The name ‘Wajãpi’ ([wajãˈpi] ~ [wãjãˈpi]) is both a glottonym and an ethnonym associated with a series of Tupi-Guarani-speaking communities living in a vast region of the northeastern Amazonian fringe, straddling the border of Brazil and French Guiana (see Figure 1). As is the case with other Tupi-Guarani groups dwelling in the eastern Guiana Shield and its environs, such as the Emerillon/Teko and the Zo’é, the Wajãpi are latecomers to the region, having crossed the Amazon River northwards during the 18th century (see Gallois 1980, 1988; Métraux 1927). Also in common with other groups in the region, the modern Wajãpi result from the coalescence of a series of politically independent communities which likely shared a common culture and which all spoke closely related speech varieties. These groups are taken to correspond to specific ethnic aggregates identified in the ethnohistorical literature, such as the Kaikusiana and the Piriu, all inhabiting the region encompassing the middle and upper courses of the Oyapock, Amapari, Araguari, and Jari Rivers (see Gallois 1980, 1988 for an extensive and nearly exhaustive discussion of the relevant historical facts).1 The 2014 SIASI/SESAI census figures give a total of 1,221 Wajãpi living in Brazil, and a nearly contemporary figure for French Guiana gives 950 members of the group living across the international border, roughly totaling the figure of 2,000 individuals given in the summary above.2

The dialect focused on in this report is known as Amapari Wajãpi (Jensen 1999: 132; Mello 2000: 120; Schleicher 1998: 11), or Wayãpi-puku (Grenand 1980, 1989: XIII). It is spoken in the Terra Indígena Wajãpi (Wajãpi Indigenous Reservation), Amapá State, Brazil. A smaller number of Wajãpi reside more or less permanently in the cities of Pedra Branca do Amapari and Macapá. Most published descriptions of the Wajãpi language have focused on dialects of the Upper Oyapock variety of Ytuwasu/Trois Sauts (Copin 2012; Grenand 1980, 1989). The variety spoken in the upper Jari River, Brazil, was the focus of attention by SIL linguists Allen Jensen and Gary Olson (see Grenand 1989: 9). In the period from 1968 to 1981 a group of Wajãpi speakers left the Cuc/Kouc River and migrated to the Upper Oyapock region, so that this variety was also, to some extent, sampled in these works (see Grenand 1989: 9). A variety of Wajãpi is also spoken in the village of Camopi, downriver from Ytuwasu, along the middle Oyapock River, close to the confluence of the Oyapock and Camopi Rivers (e.g., Copin 2012: 1). It is not clear to what extent this variety differs from that of Ytuwasu, and these two are often considered jointly as representing a ‘Guianese’ Wajãpi variety, contrasted with the Amapari or ‘Brazilian’ Wajãpi that is the focus of the present report (see Copin 2012:2).3 The map in Figure 1 provides an indication (though slightly outdated) of the locations of Wajãpi settlements near the Brazilian-French border.

The Amapari dialect of Wajãpi (henceforth, AW) has been dealt with specifically only in Jensen (1984), a diachronic investigation of Wajãpi (also including the Upper Jari dialect), and in A. Jensen (1990), a study focused on the morphosyntax of negation in Wajãpi, and which includes, as endnotes, brief considerations on the phonology of the language. The first descriptive work dedicated exclusively to the phonology of the language is Carvalho (in press). Currently no variety of Wajãpi enjoys a published, comprehensive reference grammar supported by analyzed texts. For instance, Copin (2012), a description of Ytuwasu Wajãpi, is strong on the morphosyntactic aspects but offers only a very brief discussion of phonology. His work supersedes the earlier grammar of Grenand (1980), as well as the very brief discussion of clause structure in the Upper Jari variety of the language found in G. Olson (1978). Jensen (1978) discusses non-verbal clauses in Upper Jari Wajãpi. Grenand (1989) is a very good Wajãpi-French dictionary with a French-Wajãpi vocabulary. The dictionary entries often contain etymological hypotheses (many of which are fanciful, however), and its coverage within certain ethnographically important domains, in particular of material culture, is rich and accompanied by insightful illustrations. Earlier, R. Olson (1978) produced a small vocabulary organized by semantic fields based mostly on the Upper Jari dialect. Grenand (1978) constitutes the sole publication featuring an analysis of a Wajãpi text (in this case the Ytuwasu variety of Wajãpi).

In view of the concentration on a few varieties of the language—notably the Upper Jari and Upper Oyapock dialects—it is not surprising that dialectal differences are poorly understood. AW remains the least known dialect. In fact, AW, identified as ‘Wayampipuku’, is even presented as a separate language of the branch VIII of the Tupi-Guarani language family in some comparative studies (Rodrigues 1985: 42; Rodrigues & Cabral 2002: 499). Attitudes and perceptions of the Wajãpi themselves seem to underscore the need for further study of this variety: Grenand (1989) notes that Upper Oyapock speakers tend to consider the AW the “true” Wajãpi language (also Grenand 1980: 29).

Although precise sociolinguistic census figures are unavailable for the AW variety, the author of the present report is in a position to offer some qualitative observations, over and above the merely quantitative figures offered by SIASI/SESAI, based on research informed by pedagogical activities within the Wajãpi Indigenous Reservation and by close to three years of constant interaction with the Wajãpi in the city of Macapá. Virtually every member of the ethnic group speaks Wajãpi as their first language, and children, most older people, and many women are essentially monolingual. Even among younger men, who are typically those with a greater command of Brazilian Portuguese in Brazilian Indigenous groups, one can find among the AW many individuals with whom it is almost impossible to maintain even brief conversations in Portuguese.

A conventional, practical orthography for Wajãpi is in use in schools and in the production of written material, but the degree of standardization is limited. Thus, while the 2013 Wajãpi translation of the New Testament (the Wycliffe Bible Society’s Janejare’e Ayvukwerà) employs a grave accent to represent the predictable nasalization of unaccented final a, the Wajãpi themselves most often use a tilde, based on Brazilian Portuguese conventions, while French-based conventions employ none of these symbols. Other conventions are also only unenvenly followed, such as that of writing ⟨Waiãpi⟩ for the ethnonym, reserving ⟨Wajãpi⟩ for the glottonym. The former is used, for instance, in the official personal documents issued for the Wajãpi, but not necessarily in publications and printed material. This makes the production and use of printed material in Wajãpi, whether in formal schooling or elsewhere, difficult. The Centro de Trabalho Indigenista (Center for Indigenist Work) has produced pedagogical material on themes ranging from numeracy to environmental education, but these seem not to be currently in use by the Wajãpi, and copies of these materials are in general very hard to obtain (see Gallois 2021 for some relevant references).4 There are currently a few isolated initiatives underway to produce pedagogical and written material on the language. These projects are headed either by secular or religious NGOs or by individual members of the ethnic group.

Even when visiting the major urban center of Macapá, the Wajãpi converse freely among themselves using their language. However, there are some indications that stigma from prejudiced non-Indigenous individuals is starting to have a deleterious effect on the pride and self-esteem of the group. For example, after being told by a governmental health worker that the Wajãpi should “start speaking a proper human language”, participants in a workshop offered for Indigenous professors at the Wajãpi Indigenous Reservation began expressing concerns about the language’s fitness for use in typically non-Indigenous domains. Finally, it is not uncommon to find the Wajãpi employing AW in social media and messaging apps such as WhatsApp. Some Wajãpi speakers have even started to compose songs modeled on the most popular styles among the local non-Indigenous population, and then record them and upload them to YouTube.5 More systematic investigation of the language practices and language ecology of AW is certainly needed.

Our current research on the Wajãpi language has had two main emphases: building a collaborative network with Indigenous teachers and students, and carrying out basic descriptive research on the language. The latter has been focused on the particular phonological and morphosyntactic properties of the AW variety, as it has been relatively neglected in past research on Wajãpi.

Toward the first goal, our group at the Federal University of Amapá (UNIFAP), Brazil, has enrolled the first Wajãpi teacher in a graduate course. Makaratu Wajãpi is currently developing a research project on his language under the supervision of the present author, and we have every expectation that the products of his graduate-level research will have positive repercussions for his activities as a Wajãpi language teacher in the Indigenous school at the Wajãpi Indigenous Reservation. Makaratu Wajãpi has started to develop teaching materials for the Wajãpi language, including the coinage of specialized terminology for grammatical description (e.g., parakatu-ayvu ‘tick-word’ for ‘affix’, the idea being that an affix is like a tick, an element that hangs on something else). Also of pivotal importance for these activities is the fruitful partnership developed with the dedicated workers of the Secretaria Estadual de Educação (SEED, the State Secretary for Education). This partnership yielded a workshop for Indigenous Wajãpi teachers who are in charge of classes in their native language in the local Indigenous school (see Figure 2). The workshop, which took place in 2019, successfully imparted the relevant descriptive framework and grammatical terminology necessary for their work due to the pedagogically felicitous use of Wajãpi both as the language of instruction and as the object of grammatical analysis (in contrast to previous attempts that employed Brazilian Portuguese in both functions).

As to the descriptive goals, one target of concerted effort is charting the characteristic isoglosses delineating the use vs. non-use of phonological/phonetic, lexical, and morphosyntactic features that set the AW variety apart from the other, better described Wajãpi varieties. Particularly noteworthy is the realization that a series of traits that have been attributed to the Wajãpi language as whole do not, in fact, hold for AW. Most salient among these are accentual retraction and the loss of word-final stops. On the former, while the Jari variety of Wajãpi does show an accentual retraction shift, with [ˈjawa] ‘jaguar’ as the reflex of Proto-Tupi-Guarani (PTG) *jaˈwat, the Amapari variety has [jaˈwa] ‘jaguar’, retaining the etymological accent placement. On the second feature, Amapari Wajãpi seems to retain the once present word-final stop consonants as morphophonological alternants: While -aku ‘to be hot’ is the dialect’s reflex of PTG *-akup ‘to be hot’, addition of a suffix, such as the intensifier -aɨ, yields [akuˈßaɨ] ‘to be very hot’, with the word-medial fricative [ß] reflecting a lenited alternant of PTG *p in the same context; see Carvalho (forthcoming).

AW also has some innovations that make AW unique among Wajãpi dialects and, in fact, among Tupi-Guarani languages, but these had been missed in past accounts. At the phonetic/phonological level, AW has nasalized reflexes of final, unaccented PTG *-a, as in e.g. [ˈaɾã] ‘day, time; weather; world’ < PTG *ara. This trait is particularly interesting because it has been described for some other Tupi-Guarani languages, like Araweté and Tapirapé of the Xingu-Tocantins region, to the south of the Amazon. This is precisely the region from which the Wajãpi ancestors migrated in the first decades of the 18th century. Another conservative feature of AW is the fact that it has not undergone the merger of PTG *w and *ß which took place in the other varieties of the language (compare: Jari Wajãpi [aˈwasi] ‘maize’, but Amapari Wajãpi [aßaˈsi] ‘maize’ < PTG *aßatí).

In regard to morphosyntax, AW has neutralized the contrast between Set I (for A and S_A) and Set II (for O, S_O, and Possessive) person markers for a second person singular argument. Tupi-Guarani languages are traditionally described (e.g., Jensen 1998) as having two sets of person-indexing prefixes used in verbs, nouns, and postpositions. Set I markers occur in transitive verbs coding an active or controlling argument (here referred to as A) whenever this argument is a first or second person, and when the affected argument (here denoted O) is a third person. These prefixes also index the single argument (S_A) of active intransitive verbs. Set II prefixes, in turn, are absolutive, in that they code either first or second person O argument in a transitive verb (when the A argument is third person) or the S_O argument of a subclass of intransitive verbs.6 Instead of showing ere- ‘second person singular, Set I’, still preserved in the Guianese varieties of the language, AW retains only ne- ‘second person singular, Set II’ for the indexing of a second person argument. This produces ambiguities between ‘active’ or ‘affected’ readings of the verbal argument indexed, as in ne-nupã, which may be translated (out of context) either as ‘you hit him/her/them’ (with a second person singular A argument) or as ‘he/she/they hit you’ (with a second person singular O argument).

Some impressionistic observations in the literature, such as the description of AW as a Wajãpi variety marked by the relatively stronger phonetic nasalization and glottalization, seem to be vindicated by the preliminary descriptions carried out so far (Carvalho in press). The realization of the AW glottal stop differs markedly from what is reported in existing descriptions of other Wajãpi varieties, where no mention is made of glottalization and [ʔ] is given as the sole realization of /ʔ/ in all contexts (see Copin 2012: 410; Grenand 1980: 35). For example, the noun which is phonologically /-pɔsiʔa/ ‘belly’ is in fact often realized in AW with a very brief glottal constriction but with glottalization (or creaky voice) spread all over the prosodic word, as in /ɛ-pɔsiʔa/ [ɛ̰pɔ̰s̰ḭˈˀa̰] ‘my belly’. As for nasalization, the innovative nasalization of unaccented final -a in AW mentioned above also gives this variety a noticeable ‘nasal pronunciation’ when compared to other Wajãpi varieties.

Other claims about the AW variety are harder to validate. An example is the claim that AW has incorporated fewer Cariban loanwords in comparison to other varieties of the language (Grenand 1980: 29). It is clearly the case that AW features a number of such loans. The kinship terms -parɨ ‘grandson’ and -pipi ‘father’s sister’, for instance, were likely borrowed from Apalai, a Cariban language spoken in the area of the Paru and Jari Rivers in Brazil. The cultural terms kuwakɨ ‘flour’ and marija ‘knife’ were possibly borrowed from Wayana, another local Cariban language spoken slightly to the north of Apalai in Brazil, French Guiana, and Suriname. Be that as it may, more extensive investigation is needed before a thorough assessment of the Cariban impact on Wajãpi can be produced. The present author’s ongoing research will also attempt to chart dialectal variation within AW, as some preliminary evidence exists suggesting that there might be local varieties within the Wajãpi Indigenous Reservation that are transitional, in relation to certain phenomena (e.g., denasalization of PTG nasal stops) between the better-known AW sub-varieties and another TG language of the region, Teko/Emerillon, that is spoken in French Guiana.