Papua New Guinea, specific speech contexts or aspects of contexts evoke the use of Bislama: in some communities, “Pidgin is reserved for strangers, some pseudo-sophisticated court cases, joking, and dogs” (Bee 1972:69). 

Unlike code-switching, code-mixing occurs intrasententially. This results in talk whose “host” code is peppered with words and phrases from a “guest” language. Consequently, if code-mixing serves any conversational functions, these must be more subtle than the grosser, contextual changes that code-switches typically reflect. 

Finally, to distinguish code-mixing from borrowing, one must determine if  speaker are mixing forms from two language that they control or, conversely, if they are speaking a variety of a single language that happens to have a lot of borrowed forms. Borrowing may be suspected if the alien term fills a lexical gap in the host language; if the element is a single word; if the greater part of the guest language is not available for use by speakers; if the element has been nativized phonologically and morphologically; or if monolinguals also use the word. As Pfaff (1979:297) has noted, however, this discrimination may be difficult to make, hinging as it does upon evaluation of a particular speaker’s linguistic competence.

A concern to discriminate mixes from borrowings turns around the functionalist assumption that code-mixings can add metamessages to talk. Conversational functions of several sorts have been suggested for code-mixing (cf. Ennaji 2005:142–143). These range from foregrounding a shift in register (e.g., “now we are talking ‘legal’,” or “now we are talking ‘business’”); foregrounding the relationship between speakers (e.g., “please consider me a group member”); elucidating or emphasizing a statement by juxtaposing terms with more or less the same meanings from host and guest languages (in reraha parhien, no gud tru! / ‘he is really bad, really bad!’); or neutralizing a dangerous host form (e.g., Bislama fak yu used in place of the much more offensive Kwamera term ik nehi (‘you fucked’). Plainly, use of nativized, borrowed forms (as opposed to code-mixed items) could not signal the stylistic shift that is necessary to serve these several conversational functions—at least not to the same degree. 

It is often difficult, however, to discern whether a particular case of language intermingling counts as switching, mixing, or borrowing (see Kachru 1983:193–197; Thomason 2001:133–135; Sridhar and Sridhar 1980:408–409; Winford 2003:107–108). And it is particularly difficult to distinguish code-mixing from borrowing in cases where the two languages are related (see, for example, Woolard 1987:108). In this situation, the rules of thumb for discriminating mixings from borrowings fall short. Since the semantic ranges of the two languages overlap, fewer lexical gaps appear to be filled. Lexical and syntactic correspondences sometimes make it tough to decide where, exactly, a mix begins in a sentence. And phonological and morphological similarities work to “prenativize” terms from the guest language. 

On Tanna, the mingling of a vernacular and a Pidgin is a special case of mixing related languages. Bislama is an English-lexified Pidgin: ninety percent or so of its words are drawn from English. A significant portion of Bislama’s semantics, phonology, and grammar, however, mirrors that of the indigenous Austronesian languages of Vanuatu. Camden (1979:54), for example, has suggested that ninety percent of Bislama’s lexicon semantically parallels that of Tangoa, a language of central Vanuatu, and Bislama’s semantic overlap with Kwamera is probably somewhere in the same neighborhood. Similarly, discussing Tok Pisin of Papua New Guinea, Laycock (1966:45) notes:

There are few features of the simplified structure of Pidgin that are not found in most of the native languages, and with the exception of a few consonant clusters and frequent one--for-one consonant substitution there is little problem in phonology. 

In analyzing Bislama’s penetration of talk during Tanna debates, therefore, it is not always possible to determine whether a particular Pidgin item is a code-mixed or a borrowed form (especially as used by different speakers).

More recently, linguists such as Muysken (2000) have collapsed code-shifting, code-mixing, and also borrowing into more general language “mixing” processes, analyzing the three as functionally similar:  “what might be formally characterized as borrowed elements particularly in bilingual discourse take on certain discourse functions of code-mixing” (2000:69). Muysken discriminates three sorts of mixing (2000:3–4): insertion (elements from one language inserted into the matrix of a second); alternation (speakers switch from one language to another at possible “switch points”); and congruent lexicalization (combinations of elements from two languages that share a basic grammatical structure). Kwamera/Bislama code-mixing might be considered mostly an example of Muysken’s third type (congruent lexicalization) insofar as Bislama’s grammatical structure is basically that of a simplified Austronesian language. As is apparent below, however, specific Bislama mixes might be better understood as insertions into Kwamera, or as alternations between the languages.⁸

4. BISLAMA IN DEBATING. Tannese convene frequent meetings for settling disputes and making decisions. These take place at circular clearings in the forest where people also meet to dance, to exchange goods, and to prepare men’s daily draughts of the traditional Pacific drug, kava. Depending on the issue and on the personalities involved, a debate may attract a score to several hundred participants. Only mature men possess full rights of public speech. The contributors to the four debates analyzed here, for example, ranged from around 30 to 70 years in age. Most debaters were in their 40s and 50s. To make a point, these men stand and move into the clearing, while women and younger men in the audience remain silent, seated around the periphery. Generally, talk begins in mid morning and typically continues throughout the day. Debate culminates in late afternoon when—if all has gone well—the erstwhile disputants symbolize an achieved consensus by sharing kava (see Lindstrom 1981). 

As Bee notes above, Melanesians are partial to Pidgin when engaged in “pseudo-sophisticated” court cases, moots, village meetings, or debates.⁹ Put another way, the Bislama mixes and borrowings that typically characterize Tannese debate may serve overtly to mark a specific linguistic register: “debate talk.”¹⁰ The prevalence of code-mixing alone, however, does not distinguish a debate register from others. People use borrowed or mixed terms in virtually all speech situations on the island. Bislama forms pop up in island talk about any topic. Debates, however, are particularly useful places to look for Bislama’s presence and its effects in everyday talk. The conversational functions of Bislama code-mixes here are especially conspicuous as antagonists maneuver to score points and push consensus into a shape they can live with. A count of speakers’ use of Bislama words during the four debates is presented in Table 1.¹¹ The gross percentage of Bislama terms in oratory ranged from 2.9% to 5.2%. 

Table 1. Four Debates

Debate	Topic	Date	# Words	% Bislama
1	Rights to use exchange “road"	08/03/1982	4998	2.9
2	Marital discord	05/25/1982	4489	5.2
3	Land rights	06/02/1983	14,717	4.6
4	Family discord over death of child	06/15/1983	11,275	4.2

The problem of whether inlaid Bislama words are borrowed or code-mixed—touched on above—is made particularly acute by the structure of Bislama/Kwamera language mingling, and by the fact that most Bislama obtrusions into Kwamera sentences consist of single words. Studies of code-mixing elsewhere (of Spanglish, for example) have investigated sentences composed typically of more complex elements drawn from both languages involved. Pfaff (1979:296) provides examples of this sort:  

(1)	No van a bring it up in the meeting . . .
	‘They’re not going to . . . ’

Shifts from one language to another occur at different syntactic junctures, and one can calculate the relative frequency of various sorts of syntactic elements in a mixed sentence. Although elements from practically every syntactic category (including purely grammatical morphemes such as determiners) occur in code-mixed sentences, it has been found that certain types of elements are more likely to be mixed than others. In general, the higher the constituency of the element, the more likely it is to be mixed. Thus, conjoined sentences, main clauses, subordinate clauses (including relative clauses), noun phrases, verb phrases, and prepositional phrases, are among the most frequently mixed elements  (Sridhar and Sridhar 1980:409). In the less frequently observed occurrence of mixed single words, nouns outnumber adjectives, adverbs, verbs, and then miscellaneous grammatical items respectively (1980:409–410). 

Tannese patterns of code-mixing diverge from those of Spanish/English. Kwamera possesses a nativization device that readily incorporates single Bislama words and phrases alike. This is the verb -o (‘do/make’; see Muysken (2000:184–185) and Myers-Scollon (2002:134–137) for analyses of such “helping verbs” and similar insertion devices).¹² Almost any Bislama verb, adjective, as well as many nouns, can be grafted into a Kwamera sentence simply by introducing the word with -o with appropriate person, number, and aspect prefixes. Within Kwamera itself, -o + verb is an imperative structure, and secondary verbs take echo-subject prefixes:

(2)	kmiaha	ti-o	mha-vn
	2PL(EXCL)	FUT-do	ECHOSBJ13-go
	‘You all go away.’

Instances of Bislama mixes introduced by -o include (and, in examples that follow, I italicize -o, which takes various verbal affixes):

(3)	in ro time sai work ikn mata iraha hamo lafet raka
	‘it was time to work but they had already started (made) a party’

(4)	iako against ia kmiaha
	‘I am against you all’

(5)	o ro action riti sai custom rknekn
	‘make a strong, customary action (response)’

(6)	ikamo a fool ia nermama
	‘you make a fool out of the people’

(7)	tiapwah noien hurry-up hurry-up ia nagkiariien
	‘don’t hurry up the talk’

(8)	rule in ro important
	‘the rule is important’

(9)	in ro a think iako pass naha ia kwopn
	‘he thinks I pass by that place’

(10)	iko trouble saik, rier ianha i ia public?
	‘you make your trouble, and it comes out like that in public?’

(11)	irouni mua tirouo pushem-out ianrak
	‘you two say that you will push me out’

Approximately a third of the total number of Bislama forms in the four debate texts are introduced into Kwamera sentences by means of -o. Consequently, code-mixing on the island differs structurally from that of some other language pairings. Contra Sridhar and Sridhar (1980:409–410), Tannese speakers mix many more single words than they do complex syntactic elements; and they mix a far greater proportion of alien verbs vis-à-vis nouns than has been found to be typical elsewhere (see, for example, Poplack 1979:45).¹⁴ 

In addition to the prevalence of -o + mix, Bislama/Kwamera mixing differs structurally from Spanish/English for example because, on Tanna, people are code-mixing a vernacular and a Pidgin, rather than two independent (if related) languages. Woolford, in an early analysis of syntactic constraints on the code-mixing of English and Spanish, suggested that “Lexical items can be freely drawn from either language to fill terminal nodes created by phrase structure rules common to both languages” (1983:535)—this is what Muysken would later call congruent lexicalization. If this is true, then a substantial overlap in the syntax of vernacular Vanuatu languages and Bislama may explain the prevalence of single-word mixing on Tanna. Given phrase-structure rule parallels in the two codes, speakers can readily insert single Bislama terms almost anywhere in their Kwamera sentences.

Because of the large numbers of single-word mixes, one could suppose that all Bislama terms in the debate corpus are borrowings: that Kwamera and Bislama have already fused into a single language and that the lexicon of this expanded language is composed of both Bislama and Kwamera words, many of which are now synonyms. Suspicion that the linguistic boundaries between the two have dissolved is fed by the fact that whereas people freely intermingle Kwamera and Bislama, they do not code-mix terms from the other island languages into their talk. Many of the participants at these debates were either actively or passively multilingual. The typical practice at debates is for a participant to use his own language, presuming (correctly) that this will be understood by the leading members of his audience. Although Islanders often debate in more than one island language, they are careful to maintain local linguistic boundaries. During the four transcribed debates, although all participants scrambled numerous Bislama items into their orations, only one mixing of an item from a neighboring language occurred. A man mixed a possessive form from the mother tongue of his mother and wife. He immediately caught and corrected his mistake: 

(12)	iakamuvahi pen navahagien min,	first-born *rahan...a savani, ruvahi saiou nagkiariien
	‘I gave advice to him,	his first-born . . . uh his, he took my advice’

If Bislama and Kwamera have somehow merged into an expanded code (at least at the lexical level), the appearance of a Bislama-derived term in a sentence would not carry any wider semantic load or evoke any inferential meaning, as it might in true code-mixing. Although the range of its connotations perhaps differs from that of other words a speaker could have chosen, interlocutors would not remark the fact that a shift to a different code had taken place. They would perceive only the ordinary flow of talk within a single code. 

Certainly, based on frequency of use, we can suspect that many Bislama words in the debate texts are borrowings. In the four recorded debates, speakers used 272 different Bislama terms.¹⁵ Some words were said only once. Others were much more prevalent. Table 2 sets forth the most common mixed Bislama items, ranked in order of frequency of use.¹⁶

Table 2. Common Bislama Borrowings/Mixes

Term/Number of Uses
trabol ‘trouble’/59	panis ‘punish’/34	traeb ‘tribe’/25
ting ‘think’/58	lo ‘law’/33	brata ‘brother’/24
kastom ‘custom’/58	rong ‘wrong’/32	sore ‘sorry’/22
taem ‘time’/46	wok ‘work’/31	ansa ‘answer’/21
agens ‘against’/41	tauien ‘brother-in-law’/30	poen ‘point’/20
saed ‘side’/39	paoa ‘power’/27
fasin ‘fashion’/35	raet ‘right’/25

Table 3 lists terms that, although less frequently spoken, were used at least once in all four debates. The fact that some of these Bislama words fill lexical gaps in Kwamera is additional evidence for borrowing (e.g., custom, law, work, tribe). 

Table 3. Bislama Terms Used in All Four Debates

agri ‘agree’	pas ‘pass’	trabol ‘trouble’
finis ‘finish’	raet ‘right’	wan ‘one’
kastom ‘custom’	rong ‘wrong’	wantaem ‘one-time’
lo ‘law’	saed ‘side’	wei ‘way’
panis ‘punish’	taem ‘time’	wok ‘work’
paoa ‘power'	ting ‘think’

Many of these words, however, do have good Kwamera synonyms. I suspect that these and other less common Bislama intrusions were code-mixings rather than borrowings. As such, shifts to Bislama during debate did serve several of the conversational functions that have been suggested for code-mixing. The audience during a debate would have noted that a code-mix had occurred, and could have inferred any metameanings signaled thereby.¹⁷

Aside from the diffuse, register-marking function of Pidgin mixes and borrowings in Melanesian debate situations, early analyses of code-mixing located several common conversational functions such as signaling family and ethnic connections (see Gumperz 1982:75–84; Kachru 1983:197–198). Later work, however, has suggested that code-mixing, where it is unmarked and expected, may in fact add no further meaning to what people are saying (Myers-Scotton 1993) or that speakers code-mix in order to generate ambiguously pitched utterances that hearers can construe as they may (Stroud 1992). In these debates, I return to earlier functionalist accountings of code-mixing, guessing that Bislama mixes within island oratory were heard in fairly obvious ways. Bislama code-mixes, were therefore “marked,” in Myers-Scotton’s (1993) terms, and therefore intended by speakers to be remarked by hearers. 

I. Interjection

(13)	iakni ahavn nagkiariien saik, sorry
	‘I interrupted your talk, sorry’

(14)	please, apwah nagkiariien
	‘please, stop talking’

(15)	alright, kmrhi judge me
	‘all right, appoint some judges’

(16)	irouarari mwi ia story, na okay
	‘you two turn back to the story, okay’

(17)	yes, ikamatui nukwanek
	‘yes, you are looking after my head (gave me knowledge)’

II. Reiteration (see Kulick 1992:77)

(18)	parov, sorry
	‘sorry, sorry’

(19)	government rni mua law ro standem-up, ia law ramarer
	‘the government said for the law to be established, the law established’

(20)	law kwatia sai nermama pam anan tuo followem i, nermama pam anan tuakurira i
	‘one law for all men to follow, all men to follow’

(21)	sometime, npn riti
	‘sometime, sometime’

(22)	tukwmua ik iermama parhien, ik real man ikata takata ik ia kwopin u
	‘if you are a real man, a real man, I’ll see you at this place’

(23)	ia freedom ua constitution saktaha rfni, i mean what?
	‘our freedom or constitution means what, means what?’

(24)	iako true mua nparhien nagkiariien nah
	‘I say true, say true talk there’

(25)	uncle, kwanien
	‘nephew, nephew’18

(26)	irau kroueiuaiu, ro deep ia story
	‘they two descended, descended into the story (gave detailed evidence to back a  point)’

(27)	krouavan outside, krau ia nakwai tsi u
	‘we two go outside, we are out to sea, here’

III. Reported Speech

(28)	in rfurkurn mwi iamni mua “next time last one savani?”
	‘how will he know again to say “next time is its last?”’

(29)	Kauke ragkiari, iakua “hem right”
	‘(if) Kauke speaks, I say “he’s right”’

(30)	Npn makwa ro sampam iakni pen t nah Jeffrey iakua “Jeffrey, attention”
	‘When the month is over I say to Jeffrey, “Jeffrey, pay attention” (to menstrual period sexual intercourse taboos)’

IV. Neutralization  

(31)	ia side sai family planning
	‘the issue of family planning’ (neutralizing a reference to traditional postpartum sexual taboos improper to mention in mixed company)'

(32)	rski eraha irau, mo bastard ianrau
	‘he swore at us two, did bastard at us’ (speaker avoids recitation of Kwamera taboo words by use of Bislama “doing bastard”)

V. Message Qualification

(33)	nteta savani rataka twotime
	‘his canoe (lineage) were refugees twice’

(34)	na riti mwi, number-two
	‘and another thing, number-two’

(35)	riuan naha ia kafk kapwa, iapko mha think
	‘there’s nothing in my head, I’ve no thought’

(36)	first start, iakamuvehe men naha puta
	‘at the beginning, I came and went inland’

(37)	time a Tuesday, sarakure naha pen naha mharni ratukwatukw
	‘on Tuesday, we three sat there speaking correctly’

Code-mixing allows debaters to shade the meaning of what they say. A jump to a second code in mid-sentence foregrounds wider assumptions about the two codes themselves, and this adds highlights to the message. In many similar code-mixing situations, one language is the local mother tongue; the other is a national standard or lingua franca. Early analyses of mixing noted that shifts into home codes evoke local associations and solidarities, while shifts to nonlocal languages index the larger, national arena and outside authority. Gumperz put this in terms of us and them: A speaker can choose to phrase his or her words in either a “we-code” or a “they-code” (1982:66). Closer to Vanuatu, Sankoff argued that in Papua New Guinea “use of Tok Pisin is regarded as appropriate for people of power and authority, in contexts having a relationship to the broader colonial society, especially the domains of business and government” (1980:44; see also Sankoff 1976: 303–304). 

Figuring “we code/they code” as something like “local solidarity/outside authority,” however, fails to make sense of many of the code-mixes produced by Tannese debaters.¹⁹ “Us/them” connotations are perhaps particularly problematic when the “them” language is a Pidgin. Evaluations of the social meanings and import of Bislama and other Melanesian Pidgins are contradictory. On the one hand, Pidgin has been denigrated as the bastard tongue of colonialism and of exploitative plantation agriculture. On the other, it has been celebrated as the voice of the people, a triumph of linguistic creativity in the face of adverse circumstance. These ambiguous evaluations reflect islander sensibilities as well. Sometimes, Pidgin is “white man’s language” (Sankoff 1976:304); other times it counts as local lingo. 

On Tanna, my sense is that Bislama is often more of a “we” than a “they” code (and today is less associated with Europeans than is Papua New Guinea’s Tok Pisin (see Kulick (1992:84)). Code-mixing on the island involves two kinds of “home” language. Children, nowadays, learn Bislama from their older siblings and peers. Acquiring Bislama begins in play at home and, informally, at school. Radio transmissions and popular string-band songs assist language learning. A young man or woman often acquires the language fully when he or she travels to spend a few months with relatives in Port Vila, the nation’s capital. If Bislama is not anyone’s mother tongue in southeast Tanna, it perhaps counts at least as an elder brother or sister tongue (as might Pidgins elsewhere; see Myers-Scotton 1993:72). 

 In important respects, Bislama’s “we” connotations are broader than those of Kwamera’s. Geographically and socially, Kwamera as a “we-code” is narrowly bounded. It contrasts with the four other major languages on the island, evoking cribbed social boundaries. The “we” of metropolitan Bislama, on the other hand, can encompass the entire island and nation. At the same time, Bislama does retain aspects of a “they-code.” Pidgin is a baggy, flexible language with a number of sociolects (see Mühlhäusler 1979b on Tok Pisin sociolects). Bislama absorbs English forms as easily as Kwamera does Bislama. When Tannese debaters wish to index external authority (of government, business, or religion), they shift not to ordinary Bislama but to a heavily anglicized form of Pidgin. 

(38)	iaknatarig raka, backward, forward-and-backward, mua tikuasi iermama ia right.
	‘I’ve already considered, backward, forward-and-backward, whether you beat your wife with justification’

(39)	tihatarig, be careful, hiatarig one-by-one, tio lookout . . . straight constitution havahi pehe m kmrau
	‘you all consider, be careful, consider one-by-one, you all lookout…straight constitution comes for us two (if we don’t behave, we must face the law)’

(40)	censure, iko amasan lookout
	‘censure (danger), watch it’

(41)	samagkiari takwtakwnu ia side savai chief me, mne custom. Custom  ravahi majority t noien nari riti mwi
	‘we are talking now under the side of the chiefs, and custom. Custom carries the majority of (rules) behavior in this area’

(42)	ruasi ia spiritual ua ruasi ia physical? Tirouni mua ro clear
	‘does he beat (her) symbolically or physically? You two explain so that it is clear’

(43)	iakokeikei mwipuk more than naruk
	‘I love my grandchild more than my child’

(44)	iakua ro horrible—nua kwanakwevur in treiuaiu t young fellow. Rpiko mha ikata, mata ro sound reraha.
	‘I say he is horrible—and that the greyhead (old man) has lost to the young fellow. It isn’t as you see, rather it (his argument) sounds bad.’

Items such as forward-and-backward, one-by-one, be careful, constitution, censure, spiritual, physical, more than, horrible, and majority are uncommon in rural Bislama. But they might well be heard in the Bislama of politicians, bureaucrats, and preachers. In these examples, shifts to an urban/official sociolect of Bislama do appear to cue an external, authoritative voice. In examples (38) through (41), speakers are warning others to behave themselves (cf. Kulick (1992) on people’s use of Tok Pisin within angry outbursts and harangues). Melanesian personal autonomy, achieved leadership, and so forth make “warnings” an often contentious type of speech event. Warners are concerned to justify their right to admonish others. In this regard, code-mixes of “official” Bislama terms may cue a speaker’s personal associations with governmental or religious authorities (as a chief, pastor, member of a local government council, and so on). This foregrounds his perhaps painful but unavoidable duty to caution others (cf. Gumperz (1987:92)). 

Examples (42) through (44) are personal criticisms. In (42) and (43), a local Seventh-day Adventist pastor criticized the actions of a wife-beater. He censured an overly enthusiastic bout of wife abuse and the neglect of a grandchild. In (44), a younger man derided an older leader for telling a pack of lies about rights to a land plot, suggesting that he is past his prime. Again, code shifts to official Bislama contextualized this obloquy by indexing external standards of behavior (“Christian” and “modern,” respectively). In these cases, as with Sankoff’s big-man orator, Pidgin served as a “they-code.” 

Actually, the Tannese have a fairly broad choice of “they-codes” they could shift to, including the other languages of the island and, for some, English and French. Bislama’s semantic value as a mixed code, however, stems from its generality and neutrality. Because of a close connection of language, place, and identity on Tanna (Lindstrom 1983b), people ordinarily avoid code-mixing forms from neighboring languages, or even from other micro-dialects of Kwamera (see example (12) above). Bislama’s “theyness,” however, does not carry along with it narrow, regional associations. The same neutrality was noted above, with regards to its “we-ness.” As a nonlocalized lingua franca, Bislama both belongs and does not belong to everyone. 

In other cases, debaters use the “we” connotations of Bislama. In addition to formal political or religious speech events on the island, men commonly shift to Pidgin when joking, drinking, playing football—situations of male camaraderie, solidarity, and friendly competition. Debaters, code-mixing this home-style Bislama, can thereby index solidarity (or, sarcastically, metacomment on its absence). 

(45)	my word, takousi saiou pranema takousi mousi apune!
	‘my word, I had beaten my wife (with that), I would have killed her!’

(46)	e, my friend, ko ipko mha anan nari riti, ikni auar a tekn. Pwah nermama in ro business savani.
	‘Hey my friend, if you don’t (understand) something, you only talk about its skin (you lack knowledge). Let people mind their own business.’

(47)	mata kmaha one, saiou one parhien
	‘but we are one, my one (group) truly’

(48)	mata tprena naha, suvni raka how many time? Two time finish, ua? Muvni raka no sampam.
	‘but that land, we already debated it how many times? Twice already, right? We’ve already concluded the debate.’

(49)	(49) mua irouo t company nagkiariien ianha ira, saren, iako cranky ia nagkiariien mhara cranky ia nagkiariien.
	‘but you two arrange a large debate like this, we proceed, and I’m crazy with debate, crazed by the debate.’

(50)	thank you, number-one anan nagkiariien
	‘thank you, a first rate speech’

(51)	takwtakwnu ikara meva mwi ianirak. Wanem thing olsem? I no gat head olsem!
	‘now you are stomping on me again (in debate). Why are you doing this? It doesn’t make sense!’

In these examples, shifts to the local (rather than the urban-official) sociolect of Bislama served to evoke community relations, sometimes to emphasize the fact that local solidarity was under threat. In examples (45) and (46), a debater sought the empathy of his audience, asking it to discount the point of an antagonist. Example (47) specifically commented on group solidarity. Examples (48) through (51) indexed the fact that ordinary neighborhood sociality was being undermined by the debate itself—debate that in part should have served to reestablish this. (Example (50) was sarcasm.)

In sum, because it is a Pidgin, Bislama can be both more “we” and more “they” than Kwamera. Speakers may shift from Kwamera to Bislama to evoke male solidarity or to deplore it absence; and they may shift to an anglicized Bislama to cue external authority and to remark invidious contrasts between the national center and local community. Kwamera itself as a “we-code” and the other island languages as “they-codes,” conversely, are semantically more constrained, insofar as the social unities and boundaries they evoke are entirely local. 

5. LANGUAGE CHANGE. Bislama today peppers people’s everyday Kwamera. But is this ubiquitous code-mixing leading to language convergence (Muysken 2000:122; Ramat 1995:61), or even language death as people shift over to the in-mixed code, as Kulick (1992) predicted as the likely, dismal future for Taiap in Papua New Guinea? Constant mixing, over time, may lead to nativization and borrowing if frequent appearance of a Bislama term in special code-mixing speech situations dissolves into general usage in all contexts. Here, adopted Bislama forms come to replace or supplement original Kwamera material and people, eventually, would find themselves speaking more and more Bislama (and less and less Kwamera). Many have supposed that substantial mixing may indeed spark major linguistic change: “the cumulative effect of mixing may eventually result in distinct varieties of a language” (Kachru 1983:203; see also Marasigan 1983:57; but cf. Muysken 2000:269). 

In addition to modifying a language’s lexicon, code-mixing and attendant borrowing may induce phonological and syntactic changes as well. Several studies have identified the mutational effects of Pidgin on Melanesian vernaculars. Laycock and Wurm (1977), for example, document morphological as well as lexical/semantic modifications that they trace to Pidgin’s impact on a number of languages of Papua New Guinea (see also Mühlhäusler 1979a:160). In that a Pidgin—as a Pidgin—is in general semantically, phonologically, and syntactically simpler than a vernacular, most of these changes are reductions and simplifications (Scott 1979). 

What of Kwamera? Watt’s New Testament, which provides evidence of the state of Kwamera in the second half of the nineteenth century, suggests that linguistic change here has involved the language’s lexical/semantic systems only. Unlike the case of several Indian codes (described by Kachru 1983), or that of “Mixmix” of the Philippines (Marasigan 1983), extensive code-mixing and borrowing have not produced a distinct variety. And unlike the case of several Papuan languages, a century and more of code-mixing on Tanna has not significantly altered Kwamera’s phonology, morphology, or syntax. Confinement of the impact of code-mixing and borrowing to Kwamera’s lexicon no doubt relates to the peculiar effects of a code-mixed Pidgin: an obtrusive guest code that shares basic phonological and grammatical structures with the host. 

5.1 PHONOLOGICAL CHANGE. Code-mixing and subsequent borrowing can transform the phonological system of one or both codes involved. Grace, for example, to explain phonological complexity in New Caledonia languages, has argued that there is an island-wide phonological system of which different languages possess various bits and pieces; “Borrowing will thus tend to increase the phoneme inventory of the language” (1981:267). Code-mixing, as proto-borrowing, may have the same effect: “The more frequent the use of code switching strategies, the greater the amount of phonetic overlap between the two contrasting codes” (Gumperz 1982:56–57). 

Where one code is a Pidgin, however, the situation differs significantly. Instead of both codes having balanced phonological effects upon one another, Kwamera’s phonological interference upon Bislama is far greater than Bislama’s on Kwamera. There are multiple regionalects of Bislama in Vanuatu, each shaped by local vernacular structures. People use local phonology to nativize their Pidgin (see Tryon 1979). 

In cases where marked phonological differences between Pidgin and island vernaculars are maintained, and Pidgin is not nativized, its phonological impact may be more far reaching. Laycock (1966:46), for example, wondered if the more common five-vowel system of Tok Pisin in Papua New Guinea might eventually transform, by analogic leveling, a more complex Abelam vowel system. In the case of Kwamera’s six vowels, an extra // phoneme merely interferes with how Bislama is locally pronounced. And although Kwamera possesses several more consonants than Bislama (labialized /kw/, /mw/, /fw/ and voiceless /mh/, /nh/ and /ŋh/), the consonantal sets of the two codes overlap without much discord. (Kwamera, like Bis1ama, also possesses /k/, /m/, /f/, /n/, and /ŋ/.) 

Standard Bislama does include one consonant that is absent in Kwamera: liquid /l/ (cf. Bee 1972:75). Although speakers may produce [l] when speaking Bislama, this does not carry far back into Kwamera. An [l]/[r] distinction is one of the main symbolic tokens by which people differentiate Kwamera from the larger neighboring language of east Tanna (Lindstrom 1983b). Many people are multilingual and are already competent with [l], in that this is a phoneme in most of the other languages of the island. Because lack of [l] is a recognized mark of linguistic distinctiveness, Kwamera speakers are motivated to police its appearance in their language, as they are motivated to avoid code-mixing words from neighboring languages (example (12) above). This mitigates any potential effects that Bislama’s [l] might have for Kwamera. The nineteenth-century phonological system evident in Watt’s translations is still current today.

5.2 MORPHOLOGICAL AND SYNTACTIC CHANGE. Kwamera’s morphological and syntactic systems are also conservative, at least judged in the light of Watt’s text.²⁰  Elsewhere, Pidgin has had a grammatical impact upon some local vernaculars, as Kulick (1992) and others have noted. Laycock and Wurm describe several languages in which Pidgin borrowings have induced “decay in morphological complexity, affecting especially the verb complex, noun classification and numeral systems” (1977:196). In one case, however—that of Buang—they note that the influence of Pidgin “does not appear to have resulted in a simplification of morphology or syntax—perhaps because Buang is an Austronesian language whose morphology and syntax is [sic] largely comparable with that of Pidgin” (1977:199). This is also the case for Kwamera.

The inconsequence of more than a century of Bislama borrowings and mixings upon Kwamera grammar might also be explained in terms of morphological and syntactical parallels between the two codes (see Camden 1979). Where Bislama and Kwamera phrase structure rules overlap, for example, Bislama forms drop naturally into Kwamera sentences, including comparatively unusual pronoun mixes (54). 

(52)	iemanmi u krouavahi back ira
	‘the two men here gave it back’

(53)	ramrerig back mwi
	‘returning again’

(54)	hem rani parhien nagkiariien naha
	‘he speaks truly’

(55)	matipen fwe inside ia nteta
	‘look into the inside of the canoe’

(56)	time nah tikni nagkiariien tukwe
	‘when you will talk about it’

(57)	time iroue rouakure t nkava
	‘when they two went to sit down for kava’

Bislama back (52 and 53) directly substitutes for the vernacular deictic pehe (‘towards speaker or hearer’). Bislama hem (54) replaces Kwamera third person singular subject pronoun in. Bislama inside (55) stands in for Kwamera nakwa- (‘mouth, inside’).²¹ Finally, Bislama time (56 and 57) substitutes directly for Kwamera npn (‘night, point in time’).²²

Cases where Bislama differs morphologically or syntactically from Kwamera can be handled by the -o nativization device. This absorbs Bislama forms while, syntactically and morphologically, it neutralizes their incompatibility. 

(58)	iarno pam through naha ikn kmrau Rapi
	‘Rapi and I are through with that (we two already do all through that place with Rapi)’

(59)	iko without agreement
	‘you do it without an agreement’

Unlike the previous examples, (58) and (59) are not directly translatable back into Kwamera. The grammatical disagreement here, however, is buffered by -o. This opens a window in Kwamera syntax for Bislama constructions. This device protects Kwamera syntax and morphology from potential simplification and other effects of Bislama mixings. 

5.3 LEXICAL AND SEMANTIC CHANGE. The major impact of Bislama mixings and borrowings on Kwamera, as on other Melanesian vernaculars (Laycock and Wurm 1977:196; Laycock 1966; Scott 1979), is concentrated upon native vocabulary and semantics. Over the past 150 years, changes in these areas have far outweighed modification in Kwamera phonology and grammar. 

Pidgin terms, where they come to replace vernacular lexemes, often entail semantic simplification. Should Kwamera speakers all adopt the generic Bislama verb carry [karem], for example, this could replace twelve vernacular lexemes that describe various modes of carrying. Some simplification in Kwamera vocabulary has occurred. The traditional numerical system is quinary, and Bislama terms today replace numbers greater than five. This is simplification of a sort in that, for example, eleven is used in place of kirirum kirirum kwatia (‘five five one’). Other replacements are less simplifications than they are semantic shifts. Bislama’s set of solar months has pushed back a traditional system of lunar calculation into distant memory. 

Bislama’s impact on Kwamera’s vocabulary, however, has been mostly one of lexical enlargement rather than replacement or simplification. One indication of this is the fact that only a few Kwamera words, used in Watt’s 1890 New Testament, are today unknown or archaic. Instead, Bislama borrowed terms have contributed to island vocabulary by filling lexical gaps and by naming introduced objects or abstract notions. Back-translations from Bislama also enlarge Kwamera semantics (cf. Watson-Gegeo 1987). As is typical throughout the country, Kwamera speakers now greet one another with ramasan ianpnpn (‘good morning’), and so on. They talk about hot peppered foods as apwanapwan (‘hot, i.e., in temperature’). One instant calque of this sort appeared in debate 2. After lecturing about the necessity to think “forward-and-backward” (see (38)), the speaker back translated into Kwamera: 

(60)	takatarig raka kupwn-kurira
	‘I will have already considered front-back’

Bislama, whether borrowed or code-mixed, provides useful supplemental linguistic resources. The ability to code-mix itself, of course, is important semantic capital. The Tannese enliven their debates with metaphor and figurative speech (Lindstrom 1983a). Bislama, in its “we-code” aspect, is rich in playful material of this sort.

(61)	radio no a go-ahead ia kwopn u
	‘radio (idle talk) only is going ahead here’

(62)	iakreirei mua tranan paku mo anchor paku. Ia nakwai tsi iti ua nah? Traman nmatagi trosi afi mast.
	‘If it goes on like this I don’t know where I’ll anchor. In the middle of the sea? If it goes on, the wind will break off the mast (complaint of a man losing a claim to a land plot)’

(63)	hnata pam raka corner saim
	‘they’ve all already seen your corners (crookedness)’

(64)	saik hat mama takwtakwnu
	‘your hat (knowledge) is going (ruling) now’

Although Kwamera has experienced some lexical and semantic reduction and simplification, Bislama borrowing and code-mixings in general have enriched the language.

6. CONCLUSION. Melanesians, as Grace (1981) and others have pointed out, may be particularly ardent borrowers, at least at the lexical level. Borrowings serve a number of conversational functions, such as the evasion of newly tabooed words, a demonstration of personal exotic knowledge, or the implication of a metamessage. These same conversational functions may account also for practices of code-mixing; and code-mixing—by popularizing a foreign item—can be the mother of borrowing. 

The emergence of Melanesian Pidgins, over the past 150 years, as important, nonlocalized prestigious codes with both “they” and “we” implications—has increased practices of both code-mixing and borrowing throughout the region. When code-mixing and borrowing become extensive, entirely new linguistic varieties may emerge. At the least, code-mixing can induce phonological leveling as well as morphological and syntactic mutations in one or both of the languages involved as vernaculars lose their local linguistic vitality.

This has not happened on Tanna. Kwamera, over the last century and a half at least, has been a conservative language. I have suggested that the stability of Kwamera in the face of widespread code-mixing is due to the peculiar sort of code-mixing situation at work here. This is one of a vernacular and a Pidgin—a guest code whose phonological, semantic, and grammatical systems resemble, to a large degree, those of the host vernacular. Kwamera’s nativization device -o, in addition, provides an opening in native syntax for Bislama forms that are grammatically discordant. Crowley, surveying the linguistic futures of all of Vanuatu’s extant vernaculars, was similarly optimistic:  “... in no case is any indigenous language in any obvious immediate danger of being replaced by Bislama or either of the metropolitan languages. Wherever language shift is under way, it is always some other local language that is the replacing language, and not one of the national lingua francas” (2000:125).

During the last century, Bislama’s principal impact upon Kwamera has been to enlarge its lexicon—contributing words that fill lexical gaps, that label imported items, and that serve as useful or pleasing synonyms. More important, Bislama itself as a valuable, nonlocal “they-code” and “we-code” alike, allows code-mixing debaters to inflect and to give greater texture to the meaning of what they say. 

References

Alpher, Barry, and David Nash. 1999. Lexical replacement and cognate equilibrium in Australia. Australian Journal of Linguistics 19(1):5–56.

Bee, Darlene.1972. Phonological interference between Usarufa and Pidgin English. Kivung 5(2):69–95. 

Camden, William. 1979. Parallels in structure of lexicon and syntax between New Hebrides Bislama and the South Santo language as spoken at Tangoa. In Papers in Pidgin and Creole linguistics, no. 2, ed. by Peter Mühlhäusler et al., 51–117. Pacific Linguistics Series A-57. Canberra: Australian National University. 

Crowley, Terry. 1990. Beach-la-Mar to Bislama: The emergence of a national language in Vanuatu. Oxford: Clarendon Press.

Crowley, Terry. 2000. The language situation in Vanuatu. Current Issues in Language Planning 1(1):47–132.

Crowley, Terry. 2004. Bislama reference grammar. Oceanic Linguistics Special Publication No. 31. Honolulu: University of Hawai‘i Press.

Ennaji, Moha. 2005. Multilingualism, cultural identity, and education in Morocco. New York: Springer.

Grace, George W. 1981. Indirect inheritance and the aberrant Melanesian languages. In Studies in Pacific Languages and Cultures, ed. by Jim Hollyman and Andrew Pawley, 255–268. Auckland: Linguistic Society of New Zealand. 

Gumperz, John J. 1982. Discourse strategies. Cambridge: Cambridge University Press. 

Kachru, Braj B. 1983. The Indianization of English. Oxford: Oxford University Press. 

Keesing, Roger M., and Jonathan Fifi’i. 1969. Kwaio word tabooing in its cultural context. Journal of the Polynesian Society 78(2):154–177.

Kulick, Don. 1992. Language shift and cultural reproduction: Socialization, self, and syncretism in a Papua New Guinea village. Cambridge: Cambridge University Press.

Laycock, Donald C. 1966. Papuans and Pidgin: Aspects of bilingualism in New Guinea. Te Reo 9:44–51. 

Laycock, Donald C., and Stephen A. Wurm. 1977. Observations on language change in parts of the New Guinea area. In New Guinea area languages and language study, vol. 3, ed. by Stephen A. Wurm, 195–205. Pacific Linguistics Series C-40. Canberra: Australian National University. 

Lindstrom, Lamont. 1981. Speech and kava on Tanna. In Vanuatu: Politics, economics and ritual in island Melanesia, ed. by Michael Allen, pp. 379–398. Sydney: Academic Press. 

Lindstrom, Lamont. 1983a. Metaphors of debate on Tanna. Naika: Journal of the Vanuatu Natural Science Society 12:6–9. 

Lindstrom, Lamont. 1983b. Say what? Language and political boundaries on Tanna (Vanuatu). Anthropological Linguistics 25(4):387–403. 

Lindstrom, Lamont. 1985. The politics of dictionary making on Tanna, Vanuatu. In Austronesian linguistics at the 15^th Pacific Science Congress, ed. by Andrew Pawley and Lois Carrington, 329–341. Pacific Linguistics Series C-88. Canberra: Australian National University. 

Lindstrom, Lamont. 1986. Kwamera dictionary/Nkukua sai nagkiariien Nninfe. Pacific Linguistics Series C-95. Canberra: Australian National University.  

Lindstrom, Lamont, and John Lynch. 1994. Kwamera. Munich: Lincom Europa.

Lynch, John, and Terry Crowley. 2001. Languages of Vanuatu: A new survey and bibliography. Pacific Linguistics 517. Canberra: Research School of Pacific and Asian Studies, The Australian National University.

MacDonald, Donald.1891. South Sea languages, vol. 2. Melbourne: Trustees of the Public Library, Museums and National Gallery of Victoria. 

Marasigan, Elizabeth. 1983. Code-switching and code-mixing in multilingual societies. Singapore: Singapore University Press. 

Meyerhoff, Miriam. 2001. Another look at the typology of serial verb constructions: The grammaticalization of temporary relations in Bislama (Vanuatu). Oceanic Linguistics 40(2):247–268.

Meyerhoff, Miriam. 2003. Formal and cultural constraints on optional objects in Bislama. Language Variation and Change 14(3):323–346.

Mufwene, Salikoko S. 2001. The ecology of language evolution. Cambridge: Cambridge University Press.

Mühlhäusler, Peter. 1979a. Code switching in Papua New Guinea: Local languages versus New Guinea Pidgin, Hiri Motu and English. In New Guinea and neighboring areas: A sociolinguistic laboratory, ed. by Stephen A. Wurm, 157–176. The Hague: Mouton. 

Mühlhäusler, Peter. 1979b. Sociolects in New Guinea Pidgin. In New Guinea and neighboring areas: A sociolinguistic laboratory, ed. by Stephen A. Wurm, 225–242. The Hague: Mouton. 

Muysken, Pieter. 2000. Bilingual speech: A typology of code-mixing. Cambridge: Cambridge University Press.

Myers-Scotton, Carol. 1993. Social motivations for codeswitching: Evidence from Africa. Oxford: Clarendon Press.

Myers-Scotton, Carol. 2002. Contact linguistics: Bilingual encounters and grammatical outcomes. Oxford: Oxford University Press.

Pfaff, Carol. W. 1979. Constraints on language mixing: Intrasentential code-switching and borrowing in Spanish/English. Language 55(2):291–318. 

Poplack, Shana. 1979. Sometimes I’ll start a sentence in Spanish y termino en Espanol: Toward a typology of code-switching. Working Papers in the Center for Puerto Rican Studies No.4. New York: City University of New York. 

Ramat, Anna Giacalone. 1995. Code-switching in the context of dialect/standard language relations. In One speaker, two languages: Cross-disciplinary perspectives on code-switching, ed. by Lesley Milroy and Pieter Muysken, 45–67. Cambridge: Cambridge University Press.

Salisbury, R. F. 1962. Notes on bilingualism and linguistic change in New Guinea. Anthropological Linguistics 4(7):1–13. 

Sankoff, Gillian. 1976. Political power and linguistic inequality in Papua New Guinea. In Language and Politics, ed. by William O’Barr and Jean F. O’Barr, 284–310. The Hague: Mouton. 

Sankoff, Gillian. 1980. Language use in multilingual societies: Some alternative approaches. In The social life of languages, ed. by Gillian Sankoff, 29–46. Philadelphia: University of Pennsylvania Press. 

Scott, Graham. 1979. Lexical expansion in a non-Austronesian language of Papua New Guinea. In New Guinea and neighboring areas: A sociolinguistic laboratory, ed. by Stephen A. Wurm, 101–114. The Hague: Mouton. 

Sridhar, S. N., and Kamal K. Sridhar. 1980. The syntax and psycholinguistics of bilingual code mixing. Canadian Journal of Psychology 34(4):407–416. 

Stroud, Christopher. 1992. The problem of intention and meaning in code-switching. Text 12(1):127–155.

Thomason, Sarah G. 2001. Language contact: An introduction. Washington, DC: Georgetown University Press.

Tryon, Darrell T. 1979. Bichelamar (New Hebrides). In New Guinea and neighboring areas: A sociolinguistic laboratory, ed. by Stephen A. Wurm, 77–97. The Hague: Mouton. 

Watson-Gegeo, Karen Ann. 1987. English in the Solomon Islands. World Englishes 6(1):21–32.

Watt, William. 1890. Nagkirien ruvani sumun savi savei Yerumanu saketaha ketir ramavahi urumu ketaha. Glasgow: Robert Maclehose.

Watt, William. 1919. Naresien em nupume ya nafwakien ya nagkirien Kwamera. Melbourne: Brown, Prior. 

Winford, Donald. 2003. An introduction to contact linguistics. Oxford: Blackwell Publishing.

Woolard, Kathryn A. 1987. Codeswitching and comedy in Catalonia. IPRA Papers in Pragmatics 1(1):106–122. 

Woolford, Ellen. 1983. Bilingual code-switching and syntactic theory. Linguistic Inquiry 14(3):520–536. 

Lamont Lindstrom
lamont-lindstrom at utulsa dot edu

Attribution Non-Commercial No Derivatives License