CC BY-NC-ND
35
Audrey Nikulin
University of Brasilia; andre.n.guzman@gmail.com
A phonological reconstruction of Proto-Cerrado (Je family)1
This is the second paper in a series on the historical phonology of Macro-Je languages. In this work, I examine the sound correspondences between the languages of the Central Je branch, Xavante and Xerente, in order to arrive at a reconstruction of Proto-Central Je. I further compare it to my reconstruction of Proto-Northern Je (Nikulin 2016b) and propose a phonological reconstruction of Proto-Cerrado, the most recent common ancestor of Proto-Central Je and Proto-Northern Je. The paper also includes a non-exhaustive list of Cerrado etymologies.
Keywords: Je languages, Macro-Je languages, Xavante, Xerente, language reconstruction, comparative method
1. Introduction
In an earlier work (Nikulin 2016b), I have proposed a phonological reconstruction of ProtoNorthern Je, the proto-language of the branch that comprises Apinaye, Kayapo (Mebengokre), Suya (Kisedje), Tapayuna, Timbira and Panara languages spoken in Central Brazil. In this paper I offer a reconstruction of Proto-Cerrado (PCerr)2, the immediate ancestor of ProtoNorthern Je (PNJ) and the proto-language of its coordinate branch, Central Je (PCJ).
Before proceeding to the main body of this paper, some comments on my PNJ reconstruction (Nikulin 2016b) are necessary. These concern the internal phylogenetic structure of the Northern Je branch and individual correspondences.
1.1. Internal structure of Northern Je
Recent lexicostatistical investigation has shown that the Northern branch of Je languages has the internal structure shown in Figure 1.
Figure 1. Northern Je languages
1 I am grateful to CAPES (Coordenagao de Aperfeigoamento de Pessoal de Nivel Superior) for providing a scholarship to carry out the present study.
For abbreviations, transcription conventions and lexical sources, the reader is referred to Nikulin 2016b. The following additional abbreviations are used here: DU = dual, GNR = generic possessor, INCL = inclusive, IPF = im-perfective, NOM = nominative, SG = singular, PL = plural.
2 Cerrado languages have been also called Amazonian Je (Ribeiro and Voort 2010: 549) and Northern Je (Ramirez et al. 2015: 261) in earlier literature.
Journal of Language Relationship • Вопросы языкового родства • 15/3 (2017) • Pp. 147-180 • © The authors, 2017
Contrary to my earlier claims, Timbira is not likely to form a node with Apinayé and Kayapó (called AMT in Nikulin 2016b)3. Instead, it probably was the first language to split off from Proto-Core Je. This must have occurred sometime between the III and VIII centuries CE 4 (70-82 % matches on the 110-item Swadesh wordlist). The split between Apinayé-Kayapó and Suyá-Tapayúna should be dated sometime between the VII and X centuries CE (79-86% matches). It is also possible that Timbira, Apinayé-Kayapó and Suyá-Tapayúna diverged from each other simultaneously around 700 CE, in which case Core Je would have a rake-like structure (cf. the working hypothesis in Nikulin 2015). The revision of the tree has been possible thanks to the inclusion of additional lexical data sources for modern languages not considered in Nikulin 2016b: Vasconcelos 2013, Bardagil-Mas 2016, Lapierre et al. 2016a (for Panará), Ham 1961, Albuquerque 2011, 2012 (for Apinayé), Camargo 2015 (for Tapayúna), DMK (for Suyá), Silva 2011, 2012 (for Pykobje), Castro Alves 1999 (for Apaniekrá), Popjes and Popjes 1971 (for Ramkokamekrá).
The split of Northern Je into Core Je and Panará probably occurred sometime between the III and the V centuries BC (55-60 % matches on the 110-item Swadesh wordlist), much earlier than the split of Core Je. This is in line with my earlier views on the internal structure of the Northern Je branch (Nikulin 2015, 2016a, 2016b), but contrasts sharply with an alternative hypothesis, according to which Panará would form a node with Timbira, 'Savannah Je' (Lapierre et al. 2016b, Lapierre 2017). The main argument in favor of the Savannah Je hypothesis is that both Panará and Timbira share a non-trivial, typologically rare and seemingly unnatural sound change: the devoicing of PNJ prenasalized voiced stops (*nb, *nd, *"4>, *"g). However, this sound change should be understood as part of a more general trend in Panará and Timbira historical phonologies: all PNJ voiced stops, including the oral stops *b, *4>, *g, became voiceless in both languages (Nikulin 2016b: 171-173); no mention of this fact is made by the proponents of the Savannah Je hypothesis. Note that, although devoicing of prenasalized stops is indeed uncommon cross-linguistically, as correctly observed by Lapierre (2017), there is nothing uncommon about the situation in which a general stop devoicing process targets stops in all environments, including the position after a nasal segment (cf. Old High German fintan 'to find', bintan 'to bind', where t comes from an earlier *d via the High German consonant shift). More crucially, the devoicing of voiced stops in Timbira counterfed the sound changes > h, *k > kh, unique to Timbira. This fact confirms that the voiced stop voicing in Panará and Tim-bira occurred independently.
1.2. Other additions
Major additions to my PNJ reconstruction (Nikulin 2016b) include some amendments based both on new sources and on new observations.
3 The existence of this clade has also been cautiously suggested by Carvalho (2016: 70).
4 Annotated Swadesh wordlists, created within the GLD (Global Lexicostatistical Database) project, are scheduled to be available online at <http://starling.rinet.ru/new100>). Note that the Proto-Northern Je word *ndiwi 'new' is a probable loanword and is marked as such in the database; it was considered as a native item in the lexi-costatistical calculations for Northern Je languages.
For divergence datings, I accept the flow glottochronological model advanced by Vasilyev and Saenko (2017) and the formula proposed by the authors: N(t) = e-0'61t(1 + 0.61t), where N(t) is the match percentage within the 110-item Swadesh wordlist and t is the time depth of the divergence event expressed in thousands of years. The value of the loss coefficient (0.61) was calibrated on Romance material.
1.2.1. Panará
New data that became available to me after the publication of my PNJ reconstruction, especially a recent article on the development of PNJ *r in Panará by Carvalho (2016), challenge a number of statements that can be found in Nikulin 2016b.
1.2.1.1. PNJ *r/s. Most importantly, the sound law PNJ > pnr y, based on a single cognate set PNJ *qr5t5 'toucan' > pnr yo-kwekwe, yo-sd, is now known to be erroneous. The true Panará reflex of this PNJ root is inkyo pepeti (Andrés Salanova, p.c.). The same PNJ cluster must be reconstructed for another cognate set proposed by Carvalho (2016: 59): PNJ *qr5 'to sprout' > pnr kyo, kay yro (from which PNJ %=i=yfo-to 'sprout' > api 0=i=^ro-to, tim h=i=ro-t is derived), though the absence of prenasalization in Panará remains unclear. Note that PNJ > pnr nk in all environments, including the position before nasalized vowels.
1.2.1.2. PNJ *ka- and * a-. Another fact that I failed to mention in Nikulin 2016b is that the outcomes of word-initial unstressed PNJ *ka- and *a- are exactly the same, suggesting a merger and a subsequent split development in the history of Panará, as shown in Figure 2.
Figure 2. The development of PNJ word-initial *ka- and *a- in Panará
PNJ *a- *a- (before voiceless consonants)- pnr a-
*a-
PNJ *ka- *n3- (before prenasalized consonants) (phrase-initially)
The correspondences involved in this scheme are illustrated by examples (1-2), most of which are extracted from published sources (Nikulin 2016b: 171; Carvalho 2016: 58).
(1) PNJ *ka- and *a- > pnr n3- / =r3-
a. PNJ *ka=ngro 'warm' > pnr ns=kyo / =r5=kyo;
b. PNJ *kanbro 'blood' > pnr n3pyu / =r3pyu;
c. PNJ *kaq3 'snake'5 > *kang3 > pnr n3k3;
d. PNJ *ang3 'sedge seed6' > pnr n^;
e. PNJ *a=ngro 'peccary' > pnr n3nkyo / =r~3nkyo, etc.
(2) PNJ *ka- and *a- > pnr a-
a. PNJ *ka^3t3 'cotton' > pnr as3ti 'cord';
b. PNJ *ka^uwa ~ *ka^wa 'mortar' > pnr asua 'pestle';
c. PNJ *kapri 'sad' > pnr apri-pe;
d. PNJ *kapr3t3 'turtle' > pnr apy3n;
e. PNJ *akro 'vine, fishing poison' > pnr akyo, etc.
Note that Southern Kayapo wordlists capture a stage of Panara when the prothesis of n-had not yet occurred. For example, Barbosa (1918: 62, 81, 83) gives <a(n)kio> 'warm', <ampio> 'blood', <anhan> 'snake', <ankio> 'peccary'. This means that the prothesis of n- and its subsequent flapping are very recent and should be dated to the XX century.
1.2.1.3. Correspondence Proto-Core Je *e ~ Panara 3. Carvalho (2016: 60) cites two cognate sets that cannot be easily accounted for with my PNJ reconstructions. These are reproduced below in (3).
5 Glossed erroneously as 'blood' in Nikulin 2016b.
6 This item is of significant cultural importance for the speakers of Cerrado languages. It denotes the seeds of a plant called tiririca or capim-navalha in Portuguese, which are used for making collars.
pnr r3- (phrase-medially)
(3) Proto-Core Je *e ~ Panará 3
a. Proto-Core Je *preke 'tall' ~ pnr py3 'big, tall';
b. Proto-Core Je *kanbreke 'red' ~ pnr n3npr3 / ¿3p3.
If only these comparisons are correct, we are dealing here with a correspondence Proto-Core Je *e ~ Panará 3 (note that Proto-Core Je *e is known to correspond to Panará e and Proto-Core Je *3 regularly corresponds to Panará 3). It must go back either to PNJ *e or to PNJ *3; either Proto-Core Je or Panará must have undergone a conditioned vowel change. In order to determine the directionality of this change, external data might turn out to be useful. The word for 'red' happens to have likely cognates in Central Je: PCJ *pre: 'red' > xav pre:(-di), xer pre(-di), Salinas Xavánte hoy=pye, Xakriabá (Saint-Hilaire) oi=pre-de <oipredé>, Akroá si-kuca=brd <schikutzabro>. PCJ *£ is known to correspond to PNJ *3 (see subsection 3.2), so the vowel in Panará appears to be more archaic. However, the decision to reconstruct PNJ *3 for the correspondence in question is problematic: PNJ *r is known to have yielded Panará y before non-front vowels (Nikulin 2016b: 170-171; Carvalho 2016: 66), which did not occur in PNR n3^pr3/r¡npr3. Reconstructing a front vowel for the same correspondence would not solve the issue: in this case pnr *pr3 would be expected instead of pnr py3.
The correspondence remains therefore unexplained and awaits further investigation.
1.2.2. Apinayé and Kayapó
In my description of the distribution of Apinayé and Kayapó reflexes of PNJ (Nikulin 2016b: 172) I failed to observe that the reflex ? is most consistently found in PNJ stems whose initial consonant is a non-alternating % The most salient examples are provided in (4) below.
(4) PNJ > API, KAY ?
a. PNJ 'bone' > kay ?i (Apinayé innovated and now uses both yi and i);
b. PNJ *£o 'leaf, a hair' > api ?o, kay ?o;
c. PNJ 'seed' > api ?i, kay ?i;
d. PNJ / 'to deceive' > api ?eyc / ?et (but kay et).
Note that these stems do not behave as usual class II stems in terms of Rodrigues (2012), since they suffer no modification when an internal argument expressed with a full NP immediately precedes them 7. Nevertheless, their initial consonant changes to a reflex of PNJ *y or *4> when their internal argument is expressed with a person-marking prefix (e.g. api ip-m3 a=ceyc ket-ni 'don't lie to me!', kay i=yo 'my (single) hair'). This fact leads Salanova (2011: 79) to consider that these consonants are present in the underlying representation of the person-marking prefixes in these languages, arguing that in Apinayé they surface even before consonants. However, the latter observation is valid only for the first person prefix (ic-), but not for the second person prefix (a-). Be that as it may, the presence of these consonants in the inflected forms of the stems under consideration is likely an innovation caused by analogy with true class II stems.
7 The stems that belong to class II are expected to exhibit the following initial consonants:
PNJ Apinayé Kayapó
after a full NP (uninflected): *y, % z, c У, 3
after a person-marking prefix: *y, % z, c У, 3
elsewhere (third person): *t 0 0
These predictions are valid both within frameworks that treat these segments as prefixes (Rodrigues 2012) and within frameworks that consider *y ~ *4 and their reflexes as integral parts of roots.
1.2.3. Tapayuna
Camargo (2015) offers the most complete description of Tapayuna, but this work has not been taken into consideration in my earlier research. It helps fill one of the gaps in the description of the historical phonology of Tapayuna found in Nikulin 2016b, as shown below in (5).
(5) PNJ *g > tap k, SUY k
a. PNJ *ga '2.NOM' > tap, suy ka;
b. PNJ *ga 'to roast (finite)' > tap, suy ka;
c. PNJ *gu '1INCL.NOM' > TAP ko, SUY ku.
There is one curious fact about the transcription system employed in Camargo 2015: the phonemes m and w recognized by the same author elsewhere (Camargo 2010) are systematically rendered as w, and the character m is not used in the transcriptions altogether. This is not stated explicitly anywhere in the cited work. It is not entirely clear whether the correct explanation of this fact involves transcription issues or an ongoing sound change, but it must bear some relation to the apparently irregular instances of PNJ *"b > tap w, as attested by Camargo (2010). Examples of the latter are taken from (Nikulin 2016: 175) and reproduced in (6) below.
(6) PNJ *"b > tap w:
a. PNJ *(a=)kanb3t3 'night' > tap a=gawdrS;
b. PNJ *nbe^ni 'honey' > tap wey (but tap nbey-ti 'bee').
Further fieldwork on Tapayuna should clarify the issue.
Another claim in my earlier work that has been invalidated by new data referred to the impossibility of CCC-structured onsets in Tapayuna (Nikulin 2016b: 176). These are now known to be allowed, cf. Tapayuna ng^wa-to 'moriche palm leaf' (although the former root is attested as ng^uwa 'moriche palm' in isolation).
1.3. Structure of the paper
The remainder of this paper is organized as follows. Section 2 deals with Central Je languages and the reconstruction of Proto-Central Je. Onsets (subsection 2.1), syllable nuclei (subsection 2.2) and syllable-final and word-final phenomena (subsection 2.3) are tackled separately. In section 3, I compare my reconstructions of Proto-Central Je and Proto-Northern Je and arrive at a phonological reconstruction of Proto-Cerrado onsets (subsection 3.1), nuclei (subsection 3.2) and codas (subsection 3.3). A non-exhaustive list of Cerrado etymologies is provided in section 4. I conclude with section 5 where I discuss the findings presented in this paper and suggest topics for future research in Cerrado historical phonology.
2. Proto-Central Je
Central Je (Akwe, Akuwe8) languages constitute a phylum of closely related languages formerly spoken over a vast area in Central Brazil, centered on the Tocantins watershed and reaching the headwaters of the Parnaiba and the Sao Francisco in the east (since 19th century, Xavante has also been spoken as far west as the Araguaia River due to a recent expansion
8 The endonyms of Xavante and Xerénte are, respectively, îaîuwë and akwe. These forms are reconstructible to Proto-Central Jê (PCJ *akuwe according to the correspondences described in this section below).
event). Of these, only Xavante and Xerénte are currently spoken. Akroa, Xakriaba and the Salinas dialect of Xavante are now extinct and their attestation is limited to short wordlists collected by Auguste de Saint-Hilaire, Wilhelm Ludwig von Eschwege, Karl Friedrich Philipp von Martius, Johann Baptist von Spix (Martius 1867) and Paul Ehrenreich (Ehrenreich 1895).
Akroa and Xakriaba wordlists have been tackled by Carvalho and Damulakis (2015), while some peculiarities of the Salinas dialect of Xavante are discussed by Nikulin (2015: 27-29). Fragmentary nature of available data, combined with low internal diversity of the Central Jê group, makes the contribution of Akroa, Xakriaba and Salinas Xavante data to the reconstruction of Proto-Central Jê minimal. For the most part of this subsection, only Xavante and Xeré-nte data are considered below.
Xavante and Xerénte share much of their irregular morphophonology. Phenomena like alternations and prefix substitution at the left margin of the stems, as well as nominalization via suffixing are also found in Northern Jê (and are easily projected onto Proto-Cerrado). Another phenomenon typical of Central Jê languages is the existence of special utterance-final allo-morphs: a number of stems undergo drastic changes when they are preceded by a pause. These will be considered at the end of this section (2.3).
2.1. Onset
The correspondences between Xavante and Xerénte onsets are rather straightforward and are exposed below in Table 1.
Table 1. Proto-Central Jê onsets and their reflexes in Xavante and Xerénte
PCJ XAV XER
*p p p
*pr pr pr
*t t t
*c c ~ s g ~ s
*k ? k
*kr ?r kr
*kw ?w kw
*km ?m km
*b b b
*br br br
*d d d
*3 3 ~ z9 \ ~ z, ^d ~ zd, d
*m m m
*mr mr mr
*n n n
*y y 10 n
9 Quintino (2000: 39, 46) analyzes both /z/ and /j/ as phonemes, stating that there is some variation between the two in oral contexts. However, this claim is not illustrated with any minimal or near-minimal pair.
10 According to Quintino (2000: 39-40, 2012: 125), before back nasal vowels the allophone [q] is found instead of [fi]. This phenomenon is not attested by other researchers. In fact, Pickering (2010: 63) explicitly states that in his data [j] is found both before back and front nasal vowels, noting that Quintino's data come from speakers from
PCJ XAV XER
*w w w
*r r r
*h h h
*0 ? 0
Note that voiced stops and nasals are in complementary distribution both in ProtoCentral Jê and in Xavante: at least in the onset position, voiced stops occur in syllables with oral nuclei, whereas nasals occur only in syllables with nasal nuclei (McLeod 1974; Quintino 2000: 39-41, 2012: 124-125; Pickering 2010: 62, 73-77, 76-77)11. However, just like in Nikulin 2016b, I prefer to distinguish these allophones in my transcription. Note that the allophony in question was phonemicized in Xerénte due to syllabic reduction that eliminated some of the nuclei that had triggered the choice of the allophone in Proto-Central Jê (Mattos 1973).
Although most correspondences presented in Table 1 are trivial, several comments are in order.
2.1.1. The reconstruction of PCJ *c and *3
Xavante, Akroa and Xakriaba reflexes of PCJ phonemes reconstructed here as *c and *3 suggest that they should be reconstructed as stops or affricates rather than fricatives. The phonetic realization of the reflexes is discussed below.
McLeod (1974) describes the Xavante reflexes as fluctuating between [c ~ c ~ s ~ s] and [3 ~ 3 ~ z ~ z] respectively, Quintino (2000: 44-45, 2012: 45) attests [c ~ s] and [3 ~ z] in plain speech register, while in Pickering's (2010: 63) data only fricatives ([s] and [z]) are found. According to Xavante speaker Euzebio Prowari (apud Pickering 2010: 63), the affricates [c] and [3] are a characteristic feature of the speech of the Xavantes from the Indigenous Areas Pimentel Barbosa and Areôes. I transcribe the phonemes in question as c and 3 hereinafter, since affricates are more likely to represent the more archaic state than fricatives.
In Salinas Xavante, PCJ *c is reflected as s (sometimes s), while *3 is usually reflected as 9 (sometimes s).
In Xerénte, PCJ palatal stops normally yielded fricatives. These fricatives are attested as retroflex [g, zj in earlier works (Mattos 1973; Krieger & Krieger 1994: XIII-XIV; Braggio 2004: 269), but in modern Xerénte they are mostly articulated as alveolar ([s, z]) according to more recent works (Sousa Filho 2007: 73; Souza 2008: 72-74; Grannier 2009: 252; Frazao 2013: 56; Co-trim 2016: 59). It is likely that we are dealing here with a very recent sound change. I shall transcribe Xerénte fricatives as retroflex in this paper.
In some words Xerénte displays aberrant reflexes of PCJ *3 (namely, xer %d or d), for which I am unable to offer an explanation. Note that these irregular reflexes often occur in variation with plain fricatives or with each other, e.g. kba%di-kre ~ kbadi-kre 'hammock', dako ~ %ako 'to climb', dapaka ~ %apaka 'to desire', darôtô ~ %ardtd 'to jump', %awre ~ dawer 'big', ~ da 'to stand', etc. Sousa Filho (2007: 255-256) analyzes alternating d- and as allomorphs of the same mor-
the Indigenous Area Pimentel Barbosa, whereas his own data come from residents of the community of Sao Marcos. For sake of simplicity, I follow McLeod (1974) and transcribe p.
11 Quintino (2012: 123-124) provides a minimal pair ([da-di] 'GNR-belly' vs. [na-di] 'mother-lMP'), but concludes that in the latter case the vowel of the root is underlyingly nasalized. All other researchers invariably list the form nS 'mother'.
pheme (the 'relational prefix' /z-/), suggesting that the occurrence of the forms with d- and t- is not grammatically conditioned, but does not provide any explanation for the causes of the variation in question.
It is important to note that in a number of stems stem-initial PCJ *c alternated with (in oral contexts) or (in nasal contexts). The *c-initial allomorph occurred only after the second person prefix *ay-, after the third person prefix *i- and in third person forms without an overt prefix, as is still the case in Xerente (Sousa Filho 2007: 247-255) and Xavante (Estevam 2011: 138; Hall et al. 1987: 286-287). Some examples are provided in (7) below.
(7) PCJ *3 and *c
a. PCJ *-ja 'to stand' > xav -ja, xer t ~ -da;
b. PCJ *-jadawa 'mouth' > xav -jadawa, xer -Tdawa;
c. PCJ *-poyto 'tongue' > xav -poyto ~ -potto, xer -noyto;
d. PCJ *ay-ca 'you stand' > xav a-ca, xer ay-§a;
e. PCJ *ay-cadawa 'your mouth' > xav a-cadawa, xer ay-§dawa;
f. PCJ *ay-coyto 'your tongue' > xav a-coyto, xer ay-§oyto.
While most authors agree that the phenomenon in question can be described as a stem-internal alternation, Sousa Filho (2007) considers that Xerente -t-, -d-, -n- and -§- are separate prefixes ('relational prefixes' in the author's terms). In this work I adopt the former analysis.
In Akroa and Xakriaba, PCJ *c seems to have been reflected as s (sometimes s in Xakriaba), while PCJ appears to have yielded c (sometimes c in Akroa). In the latter case the data also allow for a voiced interpretation of the reflex(es) (Carvalho and Damulakis 2015: 40).
2.1.2. Debuccalization of PCJ *k in Xavante
It is well-known that *k has been historically debuccalized in Xavante, changing to ? (Rodrigues 1999: 178; Carvalho and Damulakis 2015: 24-26). It did not affect the Salinas dialect of Xavante, as attested by Ehrenreich (1895), nor is it visible in the wordlists collected by Francis de Castelnau and Johann Emmanuel Pohl in the 19th century (Rodrigues 2004; Carvalho and Damulakis 2015: 25) 12, suggesting that the development *k > ? took place in the late 19th or early 20th century.
2.1.3. The depalatalization of PCJ in Xerente
PCJ (allophone of PCJ in nasal contexts) was unconditionally depalatalized to n in Xerente, thus phonemicizing the allophony reconstructible for PCJ. Note that the same process occurred independently in other Macro-Je languages, notably in Southern Je and in Jabuti (Ribeiro & Voort 2010: 565-566).
2.2. Nucleus
The correspondences between Xerente and Xavante vowels are straightforward, as presented in Table 2 below.
12 Note that Rodrigues (2004: 117-118) hypothesizes that the variety attested by Castelnau and Pohl is not the direct ancestor of modern Xavánte. This assumption is needed to account for the loss of k in Karajá male speech (evidenced by data from 1844), which Rodrigues attributes to Xavánte influence. If Rodrigues's hypothesis is correct, absolute dating for the sound change in question is not recoverable from historical sources.
Table 2. Proto-Central Jê vowels and their reflexes in Xavante and Xerénte.
PCJ XAV XER PCJ XAV XER
*a a a
*9 9 9 !(D A 9 9
*i i i
*D D D
*o o o *D D ô
*u u u
*£ £ £ M ■k £ ê
*e e e
*i i i *î i, i ï
Two PCJ vowels, *i and *o, were extremely rare, as are their reflexes in both languages. They are only found in a handful of words, like the postposition *jo / *co 'looking for, pursuing', the TAM marker *to, the verb *krikri 'to cry.PL' or the noun *ji:jiki // 'grasshopper'. In fact, the phonemic status of i and o is not recognized by Pickering (2010: 65, 163-165) for Xavante, whereas some authors do not distinguish between Xerente a and i (Sousa Filho 2007: 74; Frazao 2013: 54).
A non-trivial development occurred in Xavante: PCJ *i was denasalized to i after c (for example, PCJ *piciwi / *ciciwi 'on the surface of' > xav piciwi / ciciwi, xer n§iwi / §§iwi; PCJ *kmd=cici 'to fill' > xav ?mS=cici, xer kma=§i§i). For this reason, the sequence /ci/ does not occur synchronically in Xavante (Pickering 2010: 75, 153).
One very important, but poorly understood process that deeply affected Xerente phonology and phonotactics consists in vowel elision (Mattos 1973; Ribeiro & Voort 2010: 554; Frazao 2013: 85). It massively eliminated PCJ vowels under uncertain conditions (the factors possibly included stress position, consonantal environment and position within the utterance), leading to the emergence of alternating forms like hd-rd / h-ra / hd-r 'to shout.NF', kru-ku / kr-ku / kru-k 'to din.NF'. The distribution of these allomorphs is underdescribed and is of particular importance for the understanding of Xerente historical phonology. Whatever might have been the original conditions for vowel elision in Xerente, the situation observed in this language is clearly innovative when compared to Xavante. For this reason, I assume that Xavante syllabification can be projected onto the PCJ level.
vowel length is contrastive in Xavante; it has been described in autosegmental terms. Long vowels are mostly found under specific prosodic or syntactic conditions (in utterance-final allo-morphs or before the particle ha), except when they arise via compensatory lengthening that accompanies the elision of b before labials and ? (see subsection 2.3). Since nothing suggests that length contrast is innovative in Xavante, I assume it was already present in PCJ and was lost in Xerente.
2.3. Coda and utterance-final allomorphs
Xavante allows two underlying codas: a labial coda (surface realizations include [b], [m], [p], [0] and lengthening of the previous vowel) and a palatal coda (surface realizations include [y], [0] and lengthening of the following consonant). They are analyzed as /b/ and /z/ by Pickering (2010: 179-220) and treated as suprasegmental phenomena by Burgess (1971).
As a consequence of massive vowel elision, Xerénte phonotactic restrictions are very different from those of Xavante and from those that can be reconstructed for Proto-Central Jê. However, both underlying Xavante codas have their correspondences in Xerénte. These are summarized in Table 3 below.
Table 3. Proto-Central Jê underlying codas, their surface realizations and their reflexes in Xavante and Xerénte
PCJ XAV XER
y (before p, b, m, w, r, h)
*/y/ *[y] y ~ lengthening of the following C (before 7) y (utterance-medially)13
lengthening of the following C (before t, d, n, p, c, 3)
*[0] 0 (utterance-finally) 0 (utterance-finally)
p (before t, c)
*[m] m ~ b (before d, 3, r) m (utterance-medially,
*/m/ m (before n, p, h) except before labials)
b ~ lengthening of the preceding V (before 7)
lengthening of the preceding v (before labials) 0 (before labials and utterance-finally)
*[0] 0 (utterance-finally)
As is evident from Table 3 above, the coda allophony observed in Xavante can be partly projected onto the PCJ level because it has exact correspondences in Xerente. Namely, both in Xavante and Xerente both codas fail to surface in utterance-final position, and the labial coda has no consonantal realization preceding another labial consonant. Although it is possible to consider that PCJ had two underlying codas, they must have had multiple surface realizations already in PCJ. The allophony in question is exemplified in (8) below.
(8) Allophonic realizations of PCJ */-m/ and */-y/
a. PCJ */tem/ 'new':
*te (utterance-finally) > xav te, xer te;
*tem (utterance-medially, not before labials) > xav tem / teb / tep, xer tem *te: (before labials) > xav te:, xer te.
b. PCJ */tSy/ 'rain':
*ta (utterance-finally) > xav ta, xer ta;
*tay (utterance-medially) > xav tay- / tSC-, xer tay.
The existence of utterance-medial and utterance-final allomorphs is not specific to PCJ morphemes that end underlyingly in */m/ or */y/ or their modern Xavante and Xerente reflexes. This type of allomorphy is attested both in Xavante, where it is well described (Hall et al.
13 Note that in Xerente ay, Sy are in free variation with e, e (Krieger & Krieger 1994: X). Sometimes a nasalized coda (y) is found in Xerente; although the issue needs further investigation, I hypothesize that this happens before reflexes of PCerr prenasalized codas: cf. hay-§u ~ he-§u 'straw' < PCerr *foy 'bark, skin' + *coy 'leaf' and hay-kre 'to dry leather' (cf. PCerr *ngr3 'dry'), hay-pe 'with a beautiful skin' (cf. PCerr *"be^,e 'good'), hay-kro 'to warm up by the fire' (cf. PCerr *ngro 'to heat'). Given the cursory nature of this hypothesis, I do not reconstruct this apparent contrast to Proto-Central Je.
1987: 274-285; McLeod & Mitchell 2003: 54; Estevam 2011: 139-143), and in Xerente, where it is mentioned in passim (Siqueira 2003: 42). This allomorphy is systematized in Table 5 below; for each allomorphy class, a PCJ reconstruction is proposed.
Table 4. Allomorphy conditioned by the position within the utterance in PCJ, Xavante and Xerente
PCJ notes XAV XER
medial final medial final medial final
Class A. The last syllable is elided in the utterance-final allomorph
*-VkV *-V -V?V -V -(V)k(V) -V
*-VbV *-V only PCJ *wakdbd // *wakd 'price' -VbV -V -bV -V
*-VcV *-V only PCJ *pece // *pe 'good' -VcV -V -(V)gV -V
Class B. The last syllable is elided in the utterance-final allomorph; the utterance-medial allomorph is affected by nasalization14
*-VmV *-V -VmV -V -(V)mV -V
*-VnV *-V -VnV -V -(V)nV -V
Class C. PCJ *-aba // *-wa15
*-aba *-wa -aba -wa -ba -wa
Class D. The vowel of the penultimate syllable is lengthened in the utterance-final allomorph, and the consonant that immediately follows it becomes voiced (unless it is already voiced)
*-VpV *-V:bV oral nucleus -vpv -V:bV -(V)p(V) -bv
*-VtV / *-VtV *-V:dV/ *-V:nV oral / nasal nucleus -vtv / -VtV -V:dV/ -V:nV -(V)t(V) / -(V)t(V) -dV/-nV
*-vdv / *-VnV *-V:dV/ *-V:nV oral / nasal nucleus -VdV/-VnV -V:dV/ -V: nV no alternations
*-VfV *-V:rV -vrv -V: fv
*-Vb(f)V / *-Vm(r)V *-V: b(r)v/ *-V:m(r)V oral / nasal nucleus -vb(r)v / -Vm(r)V -V: b(f)V / -V: m(f)V
Class E. The final vowel is long in the utterance-medial allomorph but short in the utterance-final allomorph
*-V: *-V oral nucleus -V: -V no alternations
Notes. (1) The notation -VCV(-VpV, -V?V, ...) presupposes that the vowels that flank the consonant are identical. (2) V stands for nasal vowels. V stands for any vowel (oral or nasal), unless it is found to the left of a slash (like in *-VtV/*-VtV), in which case it stands for an oral vowel.
14 Note that the vowel in the utterance-final allomorphs of class B stems may be either oral (like in XAV tomo // to 'eye') or nasal (like in XAV mimi// mi 'firewood'). In the former case the nasality of the vowel predictably affects the realization of a preceding voiced consonant in the utterance-medial allomorphs (like in XAV nomo // du 'belly'). Moreover, in all Central Je languages the inventory of nasal vowels is smaller than that of oral vowels, and for this reason multiple vowels found in utterance-final allomorphs may correspond to one and the same vowel in utterance-medial allomorphs of class B stems. For example, Xavante stems whose word-medial allomorphs contain -o- may end in -o, -o or -u in the utterance-final position, e.g. mono// mo (habitual aspect particle), tomo // to 'eye', nomo // du 'belly'.
15 The alternation -aba // -wa may affect the realization of the preceding segment in Xavante (and presumably also in PCJ). For example, in XAV mSprebaba // mSpre:wa 'father-in-law, mother-in-law' underlying /m/ is predictably realized as [b] before -aba, but as vowel length before the labial consonant -w (this is described in the subsection above).
A minimal number of exceptions to these generalizations is found. In the Xavante reflex of PCJ *pip=tete // *pip=te:de 'to be strong' (xav pip=tete/ap=tete //pip=te:te/ap=te:te 'to get strong, to make an effort, to recover'), the medial consonant does not undergo voicing in the utterance-final allomorph, as expected; however, the Xerente reflex is regular (cf. xer nip=tete ~ nip=tte ~ nip=tet // nip=tde). In PCJ *pati // *pa:di 'southern tamandua' (xav pati // pa:di, xer padi, pat-re), unlike in other class D stems, the vowels flanking the alternating consonant are not identical. The stem PCJ *kutd // *kuhd:dd 'tapir' (xav 7utd // 7uhd:dd, xer ktd // kda) exhibits loss of a medial syllable in the utterance-medial allomorph (PCJ **kuhdtd would be expected).
Although I do not make an attempt at postulating a morphophonemic analysis for each class of the allomorphy, it is tempting to consider that the final vowels of the stems of classes A, B and D are absent from their underlying representation, since their quality is entirely predictable: they are always a copy of the vowel of the penultimate syllable16. This fact allows for a rhyme treatment of the bisyllabic sequences participating in the alternations; for example, the underlying representation of PCJ *bakoto // *bako:no 'girl' (xav ba7oto // ba7o:no, xer bakrto17 // bakno) would include the rhyme */-ot/ and the coda*/-t/.
In the remainder of this paper I will use segmental reconstruction for the alternating stems, despite being aware that it is possible to posit morphophonemic rules accounting for the allomorphy.
3. Proto-Cerrado
Now that phonological reconstructions are available for Proto-Northern Je and Proto-Central Je, it is possible to assess the reconstruction of Proto-Cerrado. Onsets, syllable nuclei and codas will be tackled separately.
3.1. Onset
An allophonic process targeting underlying nasal consonants is reconstructed by me for ProtoNorthern Je. It triggers the surface realization of underlying nasal consonants */m/, */n/, */p/ and */j/ as prenasalized voiced obstruents (wb, "d, *ng) in oral contexts (as defined by the nasality of the nucleus). It is important to note that an essentially identical process is attested in Macro-Je languages beyond the Cerrado branch (see D'Angelis 1998 for Kaingang, Pessoa 2012 for Krenak and Voort 2007: 138 for Arikapu) 18 and thus must go back to Proto-Macro-Je.
The correspondences between Proto-Northern Je and Proto-Central Je onsets, as well as the proposed Proto-Cerrado reconstructions for these correspondences, are exposed below in Table 5.
As is evident from the table, Proto-Northern Je faithfully retains the contrasts that existed in Proto-Cerrado. The only innovation that I posit for the history of Proto-Northern Je onsets is the fortition of PCerr *w (in all environments) and *y (in stressed syllables), as exemplified in (8) below.
16 Note that this phenomenon is highly reminiscent of Proto-Northern Je echo vowels (Nikulin 2016b: 168168, 182-183).
17 The variation kt ~ krt, kd ~ krd is widely attested in Xerente (see Frazao 2013: 46).
18 An Optimality Theory account of the prenasalized consonants in Macro-Je languages is offered by Damu-lakis (2010). Voort's analysis of Arikapu phonology is challenged by Ribeiro (2008).
Table 5. Proto-Cerrado onsets and their reflexes in Proto-Northern Jê and Proto-Central Jê.
PCerr PNJ PCJ
before oral vowels before nasal vowels before oral vowels before nasal vowels before PCerr oral non-high vowels before PCerr oral high vowels before PCerr nasal vowels
w W w *b(r) *m(r)
*nb(r) *m(r) *nb(r) *m(r)
*t *t *t *d *n
*nd *n *nd *n
*c
*nd, N/A
*y ■k *3 *3, *c§
*m *k(r) *k(r), *h, *k(r)
*"g(r) *V(r) *ng(r) *V(r) N/A
*g *g
*w *b *w
*r *r *r
*0 *0 *0
t *h before *i, *j elsewhere. ^ *4 in stressed syllables, *y in unstressed syllables. § *c before *i, *j elsewhere. t PCerr *k3, *ng3, *k(r)i, *ng(r)i > PCJ *ha; PCerr *ka- > PCJ *wa:- (in initial unstressed syllables).
(8) Fortition in Proto-Northern Je
a. PCerr *wi / *wt-r 'to kill' > PNJ *bi / *bi-r, PCJ *wi / *wi-n;
b. PCerr *wa 'I' > PNJ *ba, PCJ *wa;
c. PCerr *weke 'partridge' > PNJ *beke, PCJ *wiki;
d. PCerr *kuy9 'bad smell, to stink' > PNJ *ku$>9, PCJ *kuje, etc.
Note that Proto-Northern Je has both *w and *y in stressed syllables, but only in words
that lack an established Cerrado, Je or Macro-Je etymology, like PNJ *y9t§ ~ *y3t3 'sweet potato', *w£w£ 'butterfly'. These may be explained away as borrowings from an unknown source or attributed to a substrate North Amazonian language, whose existence is corroborated by certain lexical similarities between Proto-Northern Je, Nadahup and Yanomami languages (Nikulin & Carvalho 2017: 38).
Proto-Central Je appears to be much less conservative than Proto-Northern Je. The most important change that affected its onsets was the merger of voiceless and voiced occlusive consonants; it is described in more detail in subsection 3.1.1 below. Other conditioned sound changes are tackled in subsections 3.1.2-5.
3.1.1. Occlusive merger
The merger of Proto-Cerrado voiceless and voiced occlusives rearranged the distribution of most of the onsets. PCerr *k and *ng normally yield PCJ *k in all contexts, as shown in (9) below.
(9) Occlusive merger in velars
a. PCerr *kaka 'to cough' > PNJ *kaki, PCJ *kaka;
b. PCerr *kra 'offspring' > PNJ *kra, PCJ *kra: // *kra;
c. PCerr *k£t£ 'stone' > PNJ *kn, PCJ *№£ // *ke:ne;
d. PCerr *kr3y 'head' > PNJ *kr3, PCJ *kr~ay // *kr~3 'head, fruit';
e. PCerr *kr£ 'parakeet' > PNJ *kre, PCJ *kre;
f. PCerr *kre / *kre-r ~ *kre-m 'to eat' > PNJ *kre / *kre-r, PCJ *kre / *kre-ne;
g. PCerr *ku&mi 'fire' > PNJ *ku£i, PCJ *kup5md // *kucd;
h. PCerr *ngoy 'water' > PNJ *>(£-), PCJ *kuy // *ku 'still water';
i. PCerr *ngrs 'dry' > PNJ *ngr3, PCJ *kre;
j. PCerr *ngr£ 'egg' > PNJ wgre, PCJ *kre 'egg', etc.
As for PCerr labial and coronal obstruents, reflexes vary. Before nasal nuclei, both voiceless and voiced occlusives yielded nasal consonants (10).
(10) Occlusive merger in labials and coronals before PCerr nasal nuclei
a. PCerr *pimi 'firewood' > PNJ *pi 'wood', PCJ *mimi // *mi;
b. PCerr *pro 'wife' > PNJ *pro, PCJ *mro;
c. PCerr *pr5m 'hungry' > PNJ *pr3m5 'hungry, to want', PCJ *mr5m // *mr§;
d. PCerr *m3y 'greater rhea' > PNJ *m5, PCJ *m§y // *m§;
e. PCerr *mey 'to throw.SG' > PNJ *me/*me-p, PCJ *mey // *me;
f. PCerr *t£/*fg-m 'to go.SG' > PNJ *te/*t£-m, PCJ *ne-m // *ne 'to go.DU';
g. PCerr *toy 'brother' > PNJ *to, PCJ *noy // *no 'younger sibling of the same sex';
h. PCerr *no/ *n5-Pro 'to lie.SG' > PNJ *no/ *no-r, PCJ *no/ *n5-mr5, etc.
Before Proto-Cerrado high oral nuclei, voiced obstruent reflexes are found (11).
(11) Occlusive merger in labials and coronals before PCerr high nuclei
a. PCerr *puru 'field' > PNJ *puru, PCJ *buru // *bu:ru;
b. PCerr *piy 'achiote' > PNJ *pi, PCJ *bay // *ba;
c. PCerr *nbutu 'neck' > PNJ *nbutu, PCJ *butu // *bu:du;
d. PCerr *nbt 'sun' > PNJ *nb№, PCJ *bdtd // *ba:dd;
e. PCerr *nbint 'tail, penis' > PNJ "bi, PCJ *mm~a // *ba;
f. PCerr *tiki 'belly' > PNJ *tiki, PCJ *diki // *di;
g. PCerr *tu/*tu-r 'to carry' > PNJ *tu/*tu-r, PCJ *du // *du-ri;
h. PCerr *tumu 'belly' > PNJ *tu, PCJ *n5m5 // *du, etc.
Before other Proto-Cerrado oral nuclei, voiceless obstruents are the regular outcome (12).
(12) Occlusive merger in labials and coronals before PCerr non-high nuclei
a. PCerr *kupe 'to touch, to move' > PNJ *kupe, PCJ *kupi;
b. PCerr *para 'foot' > PNJ *pari, PCJ *para // *pa:ra;
c. PCerr *nba 'liver' > PNJ *nba, PCJ *pa;
d. PCerr *nbro 'ashes' > PNJ "bro, PCJ *pro 'powder, foam';
e. PCerr *kanbroy 'blood' > PNJ *kanbro, PCJ *wa:pruy // *wa:pru;
f. PCerr *tere 'fruit of a k. of palm' > PNJ *tere, PCJ *tiri // *ti:ri 'bacuri coconut';
g. PCerr *te 'tick' > PNJ *te, PCJ *ti;
h. PCerr *t3bn 'new' > PNJ *t3bni 'raw', PCJ *tem // *te;
i. PCerr *tepe 'fish' > PNJ *tepe, PCJ *tepe // *te:be; j. PCerr *p5yto 'tongue' > PNJ *p5to, PCJ *p5yto;
k. PCerr *ndomo 'eye' > PNJ *ndo(p-), PCJ *t5m5/*to, etc.
Note that the process described in this subsection occurred prior to the raising of Proto-Cerrado mid-high vowels in Central Je (PCerr *e > PCJ *i, PCerr *o > PCJ *u) and prior to the lowering of PCerr *i > PCJ *a. That way, the contrast between Proto-Cerrado syllables like *te and *ti survives in Proto-Central Je as *ti and *di (cf. 11f and 12f-g).
The situation with Proto-Cerrado palatal occlusives is slightly more complicated. Many Proto-Cerrado inflectable stems are subject to the alternation of their initial consonant: their uninflected form features a stem-initial *y- (*p- before nasal vowels) and their third person form starts with a *V, cf. PCerr *ywa 'tooth' (third person *£wa), PCerr *p5yto 'tongue' (third person %5ytj). It is likely that alternating PCerr %- and *y- / *p- remain distinct in Central Je languages. For example, PCerr *pTkopo (third person *^Tkopo) 'nail, claw' yielded PCJ *pTkopo // *pTko:bo, which appears as *-cTkopo // *-cTko:bo after the prefixes *ay- (second person) and *T-(third person). If PCJ allomorphs do reflect PCerr uninflected form and third person form respectively, the merger of palatal occlusives in PCJ has to be considered incomplete.
However, the Central Je reflexes of non-alternating PCerr %, as well as of its voiced counterpart *n^, display a behavior similar to *p, *t, wb and wd. The relevant cognate sets are provided in (13) below.
(13) Merger of PCerr palatal occlusives
a. PCerr *£oy 'leaf, a hair' > PNJ *£o, PCJ *cuy // *cu;
b. PCerr *ya%e 'nest' > PNJ *ya%e, PCJ *3aci;
c. PCerr *kan^e 'star' > PNJ *kan^e, PCJ *wa:ci;
d. PCerr *n^a / 'to bite' > PNJ ^a / "fa-r, PCJ *ca / *ca-ri;
e. PCerr mmi 'seed' > PNJ PCJ *pdmd //
f. PCerr *ku&mt 'fire' > PNJ *kuU, PCJ *kupamd // *kujd;
g. PCerr *%T 'meat' > PNJ PCJ
The following distribution is found in Proto-Central Je: a voiceless reflex (PCJ *c) is found before PCerr non-high oral vowels (13a-d), a voiced reflex (PCJ *j) is found before PCerr high oral vowels (13e-f), and a nasal reflex (PCJ *p) occurs preceding a nasal vowel (13g). A more profound investigation of Central Je historical morphology should shed light on the details of the operation of the occlusive merger at the palatal point of articulation.
3.1.2. PCerr *ka- > PCJ *wa:-
As has already been noted by Davis (1966: 14), PCerr *ka- yielded PCJ *wa:- in unstressed (nonfinal) syllables (Davis erroneously states that the condition for this sound change is the wordinitial position, but in reality it fails to occur in monosyllabic words, like PCerr *kaka 'to cough' > PCJ *kaka). Some examples are provided in (14).
(14) PCerr *ka- > PCJ *wa:-
a. PCerr *kanbroy 'blood' > PNJ *kanbro, PCJ *wa:pruy // *wa:pru;
b. PCerr *kan^e 'star' > PNJ *kan^e, PCJ *wa:ci;
c. PCerr *ka=ngro 'hot' > PNJ *ka=ngro, PCJ *wa:=kro;
d. PCerr *kat5k5 'to explode, to pop, to shoot' > PNJ *katoko 'firearm, to explode, to pop', PCJ *wan5k5 'to explode, to attack';
e. PCerr *kanga 'lazy' > PNJ *kanga, PCJ *wa:ka;
f. PCerr *kar5 'shadow, spirit' > PNJ *kar5, PCJ *wa:r5, etc.
It is probable that PCerr *k was articulated as *[g] in unstressed syllables, at least before *a; this allophonic realization is attested synchronically in Apinaye and Tapayuna. Panara also seems to retain its traces (Nikulin 2016b: 171, 176). Carvalho (2016: 70) observed that "[t]hough one might claim that the outcomes of the Panara and Central Je developments are only tenuously similar, the conditioning environment is identical for both and seems interestingly unusual", suggesting tentatively that this might be an innovation shared by Panara and Central Je. Given that other data support the inclusion of Panara into Northern Je, ruling out the possibility of a node consisting of Central Je and Panara, I am inclined to envision these facts as direct traces of a Proto-Cerrado allophony.
3.1.3. PCerr i > PCJ * hi, PCerr *yi > PCJ * ci
Certain developments of PCerr palatals before *i are evident in the cognate sets listed in (15) below.
(15) PCerr *U > PCJ *hi, PCerr *yi > PCJ *ci
a. PCerr m 'bone' > PNJ *U, PCJ *hi;
b. PCerr *yi/ *yi-r (3 person / %i-r) 'to put, to lay' > PNJ *4>i / *^i-ri (3 person / %i-ri), PCJ *hi / *hi-ri;
c. PCerr *piyi 'name' > PNJ *pi$>i, PCJ *pici // *pi:ci;
d. PCerr *piyi ~ *piyi 'one' > PNJ *pi$>i, PCJ *mici.
However, details of this development are still poorly understood. For instance, it is unclear why the third person form of the verb 'to put, to lay' (15b) was generalized in Central Je.
3.1.4. PCerr *k(r)i, *ng(r)i, *k3, *ng3 > PCJ *ha
Davis (1966: 14) already noted that Proto-Je *k and *kr are reflected as Xavante h before a. However, it is now evident that the change *k(r) > *h was fed by the occlusive merger: PCerr *k and *ng behave identically in this respect. The data presented in (16) below exemplify the merger of *kri and *ngri, *k3 and *ng3.
(16) PCerr *kri, *ngri, *k3, *ng3 > PCJ *ha
a. PCerr *kri 'cold' > PNJ *kri, PCJ *ha;
b. PCerr *kukriti 'tapir' > PNJ *kukriti, PCJ *kuta // *kuha:da;
c. PCerr *kriti 'flint, firestone; metal' > PNJ *kriti, PCJ *hata // *ha:da;
d. PCerr *ngriki 'angry, upset' > PNJ *ngriki, PCJ *haka;
e. PCerr *k3y 'skin, bark, breast' > PNJ *k3, PCJ *hay;
f. PCerr *k3 / *k3-r 'to shout' > PNJ *k3 / *ks-r, PCJ *ha / *ha-ra;
g. PCerr *ng3 'men's house' > PNJ *ng3 'men's house, courtyard', PCJ *ha 'young men's house', etc.
The development of PCerr *ki and *ngi in Central Je is less clear due to the scarcity of relevant data. One possible cognate set that suggests that PCerr *ki > PCJ *ha involves the correspondence of Proto-Southern Je *yad=ki 'mouth' to Xavante jay=ha 'lip, mouth' and probably Xerente 2fly=ha 'buttocks' (< PCJ *jay=ha); no Northern Je cognate is available. If these words are indeed cognate, they must go back to a Proto-Je stem ending in *=ki.
Note that PCerr *kr3 and *ngr3 must have developed normally, cf. PCerr *ngr3 'dry' > PNJ *ngr3, PCJ *kre.
3.1.5. Interaction with medials
It appears that Proto-Cerrado had a system of rising diphthongs, just like Proto-Northern Je. In some cases Proto-Cerrado medials triggered special reflexes of Proto-Cerrado onsets in Proto-Central Je, including PCerr *kw > PCJ *w, PCerr *yw > PCJ *kw, PCerr *Cye > PCJ *ja ~ *ca. However, the examples are very limited. They are listed in (17) below.
(17) Interaction of PCerr onsets and medials in PCJ
a. PCerr *^wak5 'coati' > PNJ *^wako, PCJ *wak5;
b. PCerr *ywa 'tooth' > PNJ %wa, PCJ *kwa;
c. PCerr *kVywa 'salt' > PNJ *ka$>wa, PCJ *kdkwa-rS;
d. PCerr *k3ykwa 'sky' > PNJ *kVykwa, PCJ *haywa 19;
e. PCerr *yada=kwa 'mouth' > PNJ *yar=kwa, PCJ *3ada=wa;
f. PCerr *kyey 'thigh' > PNJ *kye, PCJ *jay // *ja (non-alternating 3);
g. PCerr *ngye/ *ngye-^n 'to enter.PL' > PNJ *ngiya/ *ngye-dfn, PCJ *3a/ *3a-ci 'to enter.DU' (non-alternating 3);
h. PCerr *tyete 'to burn' > PNJ *tyete, PCJ *3ata // *3a:da;
i. PCerr *kye/ *kyed 'to drag, to pull' > PNJ *kye/ *kyed, PCJ *ca/ *ca-ri.
Davis (1966: 23) also identifies Xavante wam // wa 'fat' (PCJ *wam // wa) as a cognate of PNJ *tw9bni id. Since the correspondences of PNJ *w9 in Central Je are otherwise unknown, this comparison is plausible, but still uncertain.
3.2. Nucleus
The correspondences between Proto-Northern Je and Proto-Central Je nuclei, as well as the proposed Proto-Cerrado reconstructions for these correspondences, are exposed below in Table 620.
Table 6. Proto-Cerrado vowels and their reflexes in Proto-Northern Je and Proto-Central Je.
PCerr PNJ PCJ PCerr PNJ PCJ
*a *a, *a+ *a
*3 *3 *3 *3 *§
*9 *9
*i *i, *i+ *9
*D *D *D
*o *o *u *5 *5 *5
*u *u, *u+
*£ *£ *e *?
*e *e *i
*i *i *i *i *i
+ *a, *u, *? occur before *m in coda. £ PCerr *k3, *ng3 > PCJ *ha.
19 Possibly something other than PCJ *w has to be reconstructed for the reflex of PCerr *kw: in the word for
'sky' a cluster in attested in Xakriaba <akoa> (Eschwege's notes) and Salinas Xavante haijua; however, this is not the case with the word for 'mouth': Xakriaba <d'atoha> (Eschwege), <daidaua> (S. Hilaire), Salinas Xavante =sedaua (Martius 1867: 142, Ehrenreich 1895: 151, 153).
20 Oliveira (2005: 61) cites a talk entitled 'Vowel shift in Central Je' (Oliveira & Ribeiro 2005); unfortunately, I have had no access to its abstract.
As is evident from the table above, Proto-Northern Je is very conservative in its vocalism. The nasalization of *a, *u and *i before syllable-final *m in Proto-Northern Je is described in (Nikulin 2016b: 176-177). This is clearly a Proto-Northern Je innovation: in Central Je, oral vowels are found, as shown in (18) below.
(18) Proto-Northern Je innovative nasalization
a. PCerr *tubn 'old, tall (?)' > PNJ *tumu, PCJ *dum // *du 'tall', *dum-krata // *dum-kra:da 'elder sibling of the opposite sex';
b. PCerr *kubntibn 'capybara' > PNJ *kumtimi, PCJ *kumda.
Unlike Proto-Northern Je, Proto-Central Je innovated considerably. As a result of a vowel shift, discussed already by Ribeiro and Voort (2010: 554), PCerr *9 and *3 yielded PCJ *e (19a-i)21, PCerr *e was raised to PCJ *e (19j-n), and PCJ *e and *o merged with *i and *u, respectively (19o-y) (see subsection 3.1.1 for the effect of this merger on the preceding consonants).
(19) Vowel shift in Proto-Central Je
a. PCerr *nb9/*nb9-^n > PNJ ^9/^9-^ 'to carry', PCJ *kwa=pe 'to carry.DU';
b. PCerr *kur9 'smooth' > PNJ *kur9, PCJ *kure;
c. PCerr *kuy9 'bad smell, to stink' > PNJ *ku^9, PCJ *kuje;
d. PCerr *y9 'bitter' > PNJ %9, PCJ *3e;
e. PCerr *t3bn 'new' > PNJ ^bl 'raw', PCJ *tem // *te;
f. PCerr *ngr3 'dry' > PNJ *ngr3, PCJ *kre;
g. PCerr *=y3 (instrumental suffix) > PNJ *=4<3, PCJ *=je;
h. PCerr *ang3 'sedge seed' > PNJ *ang3, PCJ *ake;
i. PCerr *pr3 'ember' > PNJ *pr3, PCJ *pre;
j. PCerr *tepe 'fish' > PNJ *tepe, PCJ *tepe // *te:be; k. PCerr *ngre 'egg' > PNJ *"gre, PCJ *kre; l. PCerr *tey 'leg' > PNJ *te, PCJ *tey // *te; m. PCerr *kre 'to plant' > PNJ *kre, PCJ *kre; n. PCerr *kre 'orifice, hole' > PNJ *kre, PCJ *=kre;
o. PCerr *tere 'fruit of a k. of palm' > PNJ *tere, PCJ *tiri // *ti:ri 'bacuri coconut';
p. PCerr *te 'tick' > PNJ *te, PCJ *ti;
q. PCerr *ya%e 'nest' > PNJ *ya%e, PCJ *3aci;
r. PCerr *kan^e 'star' > PNJ *kan^e, PCJ *wa:ci;
s. PCerr *weke 'partridge' > PNJ *beke, PCJ *wiki;
t. PCerr *kupe 'to touch, to move' > PNJ *kupe, PCJ *kupi;
u. PCerr *komo 'cudgel; horn; patch of trees' > PNJ *ko, PCJ *komo // *ku 'cudgel; horn'; v. PCerr *ngoy 'water' > PNJ *ngo(£-), PCJ *kuy // *ku 'still water'; w. PCerr *£oy 'leaf, a hair' > PNJ *£o, PCJ *cuy // *cu; x. PCerr *kanbroy 'blood' > PNJ *kanbro, PCJ *wa:pruy // *wa:pru; y. PCerr *kopo 'fly' > PNJ *kopo, PCJ *kupu // *ku:bu, etc.
In addition to the shift, two other developments affected the vocalism of Proto-Central Je: PCerr *i was lowered to PCJ *a (PCJ *i occurs in words without any external etymology), and PCerr *3 and *a merged in PCJ *i. These changes are exemplified in (20) and (21), respectively.
21 Oliveira and Ribeiro (2005 apud Oliveira 2005: 61) state that *3 and *9 remained distinct in PCJ, yielding *e and *e, respectively. I was able to identify only one word with PNJ *9 corresponding to PCJ *e: PNJ *t9^i 'hard, strong' ~ PCJ *tete // te:de 'hard, strong'. This should be attributed to an additional sound law: before a coronal stop in coda, PCJ *e is found instead of *e, also as a reflex of PCerr *3 (PCerr *w3d3 'tree' > PNJ *w3r3, PCJ *wede // *we:de).
(20) Lowering of PCerr *i in Proto-Central Jê
a. PCerr *kubntibn 'capybara' > PNJ *kumtimï, PCJ *kumdd.
b. PCerr *piy 'achiote' > PNJ *pi, PCJ *bay // *ba;
c. PCerr *ku^tmt 'fire' > PNJ *ku^i, PCJ *kupàmd // *kujd;
d. PCerr *am=n& 'bee sp.' > PNJ *am=n^i, PCJ *am=ja;
e. PCerr *nbitt 'sun' > PNJ *nbitt, PCJ *bdtd // *bd:dd;
f. PCerr *nbim 'tail, penis' > PNJ *"b*, PCJ *mm~a // *ba, etc.
(21) Merger of PCerr *3 and *d in Proto-Central Jê
a. PCerr *krsy 'head' > PNJ *krs, PCJ *kr~dy // *krë 'head, fruit';
b. PCerr *pr5m 'hungry' > PNJ *pr3mS 'hungry, to want', PCJ *mrSm // *mr§;
c. PCerr *m5y 'greater rhea' > PNJ *m5, PCJ *m§y // *m§;
d. PCerr *nS 'mother' > PNJ *nS, PCJ *nà;
e. PCerr *pt / *pt-Prt 'to sit.SG' > PNJ *pi / *pi-r, PCJ *pd-mr§ (finite: 2SG *ay=cd);
f. PCerr *kwï / *kwt-r 'to chop' > PNJ *kwï / *kwt-r, PCJ *kwd / *kwd-rï, etc.
A context-specific development of PCerr *3 after velars has already been treated in subsection 3.1.4.
3.3. Coda
Proto-Central Jê codas, as well as Proto-Central Jê alternating rhymes of the classes A, B and D (see 2.3 above), have regular correspondences in Proto-Northern Jê. Proto-Central Jê classes C and E are of unknown origin. The most secure correspondences are listed in Table 7 below.
Table 7. Proto-Cerrado codas and their reflexes in Proto-Northern Jê and Proto-Central Jê.
PCerr PNJ PCJ PCerr PNJ PCJ
*-py *-pV *-pV (D) *-mV *-0 *-mV (B)
*-tV *-tV *-tV (D) *-nV *-0 *-nV (B)
*-kV *-kV *-kV (A) *-0 *-kV (A)
W (D) *-y *-0 *-y
V V *-rï *-bn/*-m *-b"ïr-mV *-m
Note that some codas that are reconstructible to Proto-Northern Je do not occur in words with secure Cerrado, Je or Macro-Je etymologies. These include PNJ *-wV and *-yi22. I suggest that such words entered Proto-Cerrado as loans from an unknown North Amazonian language: words like PNJ *kukoyi 'monkey' and PNJ *ndiwi 'new' 23 are suspiciously similar to Hup kukuy, Yanomama ku(u)ku-moxi 'night monkey' and Hup nd'dw-ay 'new', respectively (Ni-kulin & Carvalho 2017: 38).
3.3.1. Voiceless codas in Proto-Cerrado
As is evident from the table above, Proto-Northern Je voiceless stops in coda position correspond to PCJ voiceless stops in coda utterance-medially. PCJ *p and *t, on one side, and PCJ *k,
22 Possible exceptions include PNJ *bayi 'snake sp.' ~ PCJ *wa:hi 'venomous snake', PNJ *n^3yï ~ PCJ *cepc£t£ // cepc£:de 'woodpecker'. Both comparisons are dubious.
23 Erroneously glossed as 'field' in (Nikulin 2016b: 180).
on the other, behave differently. PCJ stems in *-pV and *-tV belong to class D, as shown in (22) below, whereas PCJ stems in *-kV belong to class A, as exemplified in (23).
(22) PCerr stems in *-pV, *-tV
a. PCerr *kopo 'fly' > PNJ *kopo, PCJ *kupu // *ku:bu;
b. PCerr *tepe 'fish' > PNJ *tepe, PCJ *tepe // *te:be;
c. PCerr *pip=kop5 'nail, claw' > PNJ *pip=kop5, PCJ *pi=kopo // *pi=ko:bo;
d. PCerr *nbiti 'sun' > PNJ *nbiti, PCJ *bata // *ba:da;
e. PCerr *nbutu 'neck' > PNJ *nbutu, PCJ *butu // *bu:du;
f. PCerr *p5=koto 'chest' > PNJ *p5=koto, PCJ *p5=kutu // *p5=ku:du;
g. PCerr *tyete 'to burn' > PNJ *tyete, PCJ *^ata // *3a:da;
h. PCerr *kot5 'cicada' > PNJ *(ko)kot5, PCJ *koto // *ko:do;
i. PCerr *kukriti 'tapir' > PNJ *kukriti, PCJ *kuta // *kuha:da; j. PCerr *kriti 'grasshopper' > PNJ *kriti, PCJ *kriti // *kri:di;
k. PCerr *krata 'base' > PNJ *krati 'base, stem, lower part of the body', PCJ *krata //
*kra:da 'base, beginning; near; grandparent';
l. PCerr *ritl 'to look' > PNJ *dti, PCJ *riti // *ri:ni, etc.
In one instance, the final consonant has apparently been irregularly nasalized in ProtoNorthern Je: PCerr *kete 'stone' > PNJ *kene, PCJ *kete // *ke:ne.
(23) PCerr stems in *-kV
a. PCerr *koko 'wind' > PNJ *koko, PCJ *ro:=wa=kuku // *ro:=wa=ku;
b. PCerr *tiki 'belly' > PNJ *tiki, PCJ *diki // *di;
c. PCerr *totok5 'to throb' > PNJ *totok5, PCJ *totokj // *toto;
d. PCerr *kaka 'to cough' > PNJ *kaki, PCJ *kaka;
e. PCerr *ngriki 'angry, upset' > PNJ *ngriki, PCJ *haka;
f. PCerr *katoko 'to explode, to pop, to shoot' > PNJ *katok5 'firearm, to explode, to pop', PCJ *wan5k5 'to explode, to attack', etc.
Utterance-final allomorphs of Central Je counterparts of the cognate sets (23d-f) have not been attested, but these stems are expected to belong to class A.
3.3.2. Rhotic codas in Proto-Cerrado
I reconstruct two types of Proto-Cerrado stems with a rhotic coda. Just like in Proto-Northern Je (Nikulin 2016b: 182), rhotic coda followed by an echo vowel would have been characteristic of nouns. In Central Je, it patterns with PCerr *-pF and *-tV (class D stems). On the other hand, non-finite forms of verbs would have featured a rhotic with a suppressed echo vowel. In this case, a final *-i is inserted in Central Je (*-i if the preceding syllable carries nasality). This is shown in (24) below.
(24) Distinction between PCerr *-rVV and *-r
a. PCerr *para 'foot' > PNJ *pari, PCJ *para // *pa:ra;
b. PCerr *puru 'field' > PNJ *puru, PCJ *buru // *bu:ru;
c. PCerr *tere 'fruit of a k. of palm' > PNJ *tere, PCJ *tiri // *ti:ri 'bacuri coconut';
d. PCerr *p5ro 'cord, vine' > PNJ *p5ro, PCJ *wede=p5r5 // *wede=p5:r5;
e. PCerr *wi/*wi-r 'to kill' > PNJ *bi/*bi-r, PCJ *wi/*wi-ri;
f. PCerr *pa/*pa-r 'to finish' > PNJ *pa/ *pa-r, PCJ *pa/ *pa-ri 'to finish, to erase';
g. PCerr *tu / *tu-r 'to carry' > PNJ *tu / *tu-r, PCJ *du / *du-ri;
h. PCerr *n^a/ *n^a-r 'to bite' > PNJ *n^a/ ^a-r, PCJ *ca/ *ca-ri, etc.
However, the allomorph *-ri cannot be attached to Central Je reflexes of PCerr roots ending in certain vowels, like *3 and *e. Allomorphs with echo vowels may be found in such cases, cf. PCJ *ha/ *ha-ra 'to shout' (< PCerr *k3/ *fo-r).
3.3.3. Nasal codas in Proto-Cerrado
Central Je stems that belong to class B correspond to vowel-final stems in Proto-Northern Je. I reconstruct nasal codas followed by an echo vowel. The correspondence is exemplified in (25) below.
(25) Proto-Cerrado etyma yielding Central Je class B stems
a. PCerr *pimi 'firewood' > PNJ *pi 'wood', PCJ *mimi // *mi;
b. PCerr *komo 'cudgel; horn; patch of trees' > PNJ *ko, PCJ *k5m5 // *ku 'cudgel; horn';
c. PCerr *kuUmi 'fire' > PNJ *ku£i, PCJ *kupama // *kuja;
d. PCerr mmi 'seed' > PNJ PCJ y~dm~a // *3a;
e. PCerr *ndomo 'eye' > PNJ *ndo(p-), PCJ *tomo / *to;
f. PCerr *nbini 'tail, penis' > PNJ *"bi, PCJ *man~a // *ba;
g. PCerr *tumu 'belly' > PNJ *tu, PCJ *n5m5 // *du, etc.
The Proto-Northern Je consonant-final allomorph *ndop- (25e) is preserved only in Api-naye, where it occurs before any elements in postposition (Ham 1961: 27; Oliveira 2005: 375). The final consonant -p must be understood as a relic of the Proto-Cerrado coda.
For Central Je class A stems ending in *-kF that correspond to zero in PNJ, I suggest reconstructing PCerr The denasalization of in PCJ is expected: no such phoneme can be reconstructed for PCJ, even in the onset position. The clearest examples are provided in (26).
(26) Proto-Cerrado
a. PCerr *poq5 'arrow, bamboo' > PNJ *po-^e 'bamboo', PCJ *poko // *po 'big arrow';
b. PCerr *nboVo 'lake' > PNJ *==nbo, PCJ *puku // *pu.
At first sight, the reconstruction of a velar segment for this correspondence conflicts with external data: probable Maxakali cognates of these words are pohox /puC/ 'arrow' and puxhep /piC-heP/ 'lake', with palatal codas. However, PCerr may correspond to Maxakali palatals even in the onset position (cf. PCerr *ka^n 'snake' and Maxakali kaya /kapa/ 'snake'), which renders the reconstructive hypothesis in question plausible.
3.3.4. Palatal coda in Proto-Cerrado
Proto-Central Je stems whose underlying representations end in */-y/ correspond to vowel-final stems in Proto-Northern Je. I reconstruct PCerr *-y for this correspondence. Some examples are provided in (27).
(27) Proto-Cerrado stems in *-y
a. PCerr *t5y 'brother' > PNJ *t5, PCJ *n5y // *n5 'younger sibling of the same sex';
b. PCerr *m3y 'greater rhea' > PNJ *m3, PCJ *may // *ma;
c. PCerr *ngoy 'water' > PNJ *ngo(b-), PCJ *kuy // *ku 'still water';
d. PCerr *kr3y 'head' > PNJ *kr3, PCJ *kray // *kra 'head, fruit';
e. PCerr *piy 'achiote' > PNJ *pi, PCJ *bay // *ba;
f. PCerr *kanbroy 'blood' > PNJ *kanbro, PCJ *wa:pruy // *wa:pru;
g. PCerr *%oy 'leaf, a hair' > PNJ *£o, PCJ *cuy // *cu;
h. PCerr *foy 'skin, bark, breast' > PNJ *fo, PCJ *hay;
i. PCerr *kyey 'thigh' > PNJ *kye, PCJ *jay // *3a; j. PCerr *tey 'leg' > PNJ *t£, PCJ *tey // *te;
k. PCerr *kny 'rotten' > PNJ *kn, PCJ *kny // *kn;
l. PCerr *mey 'to throw.SG' > PNJ *me/ *me-p, PCJ *mey // *me, etc.
In (27c), one Northern Je language (Apinaye) retains two allomorphs: ngo and ngoyc- (before any elements in postposition, see Ham 1961: 27); I reconstruct PNJ *ngo(%>-). The correspondence under examination also seems to be valid word-medially: PCerr *poyto 'tongue' > PNJ *p5to, PCJ *p5yto.
3.3.5. Labial coda in Proto-Cerrado not followed by an echo vowel
Stem-final */-m/ in Proto-Central Je corresponds to -bnl/-mV in Proto-Northern Je (per nasality); I reconstruct PCerr *-bn/*-m for this correspondence. It is instantiated in the stems listed in (28).
(28) Proto-Cerrado stems in *-b7 *-m
a. PCerr *pr3m 'hungry' > PNJ *pr3m3, PCJ *mrSm // *mrS;
b. PCerr *t£/*t£-m 'to go.SG' > PNJ *t£/ *te-m, PCJ *ne-m // *ne 'to go.DU';
c. PCerr *tubn 'old, tall (?)' > PNJ *tumu, PCJ *dub // du 'tall', *dub-krata // *dub-kra:da 'elder sibling';
d. PCerr *yobn 'to grind' > PNJ %obn 'flour, powder', PCJ *jum // *ju;
e. PCerr *tsbn 'new' > PNJ •fob'? 'raw', PCJ *tem // *t£, etc.
The utterance-internal allomorph of PCJ *kumdd (< PCerr *kubntibn 'capybara', cf. PNJ *kumtimV) is unattested, but the predicted form is *kumddm.
3.3.6. Complex codas in Proto-Cerrado
In a number of stems, PNJ *-rV or *-r correspond to PCJ *-brV/ *-mrV. I tentatively reconstruct PCerr *-PrV for these cases. Some examples are provided in (29) below.
(29) PCerr *-PrV and its reflexes
a. PCerr *y3 / *y3-Prs 'to enter.SG' > PNJ / *^3-r, PCJ *j£/ *3£-bre;
b. PCerr / yi-Pri 'to sit.SG' > PNJ / *pi-r, PCJ *pd-mrd (finite: 2SG *ay=cd);
c. PCerr *n5/ *no-Pri 'to lie.SG' > PNJ *no/ *no-r, PCJ *no / *no-mn;
d. PCerr *yuPru 'pus' > PNJ ~ *&uru, PCJ *jubruy;
e. PCerr *ka=to/*ka=to-Pro 'to leave, to go out' > PNJ *ka=to /*ka=to-r 'to leave, to go out, to be born', PCJ *wa=to / *wa=to-bro, etc.
It is possible that PCJ *tobro 'to get down (Xavante), to cross a body of water (Xerente)' and PCJ *cebre 'to roast (Xavante), to become soft, cooked (Xerente)' are related to PNJ *to/*to-r 'to fly, to dance' and *ga/ 'to roast', but this is far from certain. I have no explanation for Central Je non-finite forms in *-mri, like PCJ *c§/ *cS-mri 'to find', *c5 / *c5-mri 'to give', *cac5/ *cac5-mri 'to hang', *cz/ *ci-mri 'to seize, to shoot', *n5/ *n5-mri 'to lay.DU', *p§/ *pS-mri 'to weave'. These verbs must be accounted for by future research.
Yet another possible instance of a complex coda in Proto-Cerrado is suggested by a handful of verbs that have two different forms in Proto-Central Je: a form in *-kF and a form in *-rV (cf. PCJ *to=po-ko/ *to=po-ro 'to open one's eyes, to wake up'; *pu-ku/*pu-ru 'to pierce', *ddkd// *dd/ *ddrd //*dd:rd 'to die', *waptd-kd/ waptd-rS 'to fall, to be born.SG/DU', etc.). The syntactic distri-
bution of these forms is not fully explored. I abstain from reconstructing Proto-Cerrado forms in *-krV for the verbs that feature the phenomenon in question, but such a solution remains a possibility.
3.3.7. Other possible codas in Proto-Cerrado
PCerr *d also seems to have occurred in coda position, but only two examples have been identified so far; these are listed in (30).
(30) Evidence for reconstructing PCerr *-dV
a. PCerr *w3d3 'tree' > PNJ *b3r3, PCJ *wede // *we:de;
b. PCerr *yad-kwa 'mouth' > PNJ *yar-kwa, PCJ *jada-wa.
I was unable to identify regular PCJ correspondences of the PNJ codas *-fci, *-dn(i), *-dn(i), *-nV due to the scarcity of available examples. I provisionally project these PNJ codas to Proto-Cerrado, but a better solution is likely to be possible. Some examples are provided in (31).
(31) Tentatively reconstructed Proto-Cerrado codas
a. PCerr *nb£^g 'good' > PNJ *"b£tf, PCJ *pece // *pe;
b. PCerr *t9%9 'hard, strong' > PNJ *t9%i, PCJ *tete // te:de;
c. PCerr *kye/*kyed > PNJ *kye/*kyed 'to drag, to pull', PCJ *ca/*ca-ri;
d. PCerr *ngye/ ^gye-^ 'to enter.PL' > PNJ *ngiya/ *ngyed PCJ *ja/ *ja-ci 'to enter.DU';
e. PCerr *pmi 'guts, faeces' > PNJ *pmi, PCJ *^ana 'faeces';
f. PCerr *y3d3 'sweet' > PNJ %3d% PCJ *jey // *3e, etc.
Yet another problematic correspondence is found in PNJ *proto 'to run' ~ PCJ *ay=ca=mroy // *ay=ca=mro 'to run.DU', if only this comparison is valid.
3.3.8. Notes on echo vowels
The phenomenon of echo vowels, reconstructed by me for Proto-Northern Je (Nikulin 2016b: 169, 182), undoubtedly existed in Proto-Cerrado. However, certain particularities of its realization reconstructible for Proto-Northern Je might have not existed in Proto-Cerrado. For instance, the height dissimilation after the nucleus *a seems to be a PNJ innovation, since in PCJ a copy of the nucleus appears in the echo vowel position (cf. PCerr *para 'foot' > PNJ *pari, PCJ *para // *pa:ra, PCerr *krata 'base' > PNJ *krati 'base, stem, lower part of the body', PCJ *krata // *kra:da 'base, beginning; near; grandparent'; PCerr *kaka 'to cough' > PNJ *kaki, PCJ *kaka, etc.). The presence of the echo-vowel *-i after palatal and post-nasalized codas in Proto-Northern Je also appears to be innovative. That way, in my reconstruction Proto-Cerrado echo vowels were always identical to the nucleus.
Note that the suppression of the echo vowels in the non-finite forms of the verbs can be securely reconstructed to Proto-Cerrado. In Proto-Central Je, PCerr *-r yielded *-ri/*-ri (according to the nasality of the nucleus). In Proto-Northern Je, the contrast between PCerr *-rF and *-r seems to have been retained, though it has been deteriorated in most modern Northern Je languages (see Nikulin 2016b: 182 for traces of this contrast).
3.3.9. Unexpected nasalization in Proto-Central Je
A number of Central Je stems exhibit an unexpected nasalization of the stem vowel. They are listed in (32) below.
(32) Unexpected nasalization in Proto-Central Je
a. PCerr *re 'to leave' > PNJ *re, PCJ *re/ *re-me 'to leave.SG';
b. PCerr *nday 'rain' > PNJ "da, PCJ *tay // *ta;
c. PCerr *tu / *tu-r 'to urinate' > PNJ *tu / *tu-r, PCJ *to/ *to-rf,
d. PCerr *pa/*pa-r 'to kill' > PNJ *pa/ *pa-r, PCJ *pa/ *p~3-ri 'to kill.DU';
e. PCerr *ngre/ *ngre-r 'to sing, to dance' > PNJ *ngre/ *ngre-r, PCJ *ay=kre / *ay=kre-ne 'to dance.SG'.
Note that this nasalization cannot have preceded the occlusive merger (see subsection 3.1.1); otherwise, nasal occlusives would have been expected in 32b-d. Moreover, it cannot be easily explained away as a sporadic process postdating the occlusive merger: in this case PCJ **no/ **n5-ri 'to urinate' (< **du/ **du-ri) would be the expected form. Presently I have no explanation for this correspondence.
4. Cerrado etymologies
This section contains the cognate sets upon which the observations made in this paper are based.
PCerr *ang3 'sedge seed' > PNJ *ang3 (api ang3, kay a^3, tim ak3, pnr n^ks), PCJ *ake (xav ?a?e, xer ake);
PCerr *nba 'liver' > PNJ wba (api nba, kay ma, SUY, tap nba, tim pa/=mpa, pnr z"pa), PCJ *pa (xav, xer pa);
PCerr / *nba-r 'to hear, to listen' > PNJ *nba / *nba-r 'to hear, to know, to listen' (api nba / nba-r, kay ma / ma-ri, tim pa / pa-r/=mpa/=mpa-r, SUY nba / nba-y, tap nba / ba-y, pnr iya-ri), PCJ *wa=pa / *wa=pa-ri 'to hear, to listen, to obey' (xav wa=pa / wa=pa-ri, xer wa=pa /wa=pa-ri ~ wa=pa-r);
PCerr wba 'afraid' > PNJ *=ba (api nba, kay =ma, SUY =ba, tap nba, tim pa, pnr yu=pay), PCJ *pa-hi (xav pa-hi, xer pa-hi), PCJ *pi=pa (xav pi=pa, xer pi=pa);
PCerr *nb£^£ 'good' > PNJ *nb£tf (api nbec, kay mec, SUY nberi, tap nbey-, tim pey, pnr fpe), PCJ *pece // *pe (xav pece // pe 'well; to recover, to get better', xer pepe // pe 'good, beautiful');
PCerr *nb9/ *nb9d 'to carry' > PNJ *nb9/ *nb9d 'to carry' (api nb9/ nb9-yd>n ~ nb9-r, kay m9/ m9-yd>n 'to grab', tim p9/ p9-dn, suy nb9/ nb9-dl 'to grab', (?) pnr iyi-ri), PCJ *kwa=pe 'to carry.DU' (xav ?wa=pe, xer kwa=pe);
PCerr *nbiti 'sun' > PNJ *nbiti 'sun' (api nbit, kay mit, suy, tap nbiri, tim pit, pnr rpiti), *nbit-rw9 'moon' (api nbit-vr9, kay miti-rw9, SUY nbit-lw9, tim pit-wr9), PCJ *bata // *ba:da 'sun, day' (xav bata // ba:da 'sun, day, hour', xer bda 'sun, day, god');
PCerr *nbini 'tail, penis; man' > PNJ *nbi (api ya=nbi 'tail', nbi 'man', kay ya=mi 'tail', mi 'man', suy nbi 'tail, penis', me=nbi-ye 'man', tap nbi, tim ya=pi 'tail', mpi 'man', pnr y3=npi 'tail', iyi 'man'), PCJ *mma // *ba 'tail, penis' (xav m§n§ // ba, xer mna // ba 'tail'), *ay=ba 'man' (xav ?ay=ba, xer am=ba);
PCerr *nboyo 'lake' > PNJ *%=i=nbo (api ?=i=nbo, kay i=nbo, tim h=i=po, h=i:=po), PCJ *puku // *pu (xav pu?u // pu 'lake with a spring', xer pku);
PCerr *nbn 'ashes' > PNJ *nbrj (api nbro, kay mro, tim pro), PCJ *wede=pro 'coal' (xav wede=pro 'coal; coffee', xer wde=pro 'coal; coffee; sawdust'), *ka=jay=pro 'foam; beer' (xav ?a=jay=pro, xer k3=2fli=pro ~ ka=2flm=pro), *jaday=pro 'saliva' (xav jaday=pro, xer 2Ldai=pro 'saliva'), (?) *pro 'to burn' (xer pro);
PCerr *nbutu 'neck' > PNJ *nbutu (api nbut, kay mut, SUY nburu, tap muru, nbu-ti ~ mu-ti, tim put, pnr inputi 'nape'), PCJ *butu // *bu:du (xav butu // bu:du, xer bdu);
PCerr *=nbuy 'to see' > PNJ *=pu/*=pu-^n (api =bu/=nbu-y^n, kay mu/mu-yp, SUY =bu, tap mu, tim =pu /=pu-dn, pnr puq ~ pu/=yuq ~ =npu), PCJ *ja=buy // *ja=bu (xav ja=buy // *ja=bu, xer 2fl=bu), *pi=buy // *pi=bu (xav pi=buy // pi=bu 'to know how to, to control, to care, to observe, to tame'), xer pi=bu 'to observe, to visit');
PCerr *nday 'rain' > PNJ wda (api nda, kay na, SUY, tap nda, tim ta, pnr ita), PCJ *tSy // *tS (xav, xer tSy // tS);
PCerr *ndom5 'eye' > PNJ wdo(p-) (api ndo(p-), kay no, suy, tap ndo, tim to/=nto, pnr into), PCJ *tomo / *to (xav tomo / to, xer tmo / to);
PCerr *n^a/ *n^a-r 'to bite' > PNJ ^a / *n^a-r (api ja/ja-r, kay pa/pa-yp, SUY, tap ta, tim ca /ca-r/=nca / =nca-r, pnr insa / insa-ri), PCJ *ca / *ca-ri (xav ca / ca-ri, xer pa /pa-ri ~ pa-r);
PCerr 'bee', *am=n^i 'bee sp.' > PNJ (api *ji, tap nbey=ti), *am=n^i (api am=nji, kay am=yi, tim am=ci), PCJ *am=3a (xav ab=ja 'bee sp. (Tetragona clavipesf, xer am=za 'Brazilian wasp (Protonectarina sylveirae)');
PCerr wg3 'men's house' > PNJ *ng3 'men's house, courtyard' (api ng3(p-), kay q3, SUY, tap ng3, tim k3), PCJ *ha 'young men's house' (xav ha);
PCerr *ngo 'louse' > PNJ *ngo (api ngo, tap ngo, tim ko/=qko, pnr ky3=ko), PCJ *ku (xav ?u);
PCerr *ngoy 'water' > PNJ *ngo(%-) (api ngo(yc-), kay qo, suy, tap ngo, tim ko, pnr Fko), PCJ *kuy // *ku 'still water' (xav ?uy // ?u, xer kuy-);
PCerr *ngoy 'wet, to soak' > PNJ *ngo (api ngo, kay qo, SUY ngo, tap ngo ~ ngo, tim ko, pnr kow), PCJ *koy // *ko (xer koy // ko);
PCerr *ngre 'egg, testicle' > PNJ *ngre (api ngre, kay qre, SUY ngj£, tap ngK£, tim kre/=qkre, pnr Pkre), PCJ *kre 'egg' (xav ?re, xer kre);
PCerr *ngre/*ngre-r 'to sing, to dance' > PNJ *ngre/ *ngre-r (api ngre/ngre-r, kay qre/ qre-re 'to sing', suy nge-le, tap ngKe/ nge-re, tim kre), PCJ *ay=kre/ *ay=kre-ne 'to dance.SG' (xav ay=?re / ay=?re-ne);
PCerr *ngr3 'dry' > PNJ *ngr3 (api ngr3, kay qr3, SUY ngx3, tap ngv3, tim kr3/=qkr3), PCJ *kre (xav ?re, xer kre);
PCerr *ngriki 'angry, upset' > PNJ *ngriki 'angry' (api ngriki, kay qrik, SUY ng^ki, tim krik/=qkrik, pnr rkyi), PCJ *haka (xav haka);
PCerr *ngro 'to heat', *ka=ngro 'hot, fever' > PNJ *ngro (api ngro 'to bake', kay am=qro 'heat', SUY ngxo, tim kro), *ka=ngro 'hot, fever' (api ka=ngro, kay ka=qro, SUY k(h)a=ngio 'warm; to burn', tap ka=ngKo, tim ka=kro, pnr n3=kyo), PCJ *kro (xav ro:=?ro 'the end of the dry season', xer hay=kro ~ hey=kro 'to warm up by the fire'), *wa:=kro 'hot, fever' (xav wa:=?ro, xer wa=kro);
PCerr *ngye/ *ngye-^n 'to enter.PL' > PNJ *ngiya/ *ngye-^n 'to enter.PL, to put into a deep con-tainer.PL' (api ngye/ngye-yya=ngye, kay qiya/qye-yW, ya=qye-y, suy qye/ qye-le, tim a=kye-y), PCJ *ja/*ja-ci 'to enter.DU' (non-alternating 3) (xav 3a/ja-ci, xer zp-pi ~ zfl-§);
PCerr *ka 'white' > PNJ *ya=ka (api, kay, suy, tap ya=ka, tim ya=kha), PCJ *ka (xav ?a, xer ka);
PCerr *kanbroy 'blood' > PNJ *kanbro 'blood, menstruation' (api kanbro, kay kamro ~ kamro, SUY kanblo, tap kanro, tim kapro, pnr n3=pyu), PCJ *wa:pruy // *wa:pru (xav wa:pruy // wa:pru, xer wapru);
PCerr *kan^e 'star' > PNJ *ka%e (api ka^je, kay kape-ti-re, SUY ka^e-a, tap ka^e-a ~ kande-ci, tim kace-rei, pnr n3si-ti ~ n3su-ti), PCJ *wa:ci (xav wa:ci, xer wapi);
PCerr *kanga 'lazy' > PNJ *kanga, %u=kanga 'lazy, to give up, to refuse' (api kanga, kay kaqa, ju=kaqa, SUY kanga, tap tu=§nga, tim kaka 'to leave alone', pnr sw=anka), PCJ *wa:ka (xav wa:?a, xer waka);
PCerr *kaka 'to cough' > PNJ *kaki (api kaka, tim khak), PCJ *kaka (xav ?a?a, xer kka);
PCerr *karo 'shadow, spirit' > PNJ *karo 'shadow, spirit, image' (api karo, kay, tap karo 'image', tim karo), PCJ *wa:ro (xav hay=ba=wa:ro 'dead body');
PCerr *ka=to/ *ka=to-Pro 'to leave, to go out' > PNJ *ka=to/ *ka=to-r 'to leave, to go out, to appear, to be born' (api ka=to / ka=to-r, kay ka=to / ka=to-ro, SUY ka=tho/ ka=tho-lo, tap ka=p/ ka=p-ro, tim ka=to/ka=to-r), PCJ *wa=to/*wa=to-bro (xav wa=to/wa=to-bro, xer wa=to-bro ~ wa=to-br ~ wa=t-bro);
PCerr *katoko 'to explode, to pop, to shoot' > PNJ *katoko 'firearm, to explode, to pop' (api katonko, kay katjnk, SUY kato^o, tim katok, pnr ato-si 'bullet'), PCJ *wanoko 'to explode, to attack' (xav wano?o, xer wanoko);
PCerr *kete 'stone' > PNJ *kene (api kene, kay ken, SUY khene, tap kene, tim khen, pnr kiey), PCJ *kete // *ke:ne (xav ?ete // ?e:ne, xer kte // kne);
PCerr *k3/ *k3-r 'to shout' > PNJ *k3/*k3-r 'to sing (of birds)' (api k3/k3-r, kay k3/k3-r3, tim kh3), PCJ *ha / *ha-ra (xav ha/ ha-ra, xer ha/ ha-ra ~ ha-r ~ h-ra);
PCerr *k3y 'skin, bark, breast' > PNJ *k3 (api, kay fo, SUY kh3, tap k3, tim kh3, pnr k3 'skin, bark'), PCJ *hay (xav ha, xer hay ~ he // ha 'body, skin, breast'), *hdy-man§ // *hay-ba 'being' (xav hay-mana // hay-ba 'to be, to behave; being, body', xer hay-mba ~ he-mba 'to exist; soul, body, image');
PCerr *k3ykwa 'sky' > PNJ *kVykwa (api katkwa, kay k3ykwa, tap kaykwa, tim koykhwa), PCJ *haywa (xav haywa, xer haywa ~ hewa);
PCerr *koko 'wind' > PNJ *koko (api koko, kay kok, SUY koko, tap kogo, tim khok), PCJ *ro:=wa=kuku // *ro:=wa=ku (xav ro:=wa=?u?u // ro:=wa=?u, xer ro=wa=kku), *wa=kuku // *wa=ku 'to blow (of wind) (xer wa=kku);
PCerr *komo 'cudgel; horn; patch of trees' > PNJ *ko (api ko 'cudgel; patch of trees', kay ko 'cudgel; patch of trees', SUY ko 'cudgel', tap ko 'cudgel', tim kho), PCJ *k5m5 // *ku 'horn' (xav ?omo // ?u, xer kmo // ku);
PCerr *kopo 'fly' > PNJ *kopo (api kop-ti, kop-re, tap kowo, tim khop), PCJ *kupu // *ku:bu (xav ?upu // ?u:bu, xer kpu // kbu);
PCerr *koto 'cicada' > PNJ *(ko)koto (api ko(?)kot-le, tim ko(?)khot, (?) ^or), PCJ *koto // *ko:do (xav ?oto // ?o:do, xer kdo ~ krdo);
PCerr *kra 'offspring' > PNJ *kra (api, kay kra, SUY k(h)ra, tap kxa, tim khra), PCJ *kra: // *kra (xav ?ra: // ?ra, xer kra);
PCerr *krata 'base' > PNJ *krati 'base, stem, beginning; lower part of the body' (api krata 'waist, leg, beginning, medial part of a long object', krayc 'wall, stem, stalk', kay krayc 'trunk, stump, pelvis', SUY khiari 'beginning', tim khrat, pnr kyati 'waist'), PCJ *krata // *kra:da 'base, beginning; near; grandparent' (xav ?rata // ?ra:da, xer krata // krda 'previous, old; near');
PCerr *kre 'to plant' > PNJ *kre (api, kay kre, SUY khie, tim k^e 'to dig, to plant', pnr kre), PCJ *kre (xav ?re, xer kre);
PCerr *kre 'orifice, hole' > PNJ *kre (api, kay kre, tap kxe, tim kVe, pnr kre), PCJ *am=kre 'hole, pit, tomb' (xav ab=?re, xer ap=kre), *kre 'vagina, anus' (xav ?re, xer kre 'vagina');
PCerr *kre 'parakeet' > PNJ *kre (api kre-ti, kre-re, SUY ke, tap kxe, tim k^e-re), PCJ *kre (xav ?re-re, xer kre);
PCerr *kre/*kre-r ~ *kre-ne 'to eat' > PNJ *kre/*kre-r (api kre/kre-r 'to eat, to swallow', kay kre/ kre-n, SUY khje/khje-ne, tap kxe/ke-re, tim khre/khre-r, pnr kre), PCJ *kre/*kre-ne (xav ?re/?re-ne, xer kre/kre-ne);
PCerr *kr3 'dark, cloud' > PNJ *kr3 (pnr r3=ky3/ya=ky3), *ka=kr3 'dark cloud' (api ka?=kr3, kay ka=kr3, tim ka=khr3), PCJ *kra (xav ja=?ra 'darkness', ta=ja=?ra 'rain cloud', ?ra=da?a/?ra=da 'black, dark', xer kra 'black (in compounds)', ayna=ka=kra 'rain cloud');
PCerr *kr3y 'head' > PNJ *kr3 (api, kay, suy kr3, tap kxv, tim k^3, pnr iy3), PCJ *kray // *kra 'head, fruit' (xav ?ray // ?ra, xer kray // kra);
PCerr *kriti 'grasshopper' > PNJ *kriti (api kriti, tap k(x)it-ci, ki-re), PCJ *kriti // *kri:di (xav ?riti // ?ri:di, xer krti // krdi 'grasshopper sp. (likely Tropidacris cristata));
PCerr *kri 'cold' > PNJ *kri (api, kay kri, suy khii, tap kxi, tim khri, pnr kyi), PCJ *ha (xav, xer ha);
PCerr *kriti 'flint, firestone; metal' > PNJ *kriti (api kriti 'flint; to attack', suy kiiri 'metal', kxit-ci 'ax', kxit-twa 'fishhook', kiit-ho 'big knife'), PCJ *hata // *ha:da (xav hada-ra 'ax', xer hda);
PCerr *kroy 'rotten' > PNJ *kro (api, kay kro, tim kVo), PCJ *kroy // *kro (xav ?roy // ?ro, xer kroy // kro);
PCerr *kubntibn 'capybara' > PNJ *kumtimi (kay kunum, SUY kutumu, tap kofunu ~ kofuwu, tim kumtum, pnr intiq), PCJ *kumda (xav ?ubda, xer kumda);
PCerr *kukriti 'tapir' > PNJ *kukriti (api ku(?)krit, kay kukrit, SUY k(h)uk(h)iiri, tap kukxiri, tim kukhrit, pnr (i)kyiti), PCJ *kuta // *kuha:da (xav ?uta // ?uha:da, xer kta // kda);
PCerr *kupe 'to touch, to move' > PNJ *kupe (api kupe/ kupe-y^n, tim kupe/ kupe-d71 'to touch lightly; to reproach'), PCJ *kupi (xav ?upi, xer kupi 'to touch');
PCerr *kupu 'to wrap' > PNJ *kupu (api, kay, tim kupu), PCJ *kubu 'to cover' (xav ?ubu);
PCerr *kur9 'smooth' > PNJ *kur9 (tim kur9), PCJ *kure (xav ?ure, xer kure);
PCerr *ku&mi 'fire' > PNJ *ku^i (api kuvi, kay kuwi, SUY kwisi, tap kuti, tim kuhi, pnr isi), PCJ *kup§ma // *ku3a (xav ?upam§ // ?uja, xer kunma // kuza);
PCerr *kuy9 'bad smell, to stink' > PNJ *ku^9 (api kuc9, SUY, tap kut9, tim kuca), PCJ *kuje (xav ?uje, xer kuze);
PCerr *kVywa 'salt' > PNJ *ka^wa (api kacwa, SUY khatwa, tap katwa, tim ka:cwa), PCJ *kVkwa-((?) xav ?i?wa-wa:ha, xer kakwa-ra);
PCerr *kwt/*kwt-r ~ *kwi-p 'to hit, to strike' > PNJ *kwt/*kwt-p ~ *kwi-r (api kwiri, at=kwiri ~ at=kwi/pi=kwi-yp 'to break long objects', SUY ku=kwi, tim kwi/kwi-n, pnr kwi), PCJ *kwa/*kw§-ri 'to chop' (xer kwa/kwa-ri ~ kwa-r);
PCerr *kye/*kyed 'to drag, to pull' > PNJ *kye/ *kyed (api kze/ kze-dn, kay kye/ kye-dn, tim khye/ khye-dn, (?) pnr kr3-ri), PCJ *ca/ *ca-ri (xav ca/ ca-ri);
PCerr *kyey 'thigh' > PNJ *kye (api kze, kay kye, tap ce, tim khye), PCJ *jay // *ja (non-alternating *3) (xav jay // 3a, xer zda);
PCerr *mey 'to throw.SG' > PNJ *me/*me-p (api me/me-yp, kay me/me-yp, SUY me/me-ni, tap me/me-y 'to give, to throw', tim me/me-n), PCJ *mey // *me (xav mey // me, xer me);
PCerr *m3y 'greater rhea' > PNJ *m3 (api m3-ti, tap mv-ci, tim ma), PCJ *may // *ma (xav may // ma, xer ma);
PCerr *mo/*mo-r 'to go/come' > PNJ *mo/*mo-r (api mo/mo-r 'to go/come.PL', kay md/mo-ro 'to go/come.PL', suy mo/mo-lo 'to go/come.PL', tap md/mo-ro 'to go/come.PL', tim mo/mo-r, pnr mo(w)/ mo-ri), PCJ *mo/*mo-ri 'to go/come.SG' (xav mo/mo-ri, xer mo-ri);
PCerr *n3 'mother' > PNJ *n3 (api n3, kay na, SUY n3, pnr n3-py3), PCJ *na (xav na);
PCerr *no/*no-Pro 'to lie.SG' > PNJ *no/*no-r (api no/no-r, kay no/no-ro, SUY no/no-lo, tap no, tim no/no-n, pnr no), PCJ *no/*no-mro (xav no/no-mro, xer no-mro ~ n-mro ~ no-mr);
PCerr *qo/*po-r 'to give' > PNJ *qo/*po-r (api qo/po-r ~ po-t, kay qn/pd-rd, SUY qo/po-lo, tap qo, tim qo/yo-r, pnr s=o/yo-ri), PCJ *co/*co-mri (xav co/co-mri 'to give, to send', xer po/po-mri ~ po-mr);
PCerr *qoro/*pot 'to sleep' > PNJ *qoro/*pot (API qoro/pot, kay qoro/pot, SUY qoro/pono, tap pono/ poro, PNR soti), PCJ *poto // *po:no (finite: 2SG *ay=co) (xavpoto // po:no (finite: 2SG a=co), xer noto ~ not);
PCerr *pip=kopo 'nail, claw' > PNJ *pip=kopo (api pi(?)=kopo, kay pi=kop, tap kowo, pi=kowo, pi=kowo 'claw', tim yo=kop), PCJ *pi=kopo // *pi=ko:bo (xav pi=po, xer ni=krbo);
PCerr *pip=kra 'hand' > PNJ *pip=kra (apipi(?)=kra, kay pi=kra, SUYpi=kia, pi=kia, tappv=kxa, tim yo=khra, pnr yi=kya), PCJ *pip=kra (xavpip=krata // pip=kra:da (contamination with PCJ *krata // *kra:da 'stem, beginning'), xer nip=kra);
PCerr *piyi 'name' > PNJ *pi^i (api pici, kay piji, suy p=m, tap nti ~ nti=ndi, tim yici, pnr i(n)si), PCJ *pici // *pi:ci (xavpici // *pi:ci, xer nipi-ze);
PCerr *ptl*pt-Prt 'to sit.SG' > PNJ *pt/*pt-r (api ptlpt-r, KAY ptlpt-rt, SUY ptlpt-lt, TAP pi, TIM ytlyt-r, pnr pî), PCJ *pa-mra (finite: 2sg *ay=ca) (xav pa-mra (finite: 2sg a=ca), xer na-mra ~ n-mra);
PCerr *ptnt 'guts, faeces' > PNJ *pîni (api pîni, kay pîn, suy pîni, tap pîri, tim yîn, pnr yî), PCJ *pana (xav pana, xer nna);
PCerr *p5=kotô 'chest' > PNJ *po=kotô (apipo=kotô, kay pokot, tap pd=kondo, timp5=khot), PCJ *po=kutu ll *po=ku:du (xAVp5=?utu // p5=?udu, xER nô=knô);
PCerr *p5r5 'cord, vine' > PNJ *p5r5 (api poro 'vine, fiber', p3r=p5ro 'rope, vine', tim yor), PCJ *p5r5 ll *po:ro 'vein, sinew' (xAV poro // po:ro, xer wa=nro), *wede=poro // *wede=po:ro 'cord, rope, vine' (xAV wede=poro ll wede=po:ro, xer wde=nro);
PCerr *poyto 'tongue' > PNJ *poto (api po?to, kay poto, tap pop, tim yo?to, pnr soto), PCJ *poyto (xAV poyto ~ potto, xer noyto);
PCerr *pa 'arm, branch' > PNJ *pa (api, kay pa, SUY hwa, ttap hwa, tim, pnr pa), PCJ *pa-krata ll *pa-kra:da (xer pa-krta ll pa-krda), *wede=pa 'root' (xav wede=pa, xer wde=pa), (?) *payno 'arm' (xAV pano, xer payno);
PCerr *pal*pa-r 'to finish' > PNJ *pal *pa-r 'to finish, cessativelcompletive marker' (api pal pa-r, kay pa, SUY hwa, tap hwa, tim pa-r), PCJ *pal *pa-ri 'to finish, to erase' (xav pal pa-ri, xer pal pa-ri ~ pa-r);
PCerr *pal *pa-r 'to kill.PL' > PNJ *pal *pa-r (kay pa-rï, SUY hwal hwa-yï, tap hwa, pnr pa-rï), PCJ *pdl*pd-rî 'to kill.DU' (xav pa lpa-rî, xER pa lpa-rî ~ pa-r);
PCerr *para 'foot' > PNJ *parï (api para 'foot; jirau, wall', kay parï, SUY hway 'foot; jirau', tap hway 'foot; bed', tim par, pnr pa:), PCJ *para ll *pa:ra (xav para ll paira, xer pra 'foot, footprint');
PCerr *pel *pe-r ~ *pe-k 'to fart' > PNJ *pel*pe-k (api pel pe-k, SUY hwe, tim pe-k), PCJ *pil *pi-ri (xav pi l pi-ri);
PCerr *pîmî 'firewood' > PNJ *pî 'wood' (api, kay pî, suy hwî 'tree', tap hwî 'tree', tim pî), PCJ *mîmî ll *mî (xav mîmî ll mî, xer mmî);
PCerr *pty 'achiote' > PNJ *pt (api, kay pt, tap h% tim pt), PCJ *bay ll *ba (xav bay ll ba, xer ba);
PCerr *ptyi ~ *pîyi 'one' > PNJ *ptd>i (api ptci ~ pici, kay ptji, suy, tap wtti, tim ptci-t, pnr inpttï), PCJ *mîci (xav mîci, xer §=mî§ï), *pici (xer pi§i, pi§-tu ll pi§-du) 'only';
PCerr *po 'flat, wide' > PNJ *po (api, kay po, SUY ho, tim po), PCJ *po (xer po);
PCerr *poyo 'arrow, bamboo' > PNJ *po-t>e 'bamboo' (api po-?e, tim po?-he), PCJ *poko ll *po 'big arrow' (xav poko ll po);
PCerr *pro 'wife' > PNJ *pro (api pro, kay pro, SUY hlo, tap hro, tim pro 'wife (dead)'), PCJ *mro (xav mro, xer mro 'spouse, to marry');
PCerr *pr3 'ember' > PNJ *pr3 (api pr3, kay pr3 'ashes', SUY hh 'ashes', tim pr3, pr3y-pr3y 'coal'), PCJ *pre (xer pre-hika 'glowing embers', pre-nî^uri 'spark', pre-^apdo 'ember');
PCerr *pr3m 'hungry' > PNJ *pr3m3 'hungry, to want' (api pr3m3, kay pr3m, SUY h!3m3 'to want', tap hrvmv 'hungry', tim pr3m, pnr pr3nï), PCJ *mram ll *mra (xav mram ll mra 'hungry, food', xer mram ll mra);
PCerr *puru 'field' > PNJ *puru (api, kay puru, suy hulu, tap huru, tim pur, pnr pu:), PCJ *buru ll *bu:ru (xav buru ll bu:ru, xer bru);
PCerr *re 'to leave' > PNJ *rel*re-r (api relre-r ~ re), PCJ *rel*re-me 'to leave.SG' (xav rel*re-me, xer rë lrë-më ~ rë-m ~ r-më);
PCerr *rîti 'to look' > PNJ *rîti (api rîti, tim rît), PCJ *rîtî ll *rî:nî (xav rîtî ll rî:nî 'to look for', xer rîtî ~ rtî);
PCerr *tal *ta-d.n 'to tear' > PNJ *tal *ta-d.n (api ta 'to pick, to harvest', kay ta 'to cut', SUY kiv=ta 'to cut', tap kx$=ta 'to cut with one cut', tim tal tad 'to cut, to remove'), PCJ *ay=ta ll *ta-ri (xav a=tal ta-ri 'to get torn', xer ta-ri ~ ta-r 'to pluck, to pick, to tear');
PCerr *te 'tick' > PNJ *te (API te-ti, KAY te, tap [e, TIM te:-re, te:-ti), PCJ *ti: // *ti (xav ti:-?a, xer ti);
PCerr *tere 'fruit of a k. of palm' > PNJ *tere (api te-re, tim ter-re, ter-ti 'assai fruit'), PCJ *tiri // *ti:ri 'bacuri coconut' (xav tiri // *ti:ri);
PCerr *tepe 'fish' > PNJ *tepe (api, kay tep, SUY t(h)ewe, tap [ewe, tim tep, pnr tepi), PCJ *tepe // *te:be (xav tepe // te:be, xer tpe // tbe);
PCerr *tey 'leg' > PNJ *te (api, kay te, SUY the, tap [e, tim, pnr te), PCJ *tey // *te (xav tey // te, xer te);
PCerr *te/*te-m 'to go/come.SG' > PNJ *te/*te-m (api te/te-m, kay te/te-m, SUY t(h)e/t(h)e-me, tap [e/[e-me, tim te/te-m, pnr te/te-ri), PCJ *ne-m // *ne 'to go.DU' (xav ne-m // ne, xer ne-m ~ nema // ne);
PCerr *t3bn 'new, raw' > PNJ *t3bni 'raw' (api t3bn // t3m3, SUY hmi, tim fob*), PCJ *tem // *te (xav tem // te);
PCerr *t9^9 'hard, strong' > PNJ *t9$ (api t9yc/ tsyt, kay t9yc, SUY t9r9, tim t9y, PNR t9ti), PCJ *tete // te:de (xer tete ~ tte ~ tet), PCJ *pip=tete/ *ap=tete // *pip=te:de/ *ap=te:de 'to be strong' (xav pip=tete/ap=tete // pip=te:te/ap=te:te 'to get strong, to make an effort, to recover', xer nip=tete ~ nip=tet ~ nip=tte // nip=tde);
PCerr *tiki 'belly' > PNJ *tiki (kay tik, SUY t(h)iki, tim tik 'pregnant'), PCJ *diki // *di (xav di?i // di, xer dki // di 'belly; to fill the stomach');
PCerr *ti/*tiki/*ti-r 'to die' > PNJ *ti /*ti-k (api, kay ti/ti-k, suy ti/ti-ki, tap [i/[i-gi, tim ti/ti-k, pnr ti), PCJ *da?a // da/ da-ra (xav da?a // da / da-ra, xer dka // da / da-ra);
PCerr *tiki 'black' > PNJ *tiki (api tiki, kay tik, SUY tiki, tap [igi, tim tik, pnr ka=ti:), PCJ *daka // *da (xav ?ra=da, xer dka 'dark', (?) wa=k(r)=ti // wa=k(r)=di);
PCerr *totoko 'to throb' > PNJ *totoko (tim totok 'heartbeat, heart'), PCJ *totoko // *toto (xav toto, xer totoko ~ totko);
PCerr *toy 'brother' > PNJ *to (api, suy to, tim to 'brother (alive)', pnr to), PCJ *noy // *no 'younger sibling of the same sex' (xav noy // no, xer no-re);
PCerr *tu/ *tu-r 'to carry' > PNJ *tu/ *tu-r (api tu, kay tu/ tu-ru, tim tu/ tu-r, pnr tu-ri 'to carry in a basket.SG'), PCJ *du // *du-ri 'to carry.SG' (xav du // du-ri, xer du /du-ri ~ du-r);
PCerr *tu/ *tu-r 'to urinate' > PNJ *tu/ *tu-r (api tu/ tu-r, SUY tu), PCJ *to/*to-ri (xav to/to-ri, xer to-ri);
PCerr *tubn 'old, tall (?)' > PNJ *tumu 'old' (api tumu, kay tum, SUY tumu, tap [umu, tim tum, pnr =tu), PCJ *dum // *du 'tall' (xer dum), *dum-krata // *dum-kra:da 'elder sibling of the opposite sex' (dub-?rata // dub-?ra:da, xer dum-krda);
PCerr *tumu 'belly' > PNJ *tu (api tu 'belly, intestine', kay tu, tap [u, tim, pnr tu, i=tu 'tuber'), PCJ *nomo // *du (xav nomo // du, xer nmo 'fat (adj.)');
PCerr *tyete 'to burn' > PNJ *tyete (api cete, kay cet/ cere, SUY sere, tap cere ~ tere, tim cet, pnr titi), PCJ *3ata // *3a:da (xav jata // ja:da, xer zata);
PCerr %i 'bone' > PNJ *U (api i ~ zi, kay ?i, suy si, tap ti, tim hi, pnr si), PCJ *hi (xav, xer hi);
PCerr *%i 'meat' > PNJ *%i ~ *pi (api, kay, suy, tap pi, tim yi ~ hi, pnr yi), PCJ *pi (non-alternating *p) (xav pi, xer ni);
PCerr %imi 'seed' > PNJ %i (api, kay ?i ~ i, suy si, tap ti, tim hi, pnr si), PCJ *pama // *3a (xav p§m§ // 3a, xer za);
PCerr *\f,oy 'leaf' > PNJ (api ?o ~ o, kay ?o, SUY hwi=so, tap to ~ hwi=to, tim ho 'leaf, a hair', pnr pari=so), PCJ *cuy // *cu (xav we=cuy-ra, =cuy // =cu, xer pu 'a hair');
PCerr *^wako 'coati' > PNJ *^wako (api wako, SUY swako, tap toako, tim wakho), PCJ *wako (xav wa?o, xer wako);
PCerr *weke 'partridge' > PNJ *beke (tim pekhe), PCJ *wiki (xav wi?i, xer wiki ~ wiki);
PCerr *w3d3 'tree' > PNJ *b3r3 (api p3r3 ~ para 'tree, horn, plant, trunk, canoe, car', kay b3ri 'tree, horn', tim p3r 'tree, horn', pnr p3ri 'tree', ky3=si=p3ri 'horn'), PCJ *wede // *we:de (xav wede // *we:de, xer wde);
PCerr *wi/*wi-r 'to kill.SG, to extinguish' > PNJ *bi/*bi-r (api pi/pi-r, kay bi/bi-n, SUY pi/pi-Ti, tap mi/mi-ri, tim pi/pi-r 'to kill (with an arrow), to extinguish'), pnr pi/pi-ri), PCJ *wi/ *wi-ri (xav wi/wi-ri, xer wi/ wi-ri ~ w-rT);
PCerr *ya 'to stand.SG' > PNJ %a/*fo-bn ~ *$>a-m (api ca/c3-bn ~ ca-r, kay ja/ja-m, SUY ta/ta-ma, tap ta, tim ca/ cab ~ c3b ~ ca-r, pnr S3 ~ sa:y), PCJ *ja (xav3a, xer da);
PCerr *yad-kwa 'mouth' > PNJ *yar-kwa (api ya-kva, kay, suy yay-kwa, tap yay-kwa, tim yar-kwa, pnr sa-koa), PCJ *3ada-wa (xav 3adawa, xer tdawa 'mouth, door');
PCerr *yat,e 'nest' > PNJ *ya^e (api, kay yae, tim ya:he, pnr sase), PCJ *3aci (xav 3aci, xer papi 'to make a nest');
PCerr *y3 'pain, to hurt' > PNJ (api c3, SUY, tap t3, tim c3, pnr s3 /s3-ri), PCJ *3e (xav 3e, xer %e);
PCerr *y3 'urine' > PNJ 'urine, bladder' (api c3, tim c3), PCJ *3e (xav 3e, 3e-kre 'bladder', xer Tf 'bladder');
PCerr *=y3 (instrumental suffix) > PNJ (api =c3, kay =33, SUY, tap =t3, tim =c3), PCJ *=3e (xav =3e, xer =%e);
PCerr *y3/ *y3-Pr3 'to enter.SG' > PNJ / (api a=c3/ c3-r, kay wa=j3/ wa=j3-r3, SUY a=t3/ t3=t3, tim c3/ c3-r, pnr s3), PCJ *3e / *3e-bre (xav 3e/3e-bre, xer %e-bre ~ %-bre ~ de-bre);
PCerr *y9 'bitter' > PNJ (api €9, kay 39, tap t9, tim ca), PCJ *3e (xav 3e, xer %e);
PCerr *yi/ *yi-r 'to put, to lay' > PNJ %i/ %i-ri (api ci/ci-ri, kay ji/ji-ri, SUY ti/ ti-ri, tim ci/ci-r), PCJ *hi/ *hi-ri 'to put.SG' (xav hi/ hi-ri 'to put, to leave.SG', xer hi/ hi-ri 'to put, to cook, to deter-mine.SG');
PCerr *yo№ 'to grind' > PNJ *$>obn 'flour, powder' (api cob // como, kay jobn, (?) tap to=tom-ci, tim cobn), PCJ *cum // *cu (xav §u, xer §um ~ §uma // §u);
PCerr *yuPru 'pus' > PNJ %u ~ *$>uru (api cu ~ curu, kay juru, tap turu, tim cur), PCJ *3ubruy (xav 3ubruy // 3ubru, xer Tpruy // %bru);
PCerr *ywa 'tooth' > PNJ *$>wa (api cwa, kay jwa, SUY twa, tap twa, tim cwa, pnr swa ~ soa), PCJ *kwa (xav ?wa, xer kwa).
A non-exhaustive list of grammatical morphemes includes:
PCerr *wa 'I (nominative)' > PNJ *ba (api wa, kay ba, SUY wa/ pa, tap wa, tim wa/ pa), PCJ *wa (xav wa-hd, xer wa, wa-ha);
PCerr *i^n- 'I (oblique)' > PNJ (api i-/ ic-/ip-, kay, suy, tap i-, tim iy-), PCJ *i:- (xav ?i:-, xer i-);
PCerr *ga 'thou (nominative)' > PNJ *ga (api ka, kay ga, SUY, tap, tim, pnr ka), PCJ *ka (xav ?a-ha, xer ka, to=ka);
PCerr *ay- 'thou (oblique, class I)' > PNJ *a- (api, kay, suy, tap, tim a-, pnr a- ~ ha-), PCJ *ay (xav ?ay-, xer ay-);
PCerr *m3 'dative postposition' > PNJ *m3 (api, kay, suy, tap, tim, pnr m3), PCJ *ma (xav, xer ma);
PCerr *pim 'genitive postposition' > PNJ *pim (pnr yi), PCJ *pim (xavpim, xer nim);
PCerr *te 'ergative postposition' > PNJ *=te (api, kay, suy =te, tap =re, tim =te), PCJ *te (xav =te, xer =te).
The following comparisons are somewhat problematic:
PCerr *nbepë 'honey' > PNJ *nbe^nï (api nbey^n, kay meyfc1, SUY nbenï, tap nbey, tim pedn, pnr na=peyj), PCJ *pTnT/pi (xav pTnT/pi) 24;
PCerr *proto ~ *proy > PNJ *proto 'to run' (api pronto, kay pro"i, SUY hlono), PCJ *ay=ca=mroy // *ay=ca=mro 'to run.DU' (xav a=ca=mroy // a=ca=mro);
PCerr *y3^n¿ ~ *y£y 'sweet, tasty' > PNJ (api ay^, kay 33y^n, tim c3dn), PCJ *jey // *je (xav 3ey // 3e, (?) xer ^ey).
5. Conclusion
In this paper I have undertaken the first attempt at a systematic reconstruction of Proto-Cerrado phonology and lexicon. However, a number of issues remain to be tackled. These include:
— evolution of Central Jê verbal morphology and morphosyntax with a special attention to finiteness, verbal number and alignment;
— identification of the remaining correspondences between Proto-Northern Jê and ProtoCentral Jê, particularly involving diphthongs and codas;
— emergence of unexpected nasality in certain Central Jê roots;
— the origin of Central Jê stems belonging to the classes A and C.
Yet another issue that remains beyond the scope of this paper are contacts between Central Jê languages and Northern Jê languages. in addition to evident Central Jê borrowings in individual Northern Jê languages (Xavánte ?utd // ?uhd:dd 'tapir' > Tapayúna uhd-ci ~ uh3-ci, Central Jê *bdtd // *bd:dd or Xavánte bdtd // bdidd 'sun' > Panará wdtd-ti25), a number of roots reconstructible to Proto-Northern Jê and to Proto-Central Jê display completely irregular sound correspondences, suggestive of their loan origin in one of the branches (or in both), cf. 'southern tamandua' (PNJ *p3t3, PCJ *pati // *pa:di)26, 'paca' (PNJ wgra, PCJ *krawa), 'tobacco' (PNJ *kare^nï ~ PCJ *warT), 'Babassu palm' (PNJ *r5ro, PCJ *noroy). It is also necessary to
mention the curious case of the "Menren" 27 wordlist provided by Loukotka (1963: 51-54), which represents a Timbira variety close to Pykobjê (as suggested by its vocalism) and contains a considerable number of word of obvious Central Jê origin (Nikulin 2015: 26-27).
According to the principles of bottom-up reconstruction, a reconstruction of Proto-Jê is necessary in order to warrant a coherent use of Jê data in Macro-Jê comparative studies. i intend to proceed to this stage in a future paper.
References
Albuquerque, Francisco Edviges. 2011. Gramática pedagógica da lingua Apinajé. Goiânia, GO: Editora da PUC Goiás. Albuquerque, Francisco Edviges (org.). 2012. Dicionário Escolar Apinayé. PanhT Kapër - Kupê kapër. Apinayé -
Portugués. Belo Horizonte, MG: Editora da Faculdade de Letras - UFMG.
24 If Xavante pini // pi is a result of leveling of PCJ *pini // *pi, it may be compared to Xerente pi (as in to=pi 'eye rheum'), Proto-Southern Je *nbe 'liquid', Rikbaktsa pi-hik, Karaja be, Proto-Jabuti *nbi 'water'. In this case, Proto-Cerrado *nbene should be reconstructed, and the etymology of PNJ *nbe^,ni has to be viewed as unknown.
25 This etymology has been suggested by Carvalho (2016: 71-72). A native synonym, inpiti, also exists in Panara; this item is a true cognate of PCJ *bata // *ba:dd.
26 As Ribeiro and Voort (2010: 557) point out, vaguely similar words exist in Karaja, Jabuti, Tupi and Karib languages, which suggests that this item might by a Wanderwort.
27 A mistranscription of me=he 'Indians', lit. 'our meat'.
Barbosa, Alexandre de Sousa. 1918. Cayapó e panará. Manuscript. Available online at <http://etnolinguistica.wdfiles. com/local--files/biblio%3Abarbosa-1918-cayapo/barbosa_1918_cayapo.pdf>.
Bardagil-Mas, Bernat. 2016. Negation mechanisms in Panará (Jê). Amazónicas VI. Coloquio Internacional AMAZÓNICAS ("La estructura de las lenguas amazónicas") (abstract book), pp. 70-71. Leticia/Tabatinga: Universidad Nacional de Colombia; Universidade do Estado do Amazonas; Instituto Caro y Cuervo.
Braggio, Silvia Lucia Bigonjal. 2004. Revisitando a fonética/fonologia da língua Xerente Akwe: uma visäo comparativa dos dados de Martius (1866), a Maybury-Lewis (1965) com os de Braggio (2004). Signótica 17(2): 251-274.
Burgess, Eunice. 1971. Duas análises das sílabas do xavánte. In: Sarah C. Gudschinsky (ed.). Estudos sôbre línguas e culturas indígenas, pp. 96-102. Brasilia: Summer Institute of Linguistics.
Camargo, Nayara da Silva. 2010. Lingua Tapayúna: aspectos sociolingüísticos e uma análise fonológica preliminar. MA thesis. Campinas: Universidade Estadual de Campinas.
Camargo, Nayara da Silva. 2015. Tapayuna (Jê): aspectos morfossintáticos, históricos e sociolinguísticos. PhD. Campinas: Universidade Estadual de Campinas.
Carvalho, Fernando O. de. 2016. On the Development of the Proto-Northern Jê Rhotic in Panará Historical Phonology. Anthropological Linguistics 58(1): 52-78.
Carvalho, Fernando O. de, Gean Nunes Damulakis. 2015. The structure of Akroá and Xakriabá and their relation to Xavante and Xerente: A contribution to the historical linguistics of the Jê languages. Línguas Indígenas Americanas 15(1): 17-46.
Castro Alves, Flávia de. 1999. Aspectos fonológicos do Apaniekrá (Jê). MA thesis. Sao Paulo: Universidade de Sao Paulo.
Cotrim, Rodrigo Guimaraes Prudente Marquez. 2016. Uma descrigäo da morfologia e de aspectos morfossintáticos da língua Akwê-Xerente (Jê Central). PhD. Brasilia: Universidade de Brasilia.
D'Angelis, Wilmar da Rocha. 1998. Traços de modo e modos de traçar geometrias: línguas Macro-Jê & teoria fonológica. PhD. Campinas: Universidade Estadual de Campinas.
Damulakis, Gean Nunes. 2010. Fonologias de línguas Macro-Jê: uma análise comparativa via teoria da otimalidade. PhD. Rio de Janeiro: Universidade Federal do Rio de Janeiro.
Davis, Irvine. 1966. Comparative Jê phonology. Estudos Lingüísticos: Revista Brasileira de Lingüística Teórica e Aplicada 1(2): 20-24.
DMK = Dicionário Multimídia Kîsêdjê. Online resource. Available online at <http://prodoclin.museudoindio.gov.br/ index.php/etnias/kisedje/dicionario-multimidia>.
Ehrenreich, Paul. 1895. Materialien zur Sprachenkunde Brasiliens: III. Die Sprache der Akuä oder Chavantes und Cherentes (Goyaz). Zeitschrift für Ethnologie 27: 149-162.
Estevam, Adriana Machado. 2011. Morphosyntaxe du xavante. Langue jê du Mato Grosso (Brésil). PhD. Paris: Université Paris Diderot (Paris 7).
Frazao, Kêt Simas. 2013. A sílaba no Akwê-Xerente (Jê). MA thesis. Brasilia: Universidade de Brasilia.
Hall, Joan, Ruth Alice McLeod, Valerie Mitchell. 1987 [2004]. Pequeno dicionário Xavante-Português, Português-Xavante. Damreme pibuzé hä a'uwê mreme, warazu mreme na te te isapri mono zém na duré warazu mreme, a'uwê mreme na te te isapri mono zém na. Cuiabá: Sociedade Internacional de Lingüistica.
Ham, Patricia. 1961. Apinayé Grammar. Arquivo Linguistico, n. 108. Brasilia: Summer Institute of Linguistics.
Krieger, Wanda Braidotti, Guenther Carlos Krieger (orgs.). 1994. Dicionário escolar: Xerente-Português; Português-Xerente. Rio de Janeiro: Junta das Missöes Nacionais da Convençao Batista Brasileira.
Lapierre, Myriam, Bernat Bardagil-Mas, Andrés Pablo Salanova. 2016a. The nasal consonants of Panará. A talk given at the 21st Workshop on the Structure and Constituency of Languages of the Americas. Montréal: Université du Québec à Montréal.
Lapierre, Myriam, Andrés Salanova, Bernat Bardagil-Mas. 2016b. A reconstruction of Proto-Northern Jê phone-mics. Amazónicas VI. Coloquio Internacional AMAZÓNICAS ("La estructura de las lenguas amazónicas") (abstract book), pp. 110-111. Leticia/Tabatinga: Universidad Nacional de Colombia; Universidade do Estado do Amazonas; Instituto Caro y Cuervo.
Lapierre, Myriam. 2017. Post-oralized and devoiced nasals in Panará. A talk given at the Symposium on Amazonian Languages II (April 8, 2017). Berkeley: University of California.
Loukotka, Cestmir. 1963. Documents et vocabulaires inédits de langues et de dialectes sud-américains. Journal de la Société des Américanistes 52: 7-60.
Martius, Carl Friedrich von. 1867. Beiträge zur Ethnographie und Sprachenkunde Amerikas zumal Brasiliens. I. Zur Ethnographie. II. Glossaria linguarum Brasiliensium. Leipzig: Friedrich Fleischer.
Mattos, Rinaldo de. 1973. Fonemica Xerente. Série Lingüística 1: 79-100.
McLeod, Ruth. 1974. Fonemas xavánte. Série Lingüística 3: 131-152.
McLeod, Ruth, Valerie Mitchell. 1977 [2003]. Aspectos da língua xavante. Cuiabá: Sociedade Internacional de Lingüística.
Nikulin, Andrey. 2015. Verifikaciya gipotezy o zhe-tupi-karibskom rodstve [On the genetic unity of Je-Tupí-Karib]. MA thesis. Moscow: Moskovskiy gosudarstvennyy universitet imeni M. V. Lomonosova.
Nikulin, Andrey. 2016a. Proto-Je revisited: phonology, morphophonology and language contact. Amazónicas VI. Coloquio Internacional AMAZÓNICAS ("La estructura de las lenguas amazónicas") (abstract book), pp. 108-109. Leticia/Tabatinga: Universidad Nacional de Colombia; Universidade do Estado do Amazonas; Instituto Caro y Cuervo.
Nikulin, Andrey. 2016b. Historical phonology of Proto-Northern Je. Journal of Language Relationship 14(3): 165-185.
Nikulin, Andrey, Fernando O. de Carvalho. 2017. Prehistoria de las lenguas y familias lingüísticas del Gran Chaco, de la meseta brasileña y cercanías: Propuesta de base de datos léxicos y resultados preliminares. IV Encuentro de lenguas indígenas americanas (abstract book), pp. 37-38. Santa Rosa: Universidad Nacional de La Pampa.
Oliveira, Christiane Cunha de. 2005. The language of the Apinajé people of Central Brazil. PhD. Eugene: University of Oregon.
Oliveira, Christiane Cunha de, Eduardo Rivail Ribeiro. 2005. Vowel shift in Central Je. A talk given at the XVII International Conference on Historical Linguistics. Madison: University of Wisconsin.
Pessoa, Katia Nepomuceno. 2012. Análise fonética e fonológica da língua Krenak e abordagem preliminar de contos Botocudo. PhD. Campinas: Universidade Estadual de Campinas.
Pickering, William Alfred. 2010. A fonologia xavante: uma revisitagäo. PhD. Campinas: Universidade Estadual de Campinas.
Popjes, Jack, Josephine Popjes. 1971 [2009]. Phonemic Statement of Canela. Anápolis, GO: Associagáo Internacional de Linguística, SIL - Brasil.
Quintino, Wellington Pedrosa. 2000. Aspectos da fonologia xavante. MA thesis. Campinas: Universidade Estadual de Campinas.
Quintino, Wellington Pedrosa. 2012. Aspectos da fonologia xavante e questoes relacionadas: rinoglotofilia e nasalidade. PhD. Rio de Janeiro: Universidade Federal do Rio de Janeiro.
Ramirez, Henri, Valdir Vegini, Maria Cristina Victorino de Franca. 2015. Koropó, puri, kamakä e outras línguas do Leste brasileiro: revisao e proposta de nova classificagäo. Línguas Indígenas Americanas 15(2): 223-277.
Ribeiro, Eduardo Rivail, Hein van der Voort. 2010. Nimuendajú was right: The inclusion of the Jabutí language family in the Macro-Je stock. International Journal of American Linguistics 76: 517-570.
Ribeiro, Rosa Maria de Lima. Dicionário Arikapu-Portugues: registro de uma língua indígena amazónica. MA thesis. Guajará-Mirim: Universidade Federal de Rondönia, Campus de Guajará-Mirim.
Rodrigues, Aryon Dall'Igna. 1999. Macro-Je. In: Dixon, R. M. W., Alexandra Y. Aikhenvald (eds.). The Amazonian Languages, pp. 165-206. Cambridge: Cambridge University Press.
Rodrigues, Aryon Dall'Igna. 2004. Sobre a possível diferenga fonética entre a fala masculina e feminina em Karajá. Línguas Indígenas Americanas 4: 115-122.
Rodrigues, Aryon Dall'Igna. 2012. Flexao relacional no tronco linguístico Macro-Je. Revista Brasileira de Linguística Antropológica 4(2): 267-277.
Silva, Talita Rodrigues da. 2011. Descrigäo e análise morfossintática do nome e do verbo em Pykobje-Gaviäo (Timbira). MA thesis. Sao Paulo: Universidade de Sao Paulo.
Silva, Talita Rodrigues da. 2012. Análise sobre a tipologia da ordem dos constituintes do dialeto indígena Pykobje-Gaviao (Timbira). Estudos Linguísticos 41(1): 231-245.
Salanova, Andrés Pablo. A flexao de terceira pessoa nas línguas Je. Línguas Indígenas Americanas 11: 75-114.
Siqueira, Kenia Mara de Freitas. Aspectos do substantivo na língua Xerente. MA thesis. Goiánia: Universidade Federal de Goiás.
Sousa Filho, Sinval Martins. 2007. Aspectos Morfossintáticos da Língua Akwe-Xerente (Je). PhD. Goiánia: Universidade Federal de Goiás.
Souza, Shelton Lima de. 2008. Descrigäo fonético-fonológica da língua akwen-xerente. 2008. MA thesis. Brasília: Universidade de Brasília.
Vasconcelos, Eduardo Alves. 2013. Investigando a hipótese Cayapó do Sul-Panará. PhD. Campinas: Universidade Estadual de Campinas.
Vasilyev, Mikhail, Mikhail Saenko. K voprosu o tochnosti glottoxronologii: datirovanie jazykovoj divergencii po dannym romanskix jazykov. [How accurate can glottochronology be? Dating language divergence on the basis of Romance data.] Journal of Language Relationship 15(2): 114-135.
Voort, Hein van der. 2007. Proto-Jabutí: um primeiro passo na reconstruyo da língua ancestral dos Arikapú e Djeoromitxí. Boletim do Museu Paraense Emilio Goeldi. Ciencias Humanas 2(2): 133-168.
А. В. Никулин. Реконструкция фонологии праязыка нагорной ветви семьи же
Статья продолжает серию публикаций по исторической фонологии языков южноамериканской макросемьи макро-же. Автор устанавливает систему регулярных фонетических соответствий между языками центральной группы семьи же (шаванте и шеренте) и предлагает на их основе реконструкцию праязыка этой группы. Затем эта реконструкция сопоставляется с реконструкцией праязыка северной группы семьи же (№ки1т 2016Ь), и предлагается реконструкция общего праязыка для этих двух групп, пранагорного же. В статье приводится не претендующий на полноту список этимоло-гий для нагорной ветви семьи же.
Ключевые слова: языки же, языки макро-же, язык шаванте, язык шеренте, языковая реконструкция, сравнительно-исторический метод.
