3.1. Context of this study and the methods

The results reported here are based on our fieldwork in MLP over the past six years, of approximately six weeks to two months each year. The goals of this fieldwork are language documentation for Chorote, Nivaclé, and Wichí, to produce a grammar (both a practical version for the community and a technical one for scholars) and dictionary of each of the three languages (again in both a practical and a more technical version), with recording (audio and video), analysis, and archiving of many texts of many different genres. We also work with several community-based programs for preparation of educational materials for instruction in the three languages. We have trained several native speakers of each of the three indigenous languages as members of the research team; they have learned to read and write their languages, to use recording equipment and to interview and record other speakers from MLP and nearby areas, and to help with the analysis and the transcription and translation of the texts. They are integral members of the documentation and revitalization project for these languages, and continue their work when we are not in the community. We have worked with and recorded over 80 adults, who have provided oral histories, traditional tales, and other texts. We have had the opportunity to get to know personally most of the adults and a very large number of the children in the community through family connections and participation in various activities there. We have observed (and often recorded) multilingual conversations on nearly a daily basis, in many contexts, in particular among members of the research team and their families, but also in many families in many different work and social settings.

Our methods for addressing language choice and language use here have been straightforward, involving participant observation, and unstructured and structured interviews with numerous community members about language choice and use. Because this specific topic fascinated us and we wanted to understand it as much as possible, we have had many discussions with many community members about it. We also undertook a demographic survey of the community, conducted over two years by Gallardo Alejo (Secretario de MLP and Second Chief, himself Wichí with a Chorote wife and a Nivaclé mother). The demographic survey was straightforward and not difficult to administer – given the small size of the population, everyone knows which language everyone identifies with, which language they speak, and what their family members speak. Gallardo Alejo, as secretario de MLP, was able to write this information from personal knowledge, though he then went from household to household to confirm the information. The reliability of the information is not in doubt, and can be confirmed by asking almost any adult from the community. The questions posed by the survey were quite direct: what language/languages is/are spoken by each member of the household? Which language or languages are understood? How does each person identify himself/herself (with which ethnic group/language do they identify themselves)?

In the demographic survey, very few exceptions were found to the linguistic exogamy. There were no cases of Nivaclé married to Nivaclé, only four cases of straight Chorote married with Chorote, and six of Wichí married with Wichí. Wichí do not need to marry non-Wichí speakers, but many of the MLP Wichí are in fact married to speakers of one of the other two languages.Footnote ⁷ Several Wichí families from outside have moved to MLP in recent years, often to escape the flooding each year when the river leaves its banks, a problem for most places in the region except MLP. These cases usually involve Wichí married to Wichí. In addition, 30 persons were identified as having multiple ethnicity (called “cruzado” in local Spanish, where a person's identity was talked about in the surveys in terms of “race” (raza) in 2003 but as “ethnicity” (etnia) in 2007). Of these, only eleven involve spouses in a marriage speaking the same language (at least sometimes), where seven involved a Chorote-Wichí spouse married to Chorote spouse, three a Chorote-Nivaclé spouse married to a Chorote spouse, with one case of a Chorote-Guaraní spouse married to a Chorote spouse, and no cases of Nivaclé-Wichí spouses married to a Chorote, Nivaclé, or Wichí spouse.

The linguistic exogamy between Chorote and Nivaclé in MLP is not stated as an explicit rule and it appears not to be recognized consciously. The attitude seems to be that the pattern of linguistically mixed marriages just comes out that way, by chance more than by design, in spite of it being true of the vast majority of the Nivaclé and Chorote cases.

3.2. Language “identity”

In general, people “identify” with a single language and speak it with all others. They claim to understand but not speak one and in many cases both of the other two indigenous languages. This is remarkable, since the other two indigenous languages are spoken around them constantly and they usually have perfect comprehension of these languages which they claim not to speak. Personal identification with a particular language begins in childhood, perhaps around age 6 or so.Footnote ⁸ A child decides to identify with the language of one of the parents, although different children in the same family often choose to identify with a different language from that of some of their siblings, so that typically some siblings speak one of the languages and other siblings speak the other. The choice of language appears not to correlate in any significant fashion with the gender of the child, gender of the parent whose language is chosen, prestige, power, residence, or any of the social variables one might suspect from other situations. The basis for making the choice of which language to identify with is based on personal feelings, aesthetics, and early experiences. In numerous interviews where we have asked on what basis people decided to choose one language over the others to identify with, the answers were typically imprecise but along the lines of, “it just felt better,” “I liked it more,” “it seemed more comfortable,” “I thought is was nicer.” The result is that most households (typically made up of an extended family of three generations) have multilingual interactions all the time. No one of the languages, however, has greater value or prestige than the others. They are considered absolutely equal and the very question of whether one might be considered more prestigious or powerful or better in some way is so strange to inhabitants of MLP as to be almost impossible for them to comprehend.

For example, in one family, Elitye (grandfather) speaks Chorote but understands Nivaclé and Wichí (and knows Spanish, though not well). His wife Anita (grandmother) speaks Nivaclé and understands Chorote and Wichí (but knows no Spanish). Elitye speaks Chorote to Anita (and to every one else); she speaks Nivaclé to him (and to all others). Franco, their youngest son, living in the same household, speaks Chorote to his wife Valeriana and to others (his Spanish is good). Valeriana speaks Wichí to Franco and to the others (and she also knows Spanish well). (Of eight siblings, two of Franco's brothers identify with Nivacle, the others with Chorote.) Thus, in this household, Elitye speaks Chorote to all, Anita Nivacle to all, Franco Chorote to all, and Valeriana Wichí to all – all day, every day! Franco and Valeriana's children are too young to have made a choice about language yet. This family is by no means unusual. Most households in MLP have constant multilingual interactions involving spouses who speak different languages and family members who identify with different languages.

There are few exceptions to those who speak only one language but understand others. In the 2007 demographic survey (of 107 adults in 34 households), eight people were reported actually to speak (not just understand) more than one of the languages at least to some extent. In all these cases except one, Chorote was one of the languages spoken – in one case the person spoke Nivaclé and Wichí. (In one instance a person spoke both Chorote and Guaraní.) Cecilio, for example, a former lay minister, actually speaks (uses) both Nivaclé and Wichí, and formerly gave sermons in both. Two of his grown children actually speak both Chorote and Nivaclé (but identify with Chorote). These are exceptions, where most others in MLP insist they “speak” only one of the languages but “understand” the others. Other exceptions are instances where some grandchildren have not learned to understand a grandparent's language, forcing the grandparent occasionally to switch to the language of the grandchild in circumstances critical for communicating information. For example, Luis identifies with Nivaclé (his father was Wichí); he speaks Nivaclé with his wife, Ramona, who speaks Chorote to him. Two of his children identify with his language, Nivaclé, while three others identify with Chorote. Nevertheless, his youngest grandchildren (with a Chorote-speaking mother but non-Indian father), living in the same household, identify with Chorote and claim not to understand Nivaclé, forcing Luis actually to address them in Chorote for important communications, a language he claims he does not speak! Some of the children of some former community members who moved to Tartagal know only Spanish and do not speak any of the indigenous languages, though some understand; when they come to visit MLP, instances arise which force some interactions to be in Spanish that otherwise in the household would involve members speaking only the indigenous language with which each identifies personally. Other exceptions happen when people from MLP travel to other areas that are not so multilingual. For example, Franco (Chorote) and Eliseo (Wichí) both actually spoke Nivaclé, a language they claim not to speak, when visiting Nivaclé-speaking areas in Paraguay. (Both Franco's and Eliseo's mothers identify with and speak Nivaclé.) Luis claims not to speak Wichí, his father's language, and when visiting Wichí regions where only Wichí is spoken, Luis is addressed in and understands Wichí but speaks Spanish in return, though Spanish is a language he claims not to know well. Another seeming exception is that several of our consultants can transcribe and translate well recordings in the languages they claim not to speak, and some also will correct a mispronunciation or other mistake one of us makes in one of the languages they claim not to speak.

In general, however, the great majority of persons who have grown up in MLP claim to speak only one of the languages but to understand at least one and often both of the other two languages.

4.1. Geographical extent

Linguistic exogamy and multilingual conversations are not uncommon in the nearby regions, but they are also not nearly as typical as in MLP. Thus, this pattern of multilingualism is fairly common among the Chorote and Nivaclé in the indigenous sections of Tartagal and in the communities near MLP of La Bolsa and La Estrella, and not so common but still well represented in La Gracia, La Merced, Kilometro 1, and Kilometro 2. Wilbert (Reference Wilbert, Wilbert and Simoneau1985:3) reported La Merced, La Gracia, La Paz, and La Bolsa as “nowadays partially mixed villages.” Siffredi (Reference Siffredi1973:96) spoke of clan exogamy, reporting the statistically clear tendency in La Merced towards marital interchange between the different ethnic groups, which Braunstein (Reference Braunstein1983:66) interprets as meaning among the ancient tribes. Bartolomé (Reference Bartolomé and Dostal1972:430) reported that the Chorotes of Rivadavia Department “live together with Chulupí [Nivaclé], Toba and Mataco [Wichí]” and in San Martín “with Mataco and Chulupí.” He reported the “Chulupí” of these two departments as “mixed with Matacos and Chiriguanos.” Siffredi (Reference Siffredi1973:73) observed that “hoy día la convivencia armoniosa de estos dos últimos [Chorote and “Chulupí”] en la localidad de La Bolsa también revela innegables lazos de afinidad” [today the harmonious living together of these last two [Chorote and “Chulupí”] in La Bolsa also reveals undeniable connections of affinity]. Renshaw (Reference Renshaw2002:58) points out that the Nivaclé in Pedro P. Peña (in Paraguay, about 20 kms. downriver from MLP) are “largely intermarried with the Manjuy and Choroti [Chorote].” Renshaw holds the Manjuy to be a subgroup of the “Choroti” and points out that one of the groups of Manjuy is “largely intermarried with the Nivaclé” (Renshaw Reference Renshaw2002:59; see also p. 197). Further afield, Nivaclé in Paraguay marry other Nivaclé speakers (Chase-Sardi Reference Chase-Sardi2003); nevertheless, many Chorote there marry Nivaclé speakers. There are also cases in Paraguay of Nivaclé married with Guaraní, Maká, and Toba speakers, and others (Chase-Sardi Reference Chase-Sardi2003:306, 358–61, 441, etc.). Both Lafone Quevedo (Reference Lafone Quevedo and Hunt1915) and Hunt (Reference Hunt1915:xxxi), in one of the earliest reports actually to identify the Nivaclé, reported places with extensive “Choroti” [Chorote] and “Suhin” or “Chunupí” [Nivaclé] mixture. We do not know whether they engaged in dual-lingual patterns of interaction.

4.2. Historical antecedents

What is the history of this kind of dual-lingualism and language self-identification in the region? We believe the extent of passive multilingualism (dual-lingualism) and aspects of the choice of the language with which to identify probably are recent, due to the bringing together of speakers of different languages with the founding of MLP in 1944. However, there are also indications that aspects of this pattern of multilingualism were probably around long before that. There are not many historical accounts which mention these languages at all, much less so matters of language choice and multilingualism. Neither Chorote nor Nivaclé are identified clearly in colonial sources, but rather appear to have been considered part of the Mataguayo and Mataco [Wichí] groups (see Kersten Reference Kersten1904). The Chorote were first mentioned by Lozano (Reference Lozano and Altieri1941[1733]:59, 218) as “Choroties”; probably also the group he called “Xolota” (Lozano Reference Lozano and Altieri1941[1733]:82; cited also in Métraux Reference Métraux and Steward1946:235) were Chorote. However, next to nothing was known of either Chorote or Nivaclé until the late 1800s. The “Ashluslay” [Nivaclé] were first mentioned in the report of the Daniel Campos expedition of 1883 (Métraux Reference Métraux and Steward1946:236), by Cardús (1886) and by anthropologists who visited them in the early years of 1900 (Rosen Reference Rosen1904; Herrmann Reference Herrmann1908; Nordenskiöld Reference Nordenskiöld1912). Nevertheless, there are indications in the limited literature that suggest the dual-lingualism is not new among the Chorote and Nivaclé of the Pilcomayo region and did not originate per se in MLP itself. These limited reports appear to reflect a situation much as we have today in MLP where at least some Chorote and Nivaclé both intermarried and used their respective languages in multilingual interactions.

In older reports we find some clear indications; for example, Lafone Quevedo (Reference Lafone Quevedo and Massei1895:344, also 1897:53) cites Padre Doroteo Giannecchini as saying:

This seems very likely to be reporting the same kind of dual-lingualism that we see in MLP today. Lafone Quevedo cites Giannecchini further as reporting that the “Chorotis” [Chorotes] are a different tribe, but claiming “their language is a true dialect of Mataco or Noctene (Wichí) and many words are equal to those of the Noctenes [Wichí],”Footnote ¹⁰ and that “se hallan emparentados y en continua relación con los Noctenes” [they are found related to and in continuous relation with the Noctenes (Wichí)] (Lafone Quevedo Reference Lafone Quevedo and Massei1895:344; also 1897:53).

Nordenskiöld (Reference Nordenskiöld1912) spent several months in the region on an anthropological expedition (in 1908–1909) at a time when there had still been very little influence from non-Indians in this region. This is one of the earliest anthropological accounts to speak of “Tschoroti” [Chorote] and “Aschluslé” [Nivaclé, from the Chorote name for Nivaclé, /ahluhlay/] (/lh/ = voiceless “l”). Nordenskiöld (Reference Nordenskiöld1912:23) with a preamble of “since their culture is rather homogeneous, I believe they can be treated together”Footnote ¹¹, said:

He noted marriages between the Chorote and Nivaclé, and also, less commonly, between Wichí and Chorote as well as betweem Wichí and Nivaclé in the border areas (Nordenskiöld Reference Nordenskiöld1912:74). He reported, “the Nivaclé and Chorote tribes are on a par socially, as a Chorote girl can have a love affair with a Nivaclé man and a Nivaclé woman the same with a Chorote man”; this was not possible for either group with Chiriguano (Nordenskiöld Reference Nordenskiöld1912:80). He also observed that one Chorote chief was from a Nivaclé family (Nordenskiöld Reference Nordenskiöld1912:26; see also Siffredi Reference Siffredi1973:73). Wavrin (Reference Wavrin1926:42) found that “some tribes live on good terms among themselves … Suhin … called Chunapises by the Argentinians [Nivaclés], fuse a little with the Chorotis.”Footnote ¹³ Karsten (Reference Karsten1932:19) reported “much communication between the two tribes [“Choroti” and “Ashluslay”] and intermarriage is frequent” (see also Karsten Reference Karsten1932:49, 98). Similarly, Susnik (Reference Susnik1986–1987:33), from a broader geographical perspective, commented on the “Yofuaha-Chorótis” that the group along the Pilcomayo River maintained “relaciones intertribales ‘fraternizantes’ con los ‘Ashlushlai-Chulupíes’ [Nivaclé]” [‘fraternizing’ intertribal relationships with the Ashlushlai-Chulupíes (Nivaclé)], with exchange of women also with the “Noctenes” [Wichí]. Siffredi (Reference Siffredi1973:73) pointed out that the Chorote pronoun sam ‘we’ “includes indistinctly the Yojwaha Chorote and some Nivaclé groups].”Footnote ¹⁴

We frequently see in these historical and anthropological accounts the interpretation that held that the languages were mutually intelligible or were only dialects of a single language. Thus padre José Cardús (in his 1886 “Las Misiones Franciscanas”, cited by Pelleschi Reference Pelleschi1897:53) named the “Chunupis” (not to be confused with the Vilela, also called sometimes Chunupí) as one of the names by which the “Matacos” [Wichí] were known, and Padre Doroteo Giannecchini cited by Lafone Quevedo (Reference Lafone Quevedo and Pelleschi1897:53) named the “Chulupíes” among the Noctenes/Matacos groups: “The Noctenes, the Matacos, the Vejoses, the Guiznaes, the Chulupies (Nivaclés) are a single nation; all speak a single language and all have the same customs.”Footnote ¹⁵ Lehmann-Nitsche (Reference Lehmann-Nitsche1936:119) cites the common view before his time that Mataco [Wichí] and Chorote were but “codialectos” [co-dialects] as the reason why Chorote was so poorly known, almost unreported at that time.

We believe that observers misunderstood the dual-lingualsim among these groups and took it to mean they were speaking varieties of the same language to one another rather than that they were engaging in bilingual interactions. Some erroneously argue that these are “mixed” languages, based apparently in part on observations of communication among their speakers. For example, Braunstein & Miller (Reference Braunstein, Miller and Miller1999:10–11) say:

Braunstein has repeatedly advocated language mixture lying behind certain Chaco languages (Braunstein Reference Braunstein1993, Reference Braunstein, Martín and Diez1996; Braunstein & Miller Reference Braunstein, Miller and Miller1999; Martín & Braunstein Reference Martín and Braunstein1990–1991:10), as for example when he said:

Specifically he believes that there are indicators that the Maká and possibly the “Chulupí [Nivaclé] are groups that originated through the process of ethnic mixture (Braunstein Reference Braunstein, Martín and Diez1996:28). Braunstein (Reference Braunstein1993:7, Reference Braunstein, Martín and Diez1996:22–3) surveyed others who have held similar views about language mixture (Lafone Quevedo Reference Lafone Quevedo and D'Obrigny1896:134, 1915:xiii, Palavecino, Tovar Reference Tovar1981:20, etc.), and others who have favored the idea could be added (e.g. Rossi Reference Rossi2003:126, Susnik Reference Susnik1978:123–4, etc.).Footnote ¹⁷ Braunstein cites Hunt's (1915:238) statement as early recognition of this and as being very influential:

Braunstein (Reference Braunstein1993:8, 1996:23) reports that in historical research they had documented “situaciones de convivencia entre parcialidades matacas y chulupíes hacia principios de siglo en la margen izquierda del Bermejo” [situations of Mataco (Wichí) and Chulupí (Nivaclé) groups living together around the beginning of the century on the left bank of the Bermejo River], concluding that this indicated, “sin ninguna duda, algún grado de inteligibilidad” [without doubt, a degree of intelligibility].

We think those who see language mixture have just not understood the patterns of multilingualism in this region. The languages in fact show no evidence of “mixture” (heavy influence of language contact; see below).

In short, Chorote and Nivaclé were seldom referred to in colonial sources and were often thought to be dialects of Wichí (“Mataco”). Sources do, however, indicate that Chorote and Nivaclé lived together, often married one another, and seem to indicate that at least in some instances each spoke his or her own language. We believe these sources reflect a situation similar to that of MLP today where at least some intermarriage took place and where some people engaged in dual-lingualism, though the sources are neither as abundant nor as clear as could be hoped for on this.

4.3. Structural impacts?

It has been observed in other situations of intensive language contact that some languages have undergone structural changes to make them more similar to one another (cf. Aikhenvald Reference Aikhenvald, Aikhenvald and Dixon2007a). For example, in cases from India, different languages in contact have changed to become more structurally similar to one another, so that parallel morpheme-by-morpheme translations become possible (see for example Gumperz & Wilson Reference Gumperz, Wilson and Hymes1971; Nadkharni Reference Nadkarni1975). This raises the question, do the varieties of the three indigenous languages in MLP show evidence of change towards similar convergence? The answer is “no,” although the topic deserves more detailed investigation. First, these languages, as well as several others in the Chaco region, resist lexical borrowings. They have accepted very few loanwords from Spanish or other languages, and instead deploy native linguistic resources to create new words to accommodate concepts acquired through contact (acculturation). For example, ‘goat’ is not the usual loan from Spanish seen in many other indigenous languages of Latin America, but rather Nivaclé tašinštax, ‘goat’ is derived from tašinša ‘corzuela (grey brocket deer, Mazama gouazoubira)’ + -tax ‘similar to’, and Chorote sonta < sonaʔ ‘corzuela’ + -ta ‘similar to’ (Campbell and Grondona in press). That is, they acquired the concept, but accommodated the acculturation by using native resources in their languages. Second, contrary to expectations, at least in some regards, these languages, though in constant daily contact, appear to be diverging structurally in some traits. We will mention four examples which illustrate this. (The examples given here incidentally show also that though related, these languages are not especially close to one another, as mentioned above.)

For example, all three have /ɫ/, voiceless “l” phonemically, as in Chorote ɫop, Nivaclé ɫup, and Wichí ɫup ‘nest’, to cite a set of cognate forms. However, younger Iyo'wuhwa Chorote speakers in MLP are now changing this. It is not /ɫ/ for them any more; rather they have essentially changed it to a consonant cluster of /h/ or /x/ + voiced /l/, which alternates with just plain /l/ (with no /h/ or /x/ component) in some contexts, especially word initially and word finally, as in:

This change is taking place in spite of these Chorote speakers being in constant intensive contact with the other two languages which preserve their unitary voiceless /ɫ/ absolutely. This goes against areal linguistic expectations of languages in this sort of close, intensive contact.

In another example, both Nivaclé and Wichí have and frequently use the contrastive 1st person plural inclusive and exclusive pronominal forms, as seen in the contrasts in Nivaclé between the pairs (1a) and (1b), between (2a) and (2b), and between (3a) and (3b):

The contrast in Wichí is seen in:

However, Chorote speakers in MLP have lost this contrast and now have only a single first person plural category:

Contrast Chorote (8a) with the Nivaclé equivalents with cognate roots in (8b) and (8c):

Contrast Chorote (9a) with the Nivaclé cognates in (9a) and (9b):

Thus, Chorote has only ‘we’ generally, with no implications for either inclusive or exclusive readings. There are, however, frozen forms (no longer understood as different) in the language which reveal that Chorote also had the contrast, as in the case of sam and sameɫ. Historically sameɫ was exclusive, but now both are in use interchangeably, both mean just ‘we’. Again, we would not expect one of the languages to lose a morphological contrast so prevalent in the other two languages that speakers of Chorote hear and understand constantly in MLP.

The third example involves change in Nivaclé. In both Chorote and central Paraguayan dialects of Nivaclé, active verbs when they appear with the ‘prospective’ (future intent) clitic, usually translated ‘going to’, must nevertheless take the pronominal prefixes for inactive/stative verbs, as seen in the contrasts in Chorote between (10a) and (10b) and between (11a) and (11b):

This construction with the inactive/stative prefixes required with the ‘prospective’ clitic is the original state of affairs for these languages. However, MLP Nivaclé has undergone an innovation. The corresponding construction in Nivaclé here takes not inactive/stative personal pronoun prefixes, but rather only active ones, as in (12a):

It is ungrammatical with the inactive/stative subject prefix, as in (12b):

We would expect the presence of this construction in Chorote and other Nivaclé dialects, where its occurrence is reasonably frequent, to have influenced MLP Nivaclé to maintain the construction; however, this is not what happened.

A fourth example involves the contrast between the low unrounded vowels, /a/ (central) and /ɑ/ (back) in the Nivaclé of MLP, as in yiyaʔ ‘mosquito’ and yiyɑʔ ‘he drinks’. This contrast is now being lost (merged to /a/). Some older speakers still make the contrast consistently; younger speakers, however, do not (mostly have yiyaʔ for both ‘mosquito’ and ‘he drinks’), and in some words there is now free variation. They recognize that for some people there are two separate vowels, and sometimes they pronounce the back vowel, but they are unable to tell whether a particular word should have /a/ or /ɑ/, now indistinguishable in their phonemic value for them. Once again, it is unexpected to see such a contrast being lost here, since Wichí also contrasts these same two vowels (although not in all dialects, and hence not by all Wichí speakers in MLP).Footnote ¹⁸

In short, while the question of structural convergence is an important one and needs to be studied in greater detail in the MLP languages, it is clear from these examples that pressure towards conformity among the three languages because they are in intensive language contact has not been completely compelling, since they are undergoing some changes which result in greater difference, rather than greater similarity, among the three.Footnote ¹⁹

5.1. “Dual-lingualism” (“passive multilingualism”)

Multilingual interactions involving interlocutors speaking different languages to one another are reasonably common in the world, called ‘dual-lingualism’ (Lincoln Reference Lincoln1979), ‘passive bilingualism’, and sometimes ‘equilingualism’, ‘non-reciprocal (or asymmetrical) bilingualism’, ‘pseudo-bilingualism’, ‘receptive bilingualism’, ‘restrictive bilingualism’, and ‘semi-bilingualism’ (cf. Hockett Reference Hockett1958:327; Mackey Reference Mackey, Ammon, Dittmar, Mattheier and Trudgill1988:1487; Sankoff Reference Sankoff1968; Warter Reference Warter2001). Minimally some passive knowledge of the other person's language is required in these situations, though it is possible that one could have stronger knowledge of the other language and still speak only his or her own in such interactions.Footnote ²⁰

Such dual-lingual communication has been observed across generations, for example among some immigrants to the US where parents speak the language of their home country and their children answer them in English (cf. Haugen Reference Haugen1953). Dual-lingual communication is also well-known among Scandinavians involving Swedish, Norwegian, and Danish, where the languages and dialects are not very divergent from one another (cf. Warter Reference Warter2001) – there dual-lingual usage is not typically between family members or spouses, however. Lincoln (Reference Lincoln1976, Reference Lincoln1979), who coined the term, described dual-lingualism in Bougainville (Papua New Guinea) where Banoni (Austronesian) and Siwai (non-Austronesian) speakers talked with one another, each in their own language (see below). Hooley (Reference Hooley1971) suspects there was passive bilingualism among peoples of the Morobe District, Papua New Guinea, reporting it specifically for Bukaua and Jabem, two closely related languages (sometimes considered dialects of a single language). Situations with dual-lingualism are sometimes said to be common in Melanesia and Aboriginal Australia (e.g. Mühlhäusler et al. Reference Mühlhäusler, Dutton, Hovdhaugen, Williams, Wurm, Wurm, Mühlhäusler and Tryon1996:415). Thurston (Reference Thurston1987:74) speaks of extensive multilingualism and dual-lingualism in North West New Britain, but with no details given, and Wurm (Reference Wurm, Wurm, Mühlhäusler and Tryon1996a:737) asserts that in the Philippines “there are very many instances of active or passive, one-way or two-way bilingualism,” but again with no details. This notwithstanding, Dutton (Reference Dutton, Bellwood, Fox and Tryon1995:220) reports that no case of “dual lingualism” involving Melanesian languages has been adequately described and no additional cases (beyond Lincoln Reference Lincoln1979) have been reported.

A number of other cases are cited in the literature, though usually again with little specific information. These include the following: In Siberia, Dolgan and Yakut, “mutually intelligible” Turkic “languages” (Wurm Reference Wurm, Wurm, Mühlhäusler and Tryon1996b:976); among the dialects of Altai and “their Turkic-speaking neighbours … Shor, Khakas, Tuva, and Tofa” (Wurm Reference Wurm, Wurm, Mühlhäusler and Tryon1996b:975); Nanai and Samagir, Samagair and Ulchi, and Oroch and Udege (Udehe), Tungusic languages; and Chukchi and Koryak (Wurm Reference Wurm, Wurm, Mühlhäusler and Tryon1996b:976–7). In northwestern Gansu province in China, Uighur and Kazakh (Turkic languages) (Wurm Reference Wurm, Wurm, Mühlhäusler and Tryon1996c:222). In Xinjiang province, six closely related Turkic languages, especially among Uighur, Uzbek, and Tartar, but also prevalent between Uighur and Kazakh, and between Uighur and Kirghiz; also sometimes Tuva and Kazakh (though these two are not so closely related Turkic languages); and Wakhi and Sarikoli (Pamir languages, Iranian branch of Indo-European) (Wurm Reference Wurm, Wurm, Mühlhäusler and Tryon1996c:223). Intercommunication among different Chinese dialects (most being Mandarin dialects) in northern China (Wurm Reference Wurm, de Rachewiltz, Wurm, Mühlhäusler and Tryon1996:823–4). In Mongolia, Khalkha, Oirat, and Buryat, closely related Mongolian languages (Wurm & Rachewiltz Reference Wurm, de Rachewiltz, Wurm, Mühlhäusler and Tryon1996:909). In Krakalpakistan, “among the Turkic-speaking nationalities, whose languages are closely related and mutually intelligible” (Nasyrova Reference Nasyrova, Wurm, Mühlhäusler and Tryon1996:926). In Turkmenistan, Turkmen and the other Turkic languages (Baskakov Reference Baskakov, Wurm, Mühlhäusler and Tryon1996:931). In the upper Xingu region of Brazil, with extraordinary linguistic diversity, “members of each tribe speak their own language and have a passive knowledge of some of the other languages” (Adelaar Reference Adelaar, Wurm, Mühlhäusler and Tryon1996:1345; cf. Monod Reference Monod(-Bécquelin)1970).Footnote ²¹ There appears to be some dual-lingualism in the Vaupés region (see more below), but it is limited and sporadic:

Most of the cases reported in the literature (but not all) have in common that the people who engage in this kind of dual-lingualism live in small groups, where the population density is low, and in most instances the languages involved are closely related. The smallness of group size and lowness of population density is true of MLP, but the languages involved, though belonging to the same family, are not closely related.Footnote ²²

5.2. Dual-lingualism among spouses

The cases reported from elsewhere of spouses speaking different languages to one another seem sporadic and temporary, usually not regulated by formal rules of linguistic exogamy, residency, etc. It had been suggested that such a pattern might exist on Erromango, Vanuatu, but Terry Crowley (personal communication 2004) confirmed that this is by no means a regular pattern there and the only cases are where a husband might speak Erromangan and the wife speaks back to him in Bislama (the English-based creole, lingua franca) while understanding Erromangan. Women often come from elsewhere and speak a different language, but the expectation is that the wife will eventually learn and use the local language in the village. Lincoln (Reference Lincoln1979:66) reported a case from the Solomon Islands observed by Evelyn Todd of a Savosavo-speaking man and Nggae-speaking wife “conversing dual-lingually.” Savosavo is a Papuan language, while Nggae is Austronesian. Lincoln (Reference Lincoln1976:99) reported a second case, again of only a single dual-lingual marriage, in Bouganville (Papua New Guinea):

It is not clear from this how dual-lingualism may work between spouses there; however it seems clear that it is not nearly so ubiquitous as in MLP nor as institutionalized, and also the two languages here do not seem equally balanced (where Banoni learn and speak Siwai, while Siwai may only learn to understand Banoni). This is different from the situation in MLP.

In short, the cases of dual-lingualism elsewhere appear more sporadic, not as general, long-lasting, and institutionalized as in MLP.

5.3. Linguistic exogamy

While rare, various cases of linguistic exogamy have been reported (see Hill Reference Hill1978 for a summary). It is reported as probable in Arnhem Land, Australia, present among some groups from the Pacific Northwest Coast and Plateau in North America, for the Upper Xingu region in central Brazil, and most famously in the Vaupés region of the northwest Amazon.

In Northeast Arnhem Land, linguistic exogamy has been reported among the Murngin, where patrilineal, territorially-based, exogamous groups (called mada) had a single language, unintelligible with a man's wife's language (from a different mada). Children learned first the mother's language, and later shifted to the father's (Berndt Reference Berndt1964:292; Berndt & Berndt Reference Berndt and Berndt1964:63). Warner (Reference Warner1937) reported that most people spoke three or four languages, and adult men often spoke seven or eight, and that the mada and moiety language differences were actually only slight (though Berndt & Berndt Reference Berndt and Berndt1964 claim they were mutually unintelligible) (see Hill Reference Hill1978:15).

Heath (Reference Heath1978, 1981) briefly described cases of extralinguistic marriage between Ritharngu and Ngandi and between Warndarang and Nunggubuyu. Heath (Reference Heath1981:359) surmises for Arnhem Land that there was a high rate of inter-ethnolinguistic group marriages, around 50% for the average EG [ethnolinguistic group] with four clans and 200 persons, hence half the girls born into the EG would be sent at puberty to join husbands in other EG's, and half the wives obtained by men in the EG would be imported from neighbouring EG's. He says:

There is no indication of how widespread passive bilingualism was in these Australian situations; individuals appear to have learned and used other languages as well as their own. Another Australian case, in western Cape York, might be somewhat different, reported by Johnson (Reference Johnson and Baldi1990), though again we have little detail. “Marriage among the Nganhcara is most frequently between speakers of different patrilects” (Johnson Reference Johnson and Baldi1990:422). The patrilects are very closely related varieties of Nganhcara:

“It is also common to hear conversations between people each speaking their own patrilect, but at least passively competent in the other patrilects being spoken” (p. 422). These patrilects, however, are essentially dialects of the same language (or very closely related languages).Footnote ²³ In such cases a man learned and on appropriate occasions used both his father's and his mother's language.

Tindale (Reference Tindale1953:186) held that extra-“tribal” marriages (meaning by his definition also extra-linguistics ones, since for him “tribe” was equivalent to dialect/linguistic group) were common. Nevertheless, he estimates their frequency at only 15%, ranging from 7% to 21% (see Owen Reference Owen1965:679). This is very different from the situation in MLP where nearly all marriages involving Chorote and Nivaclé are cross-linguistic, as are those involving most Tukanoan groups in the Vaupés region. Heath (Reference Heath1978) describes groups which, whether by choice or not, were forced due to their small sizes to interlinguistic marriages (see also Clendon Reference Clendon2006:47).Footnote ²⁴ He reports,

Some scholars believe a similar fate fostered the linguistic exogamy in the Vaupés region and among some tribes of the Upper Xingu region of Brazil (see below).

Hill (Reference Hill1978:16) reports that:

This is not the same sort of linguistic exogamy, where failures to conform are considered incest in the Vaupés region (Aikhenvald Reference Aikhenvald2003b:2), or where conformity covers nearly all marriages as in MLP and in the Vaupés, and where status and wealth play no role in MLP.

Owen (Reference Owen1965:678) characterized Baja California patrilocal bands' typical family situation: “with the patrilateral extended family as the principal vehicle of cultural transmission, one of adult males as native speakers of one language, and some portion of the females as native speakers of different languages or at least different dialects.” However, unlike in MLP and the Vaupés region, children were enculturated principally in the language of the mother and other females, not in that of the father and other males. Thus Owen (Reference Owen1965:678–9) concluded:

It is believed that, due to their small numbers, the Trumai and others of the Upper Xingu reserve undertook many linguistically exogamous marriages. A child learned at least two languages, his father's and his mother's, though he may learn others, also (Monod Reference Monod(-Bécquelin)1970).

The Vaupes area of the northwest Amazon region offers the most famous case of linguistic exogamy, obligatory among most of the Tukanoan groups, but also involving a few Arawakan groups there. Residence is patrilocal, traditionally in long houses, where women marrying into the long house spoke different languages so that children grew up with the language of their fathers but also with knowledge of their mother's language and often several others. There was strong interest in learning new languages. (For information on multilingualism and linguistic exogamy in the Vaupés region, see Aikhenvald Reference Aikhenvald, Aikhenvald and Dixon1999, Reference Aikhenvald2000, Reference Aikhenvald, Aikhenvald and Dixon2001, Reference Aikhenvald2002, Reference Aikhenvald2003a, Reference Aikhenvald2003b, Reference Aikhenvald, Aikhenvald and Dixon2007a, Reference Aikhenvald, Aikhenvald and Dixon2007b; Chernela Reference Chernela1982, Reference Chernela1989, Reference Chernela, Rival and Whitehead2001, Reference Chernela2004; Epps Reference Epps2005, Reference Epps2007; Goldman Reference Goldman1963; C. Hugh-Jones Reference Hugh-Jones1979; S. Hugh-Jones Reference Hugh-Jones1993; Jackson Reference Jackson, Bauman and Sherzer1974, Reference Jackson and Smith1976, Reference Jackson1983, Reference Jackson and Kensinger1984, Reference Jackson1988, Reference Jackson, Urban and Sherzer1991, Reference Jackson1994; Santos-Granero Reference Santos-Granero, Santos-Granero and Hill2002; Sorensen Reference Sorensen1967, Reference Sorensen, Klein and Stark1985; Stenzel Reference Stenzel2005.) Aspects of linguistic exogamy and its impact on language choice in the Vaupés relevant to the MLP situation are revealed, at least partially, in the following citations. “In spite of the traditional multilingualism enhanced by language-based exogamy, every Indian identifies with just one language – that of the father” (Aikhenvald Reference Aikhenvald2003b:2). “A woman invariably uses the language of the longhouse – her husband's language – when talking directly with her children.” “Children are usually bilingual in both their father's and their mother's languages, but commonly use the former.” (Sorensen Reference Sorensen1967:677). “All children use the longhouse language (i.e., their father's) to both their father and mother.” (Sorensen Reference Sorensen1967:678.) In each of these, MLP contrasts with the situation in the Vaupés (see comparisons below).

The age of the linguistic exogamy in the Vaupés region is unknown, though Jackson (Reference Jackson1994:387) believes “that Tukanoan linguistic exogamy is an extremely dynamic and relatively recent institution, the result of indirect pressure from colonization efforts elsewhere in Colombia, Venezuela, and Brazil.” Jackson (Reference Jackson1983:21, 100–1) attributes the rise of linguistic exogamy to population loss, small population size, and loss of sibs. For example, the Makuna, who no longer practice it, “recognize the rule of linguistic exogamy as an ideal”; they say that “formerly the different Makuna sibs spoke different languages and that it is only recently that they have adopted a common language” (Århem Reference Århem1981:114).

While not all cases fit, typically groups practicing linguistic exogamy are small, with band-level social organization and sparse population density, hunters and gatherers or sparsely populated horticulturalists (e.g. in Papua New Guinea and the Vaupés region). The groups in MLP fit this description (or at least did until recently).

5.4. Comparisons

Multilingualism and language use in MLP differ significantly from all the cases mentioned above. We compare and contrast these, with emphasis on the similarities and differences between MLP and the Vaupés region.

The two situations are similar in that spouses marry someone who speaks a different language, and they avoid borrowings from other languages (explicitly mentioned for the Vaupés languages, Sorensen Reference Sorensen1967; for MLP, see Campbell & Grondona in press). They involve small populations and low population density (Jackson Reference Jackson1983:17 for the Vaupés area). In both regions the languages are considered of equal status. “Tukanoans in general assume that Vaupés languages are equal to one another” (Jackson Reference Jackson1983:164); “Vaupés languages are in no way ranked … Despite the fact that Tukano is a lingua franca, it is not considered preferable” (Jackson Reference Jackson1983:175). The two situations may differ from most other cases in the world in that the languages involved are not necessarily closely related. Sorensen (Reference Sorensen1967:674) believed the Tukanoan languages are less closely related to one another than Romance languages are amongst themselves, and a few instanced of non-Tukanoan languages are also involved in the Vaupés area.

However, the situations in the Vaupés region and in MLP differ significantly, in the following ways:

1. (1) In the Vaupés area, one identifies with and is loyal to the father's language. Wives are expected to learn and use the language of the husband's longhouse, and people there are genuinely happy to learn and speak the languages of other people they come to have conversations with. They learn and use other languages eagerly. In MLP people typically use only their own language, almost never shifting to the language others speak, resulting in multilingual interactions where different languages characterize the conversations. The language a person identifies with can be either the father's or the mother's.

2. (2) Men exchange women in the Vaupés (Jackson Reference Jackson1983:124–7); for the groups represented in MLP, women initiate matters and select the men they want to marry (cf. Chase-Sardi Reference Chase-Sardi2003:562, 580–25; Karsten Reference Karsten1932:53–6; Métraux (Reference Métraux and Steward1946:325, Rosen Reference Rosen1924:159–60).

3. (3) Residence is patrilocal in the Vaupés (Sorensen Reference Sorensen1967; Jackson Reference Jackson1983). “Residential exogamy, an automatic result of the rules of language group exogamy and patrilocal residence, can claim to be a principle in its own right” (Jackson Reference Jackson1983:135). Vaupés patrilocality with residential exogamy leads somewhat automatically to linguistic exogamy. This is not the case in MLP, where a man often goes to live with the wife's family initially, until after the first child is born, and then frequently the couple go to live with the husband's family, though these patterns are not rigid (cf. Chase-Sardi Reference Chase-Sardi2003; Renshaw Reference Renshaw2002).

4. (4) Women are community outsiders in the Vaupés (who come to the longhouse because of patrilocal residence rules), often not the case in MLP. In the Vaupés it is outsider women who are often seen as trouble-makers (Jackson Reference Jackson1983:130, 190; Jackson Reference Jackson1992:13); in MLP it is insider women who are considered the trouble-makers, leading some men to say they prefer women from outside MLP because they make less trouble.

5. (5) Linguistic exogamy in the Vaupés is a major principle of social organization, residence patterns, kinship, and regional interactions. Among the groups which practice it, failure to conform would be considered incest. The rule is overt, recognized, and formal. In MLP, in contrast, there appears to be nearly complete linguistic exogamy for Chorote and Nivaclé speakers, but the practice is not overtly recognized nor talked about in any consistent way. The Wichí and Nivaclé of other regions do not necessarily practice linguistic exogamy, not all Chorote elsewhere practice it, and only some of the Wichí in MLP have such marriages, and this may make the actual practice in MLP seem less noticeable or less compelling. The Chorote and Nivaclé of MLP practice linguistic exogamy, but it is not as strikingly integrated in their social fabric as in the Vaupés, and seemingly is not overtly recognized as a necessary rule.

6. (6) Dual-lingualism, while sometimes practiced in the Vaupés region, is more sporadic and short-term, not general nor characteristic of most people, as it is in MLP (see above, Sorensen Reference Sorensen1967:678). In MLP, dual-lingualism is the predominant kind of verbal interaction.

7. (7) Language identity is with the father's language in the Vaupés; language identity in MLP is a personal choice, and depends on an individual's feelings about the language selected.

8. (8) In most cases of dual-lingualism (but not all), the languages (and dialects) are quite closely related. This is not the case in MLP. It is possibly also not the case in the Vaupés, although we believe the Eastern Tukanoan languages that participate there to be more closely related to one another than the Matacoan languages of MLP are to one another. (Naturally, this does not hold for the cases of intermarriage in the Vaupés between speakers of Arawakan languages and Eastern Tukanoan languages. These non-Tukanoan language participants are much fewer in number, but nevertheless they do practice absolute linguistic exogamy as a group.)