News on the Jorá ( Tupí-Guaraní ) : sociolinguistics , description , and classification

With 45 languages, the Tupí family is one of South America’s largest families. However, several gaps still remain. Some languages are already extinct and there are others for which data can no longer be collected. The situation of Jorá has reached this point. This article aims to summarize all data concerning the Jorá people and their language, parts of which were collected by the anthropologists Hanke (1959) and Béghin (1980) and other parts by the authors. On the basis of sparse data from several sources of differing reliability we attempt to classify the Jorá language using the phoneme inventory, grammatical evidence and lexical comparison. Jorá is classified as Tupí-Guaraní, closely related to Siriono and Yuki.

WHo ARE THE JoRÁ PEoPLE? in the early 1950s, the anthropologist Wanda Hanke heard about the Jorá (ocampo Moscoso, 1982), living as a group of semi-nomadic people in Bolivian Amazonia, north of the small town of Baures, in the Beni department.they were named after one of the lagoons where they lived, Lago Jorá, near Baures.Consider the following citation from Hanke: several years ago in Bolivia i heard about an unknown indian tribe, which had lived in the area located between the lakes Victoria, Jorá, Bolsón de oro and several smaller lakes.this tribe was completely unexplored and it had never been classified.According to the information, about 500 indians were living in the lake area.they had small fields on the lake banks and on the small islands, and they were said to be good and hard-working people.they lived in nice huts, clean and well-built, which showed the diligence and skilfulness of their inhabitants.
due to the fact that the named lakes are rich in fish and crocodiles, many white hunters kept coming there, and they were disturbing the peace not only of the nature, but mostly of the indians.they wanted to seize the indian women and children and enslave or sell them.that is why they killed the men and also those women who were defending themselves and kidnapped other women with children, too.then they sold them in the town of Baures and other places.(Hanke, 1959, p. 146 [translation from Czech by Katarína Marušiaková, p.c.]).
the estimation of the number of indians that Hanke (1959, p. 146) heard and mentioned in her article (500) does not concur with the statements in her letters, based on direct observation.Compare the following excerpt from a letter written by Hanke from Baures in 1953.the number given here is much more probable, according to our own investigations.
At the Lago Yorá [=Jorá] there are still ten Yorá indians, the rest are an unstudied tribe.the inhabitants of this zone have the custom to kill the Yorá men and take the women and children with them, whom they sell as slaves soon after.(ocampo Moscoso, 1983, p. 29 [our translation]).
the small number of Jorá people may have been due, in part, to the fact that they were a small semi-nomadic group to begin with.However, the ongoing genocide had probably also decimated the Jorá people.they were chased and hunted by the local Bolivians, which also included people belonging to the Baure ethnic group, as well as European settlers in the town of Baures, founded by Jesuits in 1708 4 .Many Jorá men were killed, and women and children were generally enslaved by rich farmers (cf.Hanke, 1953Hanke, , 1959;;Kelm, 1983;nordenskiöld, 1911, p. 16; and danielsen's interview collection).the local Bolivian population was afraid of the Jorá, as the following citation shows, where the Baure woman Ld is talking to danielsen (sd) about a Jorá slave who lived with her family in the 1940s (compare also Kelm, 1983: p. 21 ff.about the siriono): Ld: "He didn't speak, he was utterly quiet, he wouldn't say anything; he would speak at all." […] Ld: "no, he only did this, with his head, he didn't speak.He was really wild.i was afraid of him."[…] Ld: "My mum was afraid of him, she said to him [my father], 'why did you bring this Chori [=indian]?',she said, 'why did you not leave him there?',you see, it was when he [the indian] remembered [his family], he was already at the door, ready to leave." the Bolivians were in the business of hunting in the area and jaguar furs and crocodile skins, in particular, were sought after on the local markets.When hunting, the people occasionally came across the indians who lived deeper in the forests and did not have direct contact with the settlements of other Bolivians.Whether or not the indians attacked the hunters with their dreaded arrows, the hunters generally tried to kill them as a precaution.other encounters with the indians of the forest happened when Baure farmers walked to their remote fields.the Baure people remember many of these occasions.if there was ever a loss on the side of the Baure, they formed groups to hunt the killers in revenge.the daughter of one murdered Baure is cited as saying: "We are killing the indians, because they killed my father" (Hanke, 1959 [authors' translation]). the same practice of attacking and hunting the "wild indians" or "bárbaros", as they called them, persisted until the 1970s, according to danielsen's informants.
When the last Jorá people, who lived around the aforementioned lakes, not too distant from Baures, were discovered, their houses were destroyed, the adults were killed, and the children were brought to Baures and distributed among various families (Hanke, 1953).there they worked as household slaves, but children were also sent to school and adopted with the surname of their host family (see also the introductory citation).
today, there are only a few Jorá people living in different villages of the region, and they do not identify as members of one ethnic group, but as "civilized barbarians", integrated into the local Bolivian society. in 1951, Béghin (1980) estimated 20 Jorá dwellings, so that szabo (2008, p. 356) suggests a number of 80-100 individuals at that time, which is probably a bit overestimated.A Baure woman calculates that roughly 60 Jorá were encountered before 1950, but probably only a few of these survived.siL members are cited as counting 8 Jorá in 1972Jorá in (cf. szabo, 2008, p. 356), p. 356), which were probably the ones already living in Baures.According to Key & Key (1967, p. 127), at the time of publication, there were an estimated 10 Jorá speakers, of which 8 were claimed to be monolingual.riester gives the number of 5 Jorá in 1976 (p. 4). in her article, Hanke wrote: "When i came to Baures in 1953, i found two female slaves and one male slave from this tribe there.A boy, named rosendo, was about 12 years old, an older girl, Marqueza, about 10 years and a younger one, Maria, about 7 years old" (Hanke, 1959, p. 147) 5 .
in fact, one of the persons mentioned by Hanke still lives in Baures.Another member of the group from Lake Jorá lives in another nearby community now, approximately four hours away from Baures, but contact with the other captured Jorá was already cut off when she was captured.other Jorá people have been sent to places like Magdalena, El Carmen and ibiato, where their descendants can still be found today.

ETHNoGRAPHiC CLASSiFiCATioN oF THE JoRÁ PEoPLE
there are two possible conclusions that can be drawn from the information on the Jorá as a people: either they are a siriono subgroup or they are a group of their own, related to the siriono and Yuki. it is easy, on the one hand, to find out which of the villagers have the experience of having been caught in the forests and not belonging to the formerly missionized, such as the itonama, Baure, Mojo and others.it is difficult, on the other hand, to determine to which group the captured indians belonged.in Baures, there are two stereotypes that classify indians: those that are light-skinned and pretty and use long bows (Jorá), and those that are dark-skinned and ugly and use short bows (mostly Yanaígua, but possibly also Ayoreo or others), to give a simplified summary (cf.table 1 for details).However, of those indians with the long bows it is almost impossible to find out whether they belonged to the siriono, who are so well-known for their long bows (cf. Holmberg, 1950), or to the Jorá people, since the siriono and the Jorá lived in relatively small clans in almost the same area (cf.figure 1 and nordenskiöld, 1922, p. 223).the identification of captured indians in the area depends crucially on knowledge about their exact place of origin.
6 the names Chiriguano and siriono are supposedly derived from one common source from a Quechua word chiri 'cold ' and wa(nu) 'excrement' (for Chiriguano) and the plural -ono from Moxo trinitario (for siriono), according to dietrich (2008, p. 39) and Combès (2010, p. 130).this term may as well be related to Jorá and Chori in turn.
figure 1. this map gives the approximate location of the Jorá in Bolivia.for more details on the specific locations, see the hand-drawn map in Béghin (1980: 96-97).
the Baure use the name Chori for the Jorá, as can be seen in the citation from Ld in the previous subsection, but this term is also used for the siriono.the word Chori is argued to be a pejorative term for indians used by the Guarayo, referring to the siriono, and meaning something like "barbarians" (Kelm, 1983, p. 14).nowadays, the Baure also use the name Chori to address other indians, which they clearly distinguish from the Jorá by their appearance (cf.table 1).As dC puts it, "the Yanaígua are also Chori ". to make the situation even more confusing, the Baure use the word Woroiy in their own language to refer to uncontacted indians, and the related term Guayaro also seems to be used in a wider area of lowland Bolivia to refer to any kind of indian (also for the Ese Ejja, cf.Vuillermet, 2012, p. 44-45 or Chapacura, cf. d'orbigny, 1839, p. 217); compare also the name Guarasu, which apparently has the same origin.However, there is the possibility that the names Jorá and Chori have the same origin, both referring to the nomadic groups of the Yuki and siriono in this part of the Bolivian lowlands (Combès, 2010, p. 184), for whom the name Jore was common since colonial times (Combès, 2010, p. 177-184)  6 .table 1 summarizes the main stereotypes of the Jorá and other indigenous groups among the local Baure people, excerpted from interviews that danielsen conducted with the Baure people.
When comparing the collected information about the Jorá to what is known about the siriono and Yuki people, we can conclude that the following similarities exist between Jorá, siriono, andYuki (compared to stearman, 1984, p. 636-639, andHolmberg. 1950): long bows, whistle language in the forest, nomadic lifestyle, minimal or non-existent shelters, and a high incidence of clubfeet.However, the Jorá differ from the siriono and Yuki, of whom it has been argued that they lost the knowledge of producing fire, did not produce watercraft or other technology for traversing or exploiting waterways, did not spin or weave or use tree cotton to make thread, and lacked any interaction with other members of their own ethnic group (stearman, 1984).However, several specific features, such as the higher genetic probability of being born with clubfeet, do seem to support the possibility that the Jorá belong to the same ethnic group as the siriono and the Yuki.this is also suggested by david Jabin, who refers to the "sirionoide" as a "conjunto étnico" (ethnic complex).
the Jorá were a small group when Béghin met them.it is possible that they remained in touch with other groups of the sirionoide, of which the siriono have presumably influenced their language more than the Yuki.if we consider that the Yuki were only first mentioned as a separate group in the 1950s (Jabin, 2012), it becomes clearer that the sirionoide complex was perceived as a large group and possibly regarded as more homogeneous than it actually was.

EVALUATioN oF THE LiNGUiSTiC DATA AND CLASSiFiCATioN oF THE JoRÁ LANGUAGE
the present corpus of the Jorá language is taken from the above mentioned sources, that is in detail: 1. Wanda Hanke (WH); 2. françois-Xavier Béghin in his own publication (fXB); 3. Béghin cited by Hanke (fXB (WH) b they used the vowel [ɨ]   c their language was difficult to learn data (sd).table 2 summarizes the numbers of entries from each source, the number of phrases therein, and also the number of uniquely occurring entries in each of the word lists7 : the Jorá database is not large, but the amount of vocabulary allows for a classification of Jorá in comparison to other possibly related languages, not only on a lexical level, but also with respect to some grammatical characteristics.our word lists contains the following languages for comparison: siriono (Gasparini, our data), Yuki (Jabin, p.c.)8 , Guarasu (generally known as Pauserna, von Horn fitz Gibbon, 1955), Guarayo (Armoye, 2009), Baure (danielsen, our data), and Proto-tupí-Guaraní (PtG, sources are indicated with the proto-forms).in our database, each entry (E) has an id in table 7 of the Appendix and throughout the text, e.g.E124 is "entry 124".Hanke (1959) was the first one to compare Jorá to tupí-Guaraní languages and Baure (Arawakan), followed by Loukotka (1963, p. 42-44), who identified Jorá as a tupí language with no further specification (Loukotka, 1963, p. 8).Later on, it was generally assumed that Jorá was a language in the same subgroup of tupí-Guaraní with its closest geographical relative: siriono (rodrigues; Cabral, 2002, p. 335).

BoLiViAN TUPÍ-GUARANÍ LANGUAGES oF SUBGRoUP ii (TG2)
our attempt to classify Jorá is challenging due to the limits of the dataset. in fact, all languages of the subgroup are poorly described and their classification may have to be revisited in the future when more data have been published and analysed.With our proposal we confirm the membership of Jorá in the tupí-Guaraní (tG) branch of the tupí language family, in particular in subgroup ii (tG2), applying rodrigues' (1984/85) and our own proposed criteria.We also discuss the areal influence that Jorá may have undergone, in particular in contact with the Baure language (cf. the subsection on areal contact below).
in rodrigues (1984/85, also adopted by Jensen, 1999), the author suggests eight subdivisions within tG, based on phonological and lexical evidence.tG2 includes Guarayo, siriono and Jorá, with Yuki being added later (rodrigues & Cabral, 2002, p. 335).Guarasu is not classified any further.for us its position is interesting, since Guarasu is spoken in geographical proximity to the languages investigated in this article and it is presumably also genetically closely related, as our latest comparison shows (Gasparini et al. 2015).dietrich presents a number of detailed studies of siriono, Yuki and their classification and comparison within Bolivian tupí-Guaraní (dietrich 2002, 2007a/b, 2008). in dietrich (2010, p. 12) he proposes a linguistic area Guaporé-Mamoré-Paraguay-Paraná with Guarayo, Guarasu, siriono, Yuki and Aché9 , not mentioning Jorá.He claims that the internal classification is based on the fact that siriono and Yuki are languages with a great number of lexemes which are apparently not of tG origin (dietrich, 2002, p. 362), but his analyses of schermair (1949) resulted in the conclusion that siriono is "less 'extravagant' than what was formerly thought" (dietrich, 2002, p. 359), i.e. that it can be clearly classified as tupí-Guaraní 10 .
We can also infer grammatical information from Jorá words and phrases, because several entries are morphologically complex. in this article, grammatical features are labelled with a G and are numbered consecutively.
16 it is possible that there are actually prenasalized variants in Yuki, as suggested by Garland (o'Hagan, p.c.).
the Jorá 1sG and 2sG prefixes coincide with the PtG forms.the 1sG in Yuki and Guarayo includes the reconstructed affricate, whereas siriono and Guarasu have an initial fricative; Jorá has examples of both consonants.for the 2sG, there is only information about the set ii marker in Jorá, which is related to PtG, but did not retain the PtG nasal quality of the consonant.A rather oral realization of the 2sG marker is also observed in Yuki 16 .there is possibly also third person marking in the Jorá corpus, considering the stative verb/adjective in (5) with the initial vowel e-. the form coincides with the PtG third person marker of set ii, also found in other tG2 languages.furthermore, there is zero third person marking of set ii in Jorá, as example (6) shows, compared to siriono in (7) (dietrich, 2002, p. 365).
the person marking patterns found in Jorá are a strong argument for the tG membership from a grammatical perspective.
→ (G_5) Modifier following noun within NP: syntactically, the position of a modifier in an nP in Jorá is the typical tG position: noun-modifier, as shown in ( 12) for Jorá, and in (13) for siriono: in addition to the above-discussed criteria, we have some semantic information about the kinship terminology in Jorá. in tG there are typically gender distinctions in core kinship terms, based on the referent and on the relative genders.A gender opposition of feminine and masculine gender of referent and possibly also speaker can be expected for sibling terms in Jorá (E6 in example (1). in addition, age is used to distinguish siblings in tG.some tG kinship forms also have vocative forms. in Jorá, there is evidence of the attested tG vocative form tain 'mother' (E1), but we do not know whether other kinship terms in Jorá also show an opposition of vocative and relational forms.it is also unclear whether Jorá did not only preserve one form for 'mother', which does not necessarily have to be a vocative.We only have one form for 'son' (E5), which may also mean 'offspring', and thus refer to 'daughter' as well, which is common in tG (see also stearman, 1984 for Yuki and siriono kinship).

EVALUATioN oF JoRÁ REGARDiNG PRoPoSED CRiTERiA FoR TG2 LANGUAGES
rodrigues & Cabral (2002, p. 329) suggested some very specific phonological criteria for distinguishing subgroups of tG.Villafañe (2004, p. 196) reviewed these characteristics with respect to Yuki, siriono, Guarayo and Chiriguano, and Gasparini (2012, p. 99) reviewed them for siriono.We present an evaluation of these phonological (P) criteria in Jorá and, in passing, in Guarasu, where information is available, summarized in table 4 at the end of this section.
→ (P_1) Loss of word final consonants: tG2 languages have lost PtG final consonants17 , as seen in siriono (Gasparini, 2012).Guarayo only has final -r (dietrich, 1990, p. 13) and Yuki apparently only has final - (d)n (Villafañe, 2004, p. 197) or devoiced consonants (Jabin, p.c.). the final consonants we find in Jorá and Guarasu may also be devoiced.Jorá final consonants are -t, -s, a liquid -l/-r, and the nasals -n and -m (which may well be interpreted as marking nasality of preceding vowels), but they can generally not be related to PtG reconstructions or the other tG2 languages (e.g.E14, E98).Guarasu preserves the final PtG consonant in (E83).
→ (P_2) Merger of affricates *tʃ and *ts into one sound, either the affricate /ts/ or the fricative /s/: Even though the merger is argued by rodrigues & Cabral (2002, p. 329) to be a uniform process from PtG into tG2, the developments display differences within tG2. in siriono *tʃ and *ts evolved into /s/, but in Yuki this only happened in some cases, while in others /tʃ/ is preserved.for example, the 1sG is tʃe in Yuki and Guarayo and se in siriono and Guarasu (cf.table 3).Jorá includes examples of both consonants in 1sG.According to our analysis, Jorá underwent phonological changes similar to those in siriono and Yuki, with the rare segment tʃ as potential evidence for an incomplete merger.Guarayo conflated PtG *tʃ and *ts into /ts/ (dietrich, 1990, p. 29). in Guarasu *tʃ and *ts partly lenited into /h/, like languages in tG4 and tG6, e.g.kiha 'knife' < *kɨtse/ *kɨtʃe (E37) and piha 'night' < *pitsa (E49b). in addition, there are examples like *tsɨ, *tʃɨ 'mother' (E1), where some tG2 languages have an initial plosive t-, which shows that the position of the consonants probably played a role, and that the existence of the affricate may also result from different degrees of palatalization, which is shown to vary even within single languages of tG2 (P_10).
→ (P_3) Change from *pw to [kw] or [k]: this switch is confirmed in siriono and Guarayo. in Yuki, Villafañe claims it changed to [gʷ]  (2004, p. 201).one example in our corpus is Guarayo kwã 'finger/hand' (E15) < *pwã.Another hypothesis is that there may also have been an original *p(w)a sequence, with /p w / and /p/ as a phoneme alternation in some cases.this may be the case for 'lake/river' (E52), which is reconstructed with -pa-as *ɨpab. in Guarasu the reflex is labialized in íkua, and in Jorá ig(u)a there is an alternation between [g] and [g w ] (cf.E135).there is also the bird name tsu-ru-kwa ('(kind of) grey pigeon', E165, fXB) in Jorá, but this is surely not enough evidence to claim that the labialized velar consonant /kw/ was phonemic, and that it is related to PtG *p(w)a.
→ (P_4) Retention of *pj(?): this proto-phoneme suggested by rodrigues & Cabral (2002, p. 329) only occurs in one reconstruction, *epyak 'to see' (E167) and it does not occur in the Jorá corpus.the sound [pʲ] does not exist in the siriono, Yuki and Guarayo phonetic inventories either.in Guarasu and Guarayo it may be present in the verb 'to see', but there it could also be analysed as a consonant-vowel sequence.this criterion does not hold for the whole subgroup.
→ (P_5) Moving of stress to the penultimate syllable: PtG supposedly had word final stress, and tG2 moved stress onto the penultimate syllable.this criterion holds for siriono, Yuki, Guarasu and Guarayo18 , but it does not seem to be true for Jorá. the corpus provides many examples with an accent on the ultimate syllable -62 entries out of 165.since these 62 make up the majority of disyllabic words and since many of the words are clearly cognates of PtG forms (as e.g.Jorá tatá 'fire' (E41), this criterion cannot be confirmed for Jorá.
for a better understanding of the internal relations of tG2, we suggest additional P_criteria: → (P_6) Loss of glottal stop: dietrich (2007a) remarks that Bolivian tG generally lost the PtG glottal stop.Based on available reconstructions, four examples of Jorá illustrate this loss: E10, E31-33, E51, E59-60, E139/149.While siriono and Yuki also lost the glottal stop, Guarayo did not.
→ (P_8) Nasals and post-oralization: this feature is presented in Cabral & rodrigues (2011) as a general tG feature, where they claim that in tG1, tG2, tG3 and tG4 all languages "display alternation of nasal and post-oralized stops, the former occurring in nasal contexts and the lat[t]er in oral ones" (Cabral & rodrigues, 2011, p. 74) namely "[t]he alternation of nasal versus post-oralized consonants is motivated by the nasal or oral context following such consonants" (ibid: 76).rodrigues & dietrich (1997, p. 268), reconstruct the PtG nasal consonants *m, *n, and *ŋ.these phonemes may show alternative post-oralized variants mb, nd, and ng in a single language or vary in realization in different languages.Yuki, for example, seems to be less nasal in general, i.e. the development may extend to b, d, and g, respectively (dietrich, 2008, p. 41).this is presumably also the case in Guarasu. in our specific comparison, we observed a relative loss of nasal quality in Jorá and Yuki -e.g. in 'snake' (E84) and '2sG' (E116) -, whereas siriono preserves it (cf.E7, E18, E135, E157, E166).
strikingly, there are also some counterexamples.in Jorá (E157), there is one example where m corresponds to b in siriono.the presentation of alternative forms, shows the variability of the phonetic realization of the same word with respect to post-oralization of nasals, as e.g. in Jorá 'people' (E7) may be realized as mbia or ma, and 'fly' (E166) can be mberu and meru in Guarayo.→ (P_9) Reconstruction or fusion of /i/ and /ɨ/: the two phonemic vowels *i and *ɨ of PtG have partly merged in tG2.Yuki lost the central vowel completely (dietrich, 2007b, p. 206), whereas in siriono the opposition has been reversed (i.e.*ɨ in PtG is now i and *i is now ɨ; see Crowhurst (2002) and E42, E44, E81, E87, E111, E133).E111 and E133 show the reversal of the vowel opposition in siriono, on the basis of a reconstructed plain *i.Guarayo keeps the reconstructed central vowel.Guarasu seems to have retained the central vowel, too.Jorá, however, does not show any evidence of the vowel ɨ. the importance of the absence of this phoneme in the data, though, may be trivial.We assume that ɨ existed in Jorá, just like in siriono, because it is a characteristic that Baure people usually point at when they talk about this language: sd: "You didn't understand (what they spoke)?"rP: "Because they didn't speak well, only this ɨ, ɨ, ɨ was what the indians made."(Baure speaker rP in Baures, 5 of september, 2009, in interview with danielsen) We can imagine that Hanke and Béghin were probably not able to distinguish the two vowels.→ (P_10) Palatalization and friction of *k and *t: A distinguishing feature within tG2 is the affrication of *k as [tʃ], sometimes only palatalized as [k j ] in some of the languages.in addition, there may be more or less palatalization and friction of *t, generally before a high vowel (dietrich, 2007b, p. 208). dietrich (2007b, p. 208) noticed a general tendency towards palatalization in the word 'maize, corn' (E111).the tendency is even more advanced in siriono [s] and Jorá [ʃ].Particularly interesting is the case of 'milk' (E18) and 'breast' (E10), where Jorá shows an affricate, siriono and Yuki have palatalized consonants, and Guarasu and Guarayo are close to PtG without any palatalization.further possible examples are E54, E63, E101, E108, E111. it seems that the process is most advanced in Jorá which is similar to siriono and Yuki.We can also observe that palatalization proceeded in time and occurred more recently, and older forms in siriono present less palatalized variants (compare schermair 1957).→ (P_11) Evolution of *(n)y: the example of 'moon' (E44) shows different developments of word initial *y. in fact, it seems that we can talk about two major processes, one starting off from a *ñ, changing into [n(y)] in Jorá (E96, E31-33), and one starting with an oral *y, corresponding to [dʒ] in Jorá (E44, E62, E76) (o'Hagan, p.c.). the word proposed for comparison, *yuka '(i) kill ' (dietrich, 1990, p. 21), dutʃa in Jorá (E152), may be evidence of the oralization of the initial consonant here.Also Guarasu changes the oral *y into [d].Unsurprisingly (compare P_8), siriono is more nasal than Yuki (E31-33, E44, E96).Yuki corresponds more or less to Guarayo, generally the most conservative language of tG2.realization of *ñ shows a different correspondence with [s] in Jorá kusa 'woman' (E2b), however, this is in word-medial position (o'Hagan, p.c.).
All  the phonological criteria discussed above suggest three clusters within tG2, where Guarayo is different from all the others, but more conservative with respect to the PtG features (P_2, P_3, P_6, P_9).Guarasu often shows tendencies different from the rest, in particular for less extreme palatalization (P_10) diachronically, and the change of PtG *tʃ and *ts to /h/ (P_2).the latter makes Guarasu more similar to tG4 or tG6 languages, which could mean that either it belongs to another subgroup of tG or that it changed due to contact with languages of other subgroups.siriono, Yuki and Jorá seem to form a subgroup (P_2, P_6, P_7, P_10).Jorá stands out in particular for the lack of stress shift (P_5).
the idea of the split of tG2 is also supported by the lexicon, as compared in table 5, where again, Guarayo and Guarasu are different from Jorá, siriono, and Yuki: SPECiFiC CHARACTERiSTiCS oF JoRÁ there are several specific features of Jorá that the other languages do not have.of course, with the sparse data we are dealing with, it is difficult to find particular features that are not described for any other related tG language.We want to be cautious with our claim here because examples may also have been translated badly or misunderstood.
firstly, there are some examples of vocabulary that have been reconstructed for PtG for which Jorá seems to have a completely unrelated form (E14, E37).
We also noticed a recurring final syllable in Jorá, for which we suggest an analysis as a kind of derivational suffix, possibly adjectivizing: -(t)ʃi ~ - (d)ʒi. it occurs in the construction of 'star' (= lit.'fire of the moon') on the dependent noun in the construction ( 14), and in didintʃi 'child, little' (E4), dadatʃi 'pretty' (E129), taratʃi 'foreigner' (E9), and a few more. in contrast to Jorá (14), in siriono, nyasi tata 'star' does not contain any affix presumably constitutes a compound of two nouns like in the other tG languages.the Jorá suffix occurs in some expressions that could be adjectival (or stative verbs), but also in nouns, which could have been derived (E24, E40 idadʃi 'path'). in some cases, we may be dealing with a palatalized variant only (P_10).stress in Jorá generally falls on the last syllable of the stem (P_5) and it is never marked on this syllable, pointing to the analysis of -tʃi as a suffix more than a part of the stem.An alternative analysis could be that the suffix is actually a diminutive, possibly related to Baure -tʃi 'diM', at least in some cases.

AREAL CoNTACT
the tG languages remained in contact with one another and with genetically unrelated languages in their region (dietrich, 2007b).the Mamoré and Guaporé rivers form a zone of intense contact (Crevels & van der Voort, 2008) and Jorá is located in this area.towards the south of tG2, there are Chiquitano (known as Bésɨro, Macro-Gê) and Ayoreo (Zamucoan).to the west, there are the Mojeño languages (Arawakan), and the isolates Yurakaré, Movima, and itonama.to the north, close to Jorá, there is the Baure language (Arawakan), and in former times there were presumably Chapakuran groups to the north (Moré) and south (tapakura, napeka, Kitemoka) of the Jorá (Wienold 2012).Hanke (1959) and Loukotka (1963, p. 42) propose that Jorá shows "some influence from the Baure language".At that time, however, the Baure language had not been investigated yet.since danielsen has studied Baure in particular detail, we revisit this proposal again here.in her word list (1959, p. 150-154), Hanke marks three entries as possibly related to Baure (the comparative Baure lexicon has been updated from danielsen's corpus), which are E5, E10, and E156.Loukotka (1963, p. 43) adds that the Jorá word in 'water' (E42) could likewise be related to Guaraní y (/ɨ/), tG, or Arawakan Baure in. it is difficult for us to decide on this point, but an additional alternative could be that the Baure word in 'water' originates in tG as well.danielsen counts 11 items with a possible Baure source; however, most of them are tentative at best: E2a, E5, E10, E17, E42, E83, E98, E119, E140, and E156 (most probable forms in bold and Baure notes in table 7 in the Appendix).
for E42, 'water', we could probably say that Hanke noted a Baure word, but Béghin did not.interestingly, Crevels & van der Voort (2008, p. 164) mention that "the only possible similarity found in the 11 Bolivian languages [of the Guaporé-Mamoré area] is the word for 'water' in Baure (Arawakan) and Yuki (tupí-Guaraní)".the word mitʃkiri 'small' in Baure may be related to Jorá mitʃi (E119), but this root is largely diffused in all south America for the word 'cat' (Kiddle, 1964).
for the phrase in (15), we cannot say for sure if the two words separated by a comma were meant to be alternatives or mark a pause. in any case, the first word could be related to Baure (16).if taken as one phrase, we could imagine this to be an example of language mixing in Jorá: (15) three possible contact scenarios of Jorá with Baure could be proposed: 1. bilingualism in the area for a relatively long time; 2. language mixing with Baure people integrated into a Jorá group; 3. language decay in Jorá and contact with the Baure language in Baures after the capturing of the Jorá.We do not know how much the Jorá children had already been in contact with Baure speakers, and therefore when they gave a word, whether they might have given a Baure translation in some cases.We have to be careful in assuming mixing with the Baure language.As our data analysis has shown, the tG character can definitely be identified, and the lexicon seems to be clearly tG2.
finally, while there is no evidence of spanish language contact with Jorá in our corpus, there is one possible case of Portuguese language contact: inóit 'night' (E49) was possibly borrowed from the Brazilian Portuguese phrase CLASSiFiCATioN in this article we have shown in detail that Jorá is a tG2 language, and we have also discussed the relation of languages within this subgroup.As claimed above, Guarayo is usually the most conservative language within tG2, and Guarasu, even though geographically proximate, contrasts strongly with the Bolivian tG languages.one of the striking observations for tG2 was the shift of word stress to the penultimate syllable (criterion P_5), a claimed areal tG feature that is supposedly "already quite old" (dietrich, 2007b, p. 215, [the authors' translation]).nonetheless, Jorá does not seem to have shifted its stress, but keeps it on the final syllable (cf. the subsection: Evaluation of Jorá regarding tG2 criteria).this could mean that Jorá split off earlier from the other languages of tG2, and similarities could be the results of language contact afterwards.However, it is in general difficult to compare languages of tG2, because they diverge so strongly for some characteristics.
on the basis of our findings, we assume that all three languages, Jorá, siriono and Yuki, derive from one protolanguage and form a sirionoan group.Ethnographic descriptions support the thesis of the sirionoide group (Jabin, 2012) and the membership of Jorá in this group (cf.table 1), and linguistic comparison reinforces this claim of a group forming a large dialectal continuum.
stearman (1984, p. 649) presents a possible relation among Bolivian tG ethnic groups and dietrich (2008, p. 47) tried to unite all Bolivian languages.A new collaboration arose to clarify this claim with the team developing the tG Comparative Lexical database (o'Hagan et al., 2015).We integrated Jorá into a new phylogenetic classification and searched for borrowings through a list of innovations for sirionoan languages (Gasparini et al. 2015).it appears that tG1 and tG2 do not form two nodes but three clades: Guaranian, Guarayo-Guarasu and the sirionoan languages.
With regard to lexical data, Guarayo and Guarasu show less differentiation than sirionoan languages, which is illustrated by the length of the lines.A project started late 2014 by the same team compares morphosyntactic data of tG and may shed additional light on certain aspects of the classification.Unfortunately, this new project will not include Jorá and Guarasu because of the scarcity of morphosyntactic data for those languages.

CoNCLUSioNS oN THE JoRÁ LANGUAGE AND PEoPLE
it is always difficult to classify and characterize a language on the basis of sparse data.However, it was possible through systematic comparison to challenge the phonetic criteria for the tG2 group and to describe and compare the pronunciation of the Jorá language with other tG2 languages.Jorá data are not only wordlists with lexical elements but contain a few phrases that shed light on some grammatical criteria including word order within a nP and person marking on verbs and nouns.the lexicon was enough to propose cognates with the tG languages in a lexical comparison and include the language in a phylogenetic classification based on lexicon.
Phonetic criteria, grammatical criteria and lexical comparison allow us to integrate Jorá into a larger complex, together with siriono and Yuki.All the languages of tG2 still need more intensive investigation where possible.Guarasu, Guarayo and siriono, are now the focus of documentation projects with grants from the Endangered Languages documentation Programme (ELdP).
We have tried to gather as much data for Jorá as possible.nonetheless, there is a small possibility that we will be able to collect more data in the field from one remaining Jorá woman.this means that more studies of tG2 can still be expected in the future.And in addition to this, there will be more descriptions of other tG languages and Bolivian languages of the area, so that a broader comparison is another task.
the moon'; translated as 'star'

table 1 .
stereotypes among the Baure about indigenous groups of the area (interviews conducted by danielsen in 2009 and 2011).