Reconstructing the category of 'associated motion' in Tacanan languages (Amazonian Bolivia and Peru)

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages (Amazonian Bolivia and Peru)*

Antoine Guillaume Laboratoire Dynamique du Langage (CNRS & Université Lumière Lyon 2)

This paper deals with the diachrony of the associated motion (AM) category in the Tacanan languages of the Amazonian lowlands of Bolivia and Peru. The category of AM consists of grammatical markers that attach to non-motion verbs and specify that the verb action (V) occurs against the background of a motion event (e.g., ‘go and V’, ‘V while going’, etc.). The AM systems of Tacanan languages are noteworthy for their remarkably high degree of complexity, reaching levels rarely found in languages of other parts of the word. This raises several questions, one of which is how they evolved historically. In this paper, I gather for the first time the information available on the AM systems of the five Tacanan languages (Araona, Cavineña, Ese Ejja, Reyesano and Tacana) and attempt to reconstruct their past. I first argue that between one and four AM markers can be reconstructed in Proto-Tacanan. Secondly, I argue that several AM markers that are not reconstructible have developed recently out of independent verbs of motion and I make the hypothesis that these have been directly copied into already well established AM paradigms.

1. Introduction

This paper is about reconstructing the origin of a particular type of motion-and-direction markers in the Tacanan languages, a small family of five languages

* I am indebted to the Cavineña, Reyesano and Tacana people for generously sharing with me data and thoughts on their respective languages. I am thankful to Spike Gildea for provid-ing me with useful feedback on the reconstruction that is presented here. Finally, I am grateful to the following people who have read and commented on previous versions of this paper: Carola Emkow, Harold Koch, Marine Vuillermet, two anonymous reviewers, as well as Ritsuko Kikusawa and Lawrence Reid, the editors of the present volume. The usual disclaimer applies.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

130 Antoine Guillaume

spoken in the Amazonian lowlands of Northern Bolivia and Southeastern Peru. The Tacanan languages are listed in Table 1, including estimates about their cur-rent number of speakers as compared to the number of ethnic group members, and with references to the main studies available on these languages.

Figure 1 gives Key’s (1968) and Girard’s (1971) internal classification of the Tacanan languages consisting of three branches (Kavinik, Chamik and Takanik), all placed at the same level within the family tree. Even though I will make use of this classification in this paper, it must be stated that it is exclusively based on phonological and lexical reconstructions, and on fairly limited material. Work remains to be done in order to fully confirm it and to investigate whether the branches can be put into a more complex hierarchy.

The type of system that will be reconstructed is cross-linguistically very rare and not discussed in the general typological literature. It corresponds to what

Branches Languages

Kavinik Cavineña

Proto-Tacanan Chamik Ese Ejja

Araona

Takanik Reyesano

Tacana

Figure 1. Internal classification of Tacanan languages (Key 1968; Girard 1971)

Table 1. Tacanan languages (figures from Crevels & Muysken 2009)

Name Location No. speakers No. ethnic group members

Main grammatical studies on the languages

Araona Bolivia 111 158 Pitman (1980), Emkow (2006)Ese Ejja Bolivia & Peru 518 732 Vuillermet (2012)Cavineña Bolivia 601 1683 Camp & Liccardi (1989),

Guillaume (2008)Reyesano Bolivia 12 4019 Guillaume (2009a, 2012,

field notes 2004–2008)Tacana Bolivia 50 7345 Guillaume (field notes

2009–2012)

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 131

has been described under the term “associated motion” (henceforth AM) in the literature on Australian aboriginal languages (Koch 1984, 2006; Tunbridge 1988; Wilkins 1991) for referring to grammatical markers that attach to non-motion verbs and specify that the verb action occurs against the background of a motion event with a specific orientation in space.

The AM systems of Tacanan languages and of many of their neighboring lan-guages are noteworthy for their remarkably high degree of elaboration and com-plexity, reaching levels rarely found in languages of other parts of the world. This raises several questions, one of which is how they evolved historically. In this pa-per, I gather for the first time the information available on the AM systems of the five Tacanan languages in order to compare them synchronically, and I attempt a reconstruction of their development.

The paper is organized as follows. Section 2 provides a working typological framework for the analysis of complex AM systems in general. In Section 3 this framework is used to present and compare the synchronic systems of AM in the five Tacanan languages. The results show that, except for Reyesano, the Tacanan languages have very complex AM systems, ranging from 7 markers in Araona to 13 markers in Tacana. Section 4 briefly discusses the category of AM in neigh-boring languages. It shows that complex AM systems are not restricted to the Tacanan languages in the region but are also found in many other languages from several distinct genetic groupings. This fact suggests that the phenomenon of dif-fusion must have played a major role in the development of the AM systems in the languages of the area, and must therefore be seriously taken into account when attempting to reconstruct the AM system of any of these languages. Section 5 is dedicated to the reconstruction of the AM systems of the Tacanan languages. I first propose a reconstruction of four AM markers in proto-Tacanan, using the comparative method (§5.1). I then show that at least some of the remaining non-reconstructible AM markers in the different language must have originated in independent verbs of motion, and put forward the hypothesis that this happened through a spontaneous process of copying within a well established construction of AM present in the different languages (§5.2). Finally, a conclusion summarizes the findings.

2. The category of associated motion: Definition and typological framework of analysis

The term “associated motion” first occurred in the literature on Australian ab-original languages, in particular in the work of Koch (1984, 2006) and Wilkins (1991), for referring to grammatical markers that modify non-motion verbs and

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

132 Antoine Guillaume

specify that the verb action occurs against the background of a motion event. In previous publications (Guillaume 2006, 2008: 212–236, 2009b), I have shown that very similar types of markers are found in the Tacanan language Cavineña, such as the suffix ‑ti, which attaches to the non-motion verbs ‘bathe’ in (1a) and ‘see’ in (1b), and expresses the fact that the actions of ‘bathing’ and ‘seeing’ are preceded by a going motion of their subject argument.1, 2

(1) Prior motion (andative) a. Intransitive non-motion verb Nawi-ti‑wa=jari =pa. bathe-go.temp&do-perf=still =rep ‘He has gone to bathe.’ b. Transitive non-motion verb Tudya =ekwana ba-ti‑kware takure. then =1pl see-go.temp&do-rem.past chicken ‘So we went tosee the chicken (in the back of the bus).’

The comparison that I was able to make between the system of AM in Cavineña (12 markers) and other complex systems of AM in Australian languages revealed that a number of organizing parameters are recurrent in the different languages. These parameters can form the basis of a typological framework for analytical and comparative purposes. In what follows, I will discuss and illustrate these param-eters and the proposed analytical framework with data from Cavineña.

A first and central organizing parameter has to do with the temporalrelation that holds between the motion indicated by the suffix and the action expressed by the main verb. Along this parameter, there can be three different types of AM markers:

1. The phonetic value of the graphemes used for writing the different Tacanan languages in this study (coded as follows: A = Araona, C = Cavineña, E = Ese Ejja, R = Reyesano, T = Tacana) is that of their equivalent IPA symbols, except for the following: b[ɓ] in E and [mb] in R; ch [ʨ]; dh[ð] in R; j [h]; r [ɺ] in C; s[s] in T; sh [ɕ]; u [ʊ]; wi [ßi] in C; y [j]; x [χ] in E; ‘ [ʔ] in E (sources: Pitman & Pitman 1970 and Emkow 2006 for Araona; Guillaume 2008 for Cavineña; Vuillermet 2012 for Ese Ejja; Guillaume 2012 and field notes 2004–2008 for Reyesano; and van Wynen & van Wynen 1962 and Guillaume field notes 2009–2012 for Tacana).

2. Abbreviations used in this paper are: a = agent/subject of transitive verb; am = associated motion; bm = boundary marker; dc = deictic center; do = do the verb action; foc = focus; imp = imperative; o = object/patient of transitive verb; perf = perfect; perm / perm. = per-manent; reitr = reiterative; rem.past = remote past; rep = reportative; restr = restrictive; s = unique argument of intransitive verb; strg.emph = strong emphasis; temp = temporary.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 133

1. markers that specify a motion prior to the action, as with the suffix ‑ti in (1a, b);

2. markers that specify a motion concurrent with the action, as with the suffix ‑aje in (2), which expresses the fact that the action of ‘screaming’ is realized concurrently to a going motion of its subject argument;

3. markers that specify a motion subsequent to the action, as with the suffix ‑kena in (3), which expresses the fact that the action of ‘drinking’ is followed by a going motion of its subject argument.

(2) Concurrent motion (andative) Tume =pa juje kiketere‑aje‑kware. then =rep duck scream-do.ipfv.going-rem.past ‘The duck went screaming.’

(3) Subsequent motion Refresco=kamadya =tuke =Ø iji-kena‑wa. soft.drink=restr =3sg =1sg drink-do&go-perf ‘I just hadasoft‑drink as I was leaving (my house).’

A second recurring parameter concerns the identification of the moving argu‑ment. There can be markers that indicate the motion of the subject (transitive A or intransitive S), as is the case with the three Cavineña suffixes discussed above (‑ti, ‑aje and ‑kena). By contrast, there can be markers that are exclusively used with transitive verbs and indicate the motion of the object (O) argument, as is the case with the Cavineña suffix ‑dadi in (4), which expresses the fact that the action of ‘seeing’ is associated with a going motion of its O argument.

(4) Motion of O argument (andative) [Peadyaekwita]=tuke=Ø ba-dadi‑wa… one person =3sg =1sg see-go(O)-perf ‘Isawa man going away from me.’

A third recurring parameter has to do with the directionality of the motion (i.e., Talmy’s typological category of Path).3 The most common path distinction ex-pressed in AM systems appears to be that of deixis, i.e., the orientation of the mo-tion vis-à-vis the deictic center (DC), typically the position of the speaker. There can be markers that specify an andative motion, namely one that is directed away from the DC, as is the case with ‑ti and ‑aje in (1) and (2) above. By contrast there can be markers that specify a venitivemotion, namely one that is directed toward

3. “The Path […] is the course followed or site occupied by the Figure object with respect to the Ground object” (Talmy 1985: 61).

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

134 Antoine Guillaume

the DC, as is the case with ‑na and ‑be in (5) and (6), which expresses the fact that the actions of ‘entering’ and ‘eating’ are associated with a coming motion of their subject arguments.4

(5) Prior motion (venitive) Nubi-na‑kwe=dya =di! enter-come.temp&do-imp.sg=foc = strg.emph ‘Come and enter!’

(6) Concurrent motion (venitive) Jadya =tu amena ara-be‑kware era. thus =3sg bm eat-do.ipfv.coming.temp-rem.past 1sg.erg ‘So I came eating along the way.’

Finally, depending on the languages and the level of complexity of particular AM systems, finer distinctions can be found, such as the aspectualrealization of the verb action, whether it is repetitive/continuous (imperfective), as with ‑aje in (2) and ‑bein (6), or punctual (perfective), as with ‑nati in (7), or the stabilityofthemotiontarget, with motion markers that specify a ‘temporary/unstable’ target, as with ‑ti in (1a, b) and ‑na in (5), and others that specify a ‘permanent/stable’ tar-get, as with ‑diru in (8).

(7) Concurrent motion (andative, punctual realization) [Jukuri turu ebari]=tuke=Ømee=ju ba-nati‑kware. coati big.male big=3sg=1sg saltlick=loc see-do.pfv.going-rem.past ‘I saw a big male coati in a saltlick onmyway.’

(8) Prior motion (andative, permanent target) Yuwa ani-diru‑kwe waku=ju! over.there sit-go.perm&do-imp.sg bench=loc ‘Go and sit (and stay) over there!’

The most crucial parameters for making sense of the semantic and functional distinctions made by the different members of AM systems are the first three, namely (1) the temporal relation holding between the motion and the action, (2) the identification of the moving argument and (3) the directionality/path of the motion. These are the parameters that I will use below for my cross-Tacanan comparative study of AM.

4. The motions implied by the suffixes ‑dadi ‘go(O)’ in (4) and ‑kena ‘do&go’ in (3) are not deictically anchored. In the case of ‑dadi, the motion is oriented away from the position of the referent of the A argument of the verb, regardless of the position of the speaker. In the case of ‑kena, the motion does not imply any particular orientation.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 135

The first parameter (temporal relation) is crucial for distinguishing AM sys-tems from other types of better known and seemingly similar systems of spatial expression such as, for example, ‘spatial preverbs’ of European languages (e.g., Rousseau 1995), or ‘directionals’ in languages from other parts of the words, such as Africa (e.g., Creissels et al. 2007: 148), Australia (e.g., Dixon 2002: 419–420), Austronesia (e.g., Senft 1997, 2004), Mesoamerica (e.g., Haviland 1991, 1993; Craig 1994), North-America (e.g., Mithun 1999: 139ff.) and Papua-New Guinea (Foley 1986: 148–152). To my knowledge, these systems consist of markers which primarily operate on motion verbs and (unless they have developed non-spatial meanings) specify the path of a motion already encoded by the verb. Thus, ‘spatial preverbs’ and ‘directionals’ do not contribute a motion component which could have a temporal realization distinct from that of action (see Guillaume 2006 for a discussion of major differences between the two types of systems).5

3. Associated motion systems in Tacanan languages: A comparison

Having clarified what is meant by AM and having provided a typological frame-work for the analysis of AM systems in general, I will now proceed with the presentation of what is known to date of the AM systems of the five Tacanan lan-guages. It is important to keep in mind that most Tacanan languages are relatively poorly studied and that the information available on their AM systems is far from being exhaustive.

3.1 Kavinik branch

To date, Cavineña is the best studied (and documented) language of the Tacanan family. In particular, its system of AM markers has received quite a lot of atten-tion, starting with the identification and description of some of the markers by SIL missionaries Elisabeth Camp and Millicent Liccardi in the 1970s and 1980s (see Camp 1982 and Camp & Liccardi 1989: 316–318), and followed by my own work on the system, presented at a number of conferences (Guillaume 1999, 2000a, b, 2009c, 2010) and published in several journal and book articles (see

5. This characterization of the two categories, AM and directionals, does not rule out the pos-sibility of the existence in some languages of polysemous markers that can function both as AM (with non-motion verbs) and directionals (with motion verbs). In South America this is the case for example with the suffix ‑mu in Quechua (Kerke & Muysken 1990). See other examples in Guillaume (2011).

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

136 Antoine Guillaume

Guillaume 2000c, 2006, 2008: 212–236, 2009b, 2013). My current understanding is that Cavineña has 12 AM markers.6 They are presented in Table 2.�

3.2 Chamik branch

Table 3 presents the AM system of the Portachuelo dialect of Ese Ejja, following the most recent analysis by Marine Vuillermet, who has worked under my direct supervision, and has dedicated a fairly important part of her study of the language on the expression of space. As she states in Vuillermet (2012, 2013), the language has 11 AM markers. The exact meaning of these markers requires further in-vestigation. It seems that this dialect of Ese Ejja basically displays all the major distinctions found in Cavineña, although, apparently, the system is less sensitive to deixis.

6. This figure actually includes a form, ‑(ne)ni, that I did not classify as an AM marker in my previous work, but as a distributed aspectual marker (Guillaume 2008: 206–209). The fact that (among other meanings) this marker can express spatial distribution of actions across space, implying that the action(s) occur in the background of a motion event, suggests that it is func-tionally very similar to the other AM markers of the system.

�. Note that the three motion suffixes ‑diru, ‑na and ‑eti are polysemous and can be used for expressing both prior and concurrent motion. In the latter function, they express the fact that the verb action occurs once in the middle of a motion path.

Table 2. Cavineña system of associated motion (Guillaume 2006, 2008, 2009b)

Deixis Motion of S/A Motion of O

prior concurrent subsequent

andative ‑tigo.temp&do ‑natido.pfv.going ‑dadigo(O)* ‑dirugo.perm&do ~ do.pfv.going.perm

‑ajedo.ipfv.going

venitive ‑nacome.temp&do ~ do.pfv.coming.temp ‑tsacome(O) ‑eticome.perm&do ~ do.pfv.coming.perm

‑bedo.ipfv.coming.temp‑etibedo.ipfv.coming.perm

unspecified ‑(ne)ni do.here&there ‑kenado&go

* Strictly speaking, the suffixes that express the motion of the O argument in all the Tacanan languages that have them, although oriented, are not deictically anchored. As mentioned above, the motion is oriented away from or toward the position of the referent of the A argument of the verb, regardless of the position of the speaker.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 13�

3.3 Takanik branch

Table 4 tentatively gives the inventory of AM markers of Araona. The informa-tion available on these morphemes in this language is very scanty. There are two grammatical descriptions on the language, Pitman (1980) and Emkow (2006), but none of them is very clear with respect to the number of morphemes that could be described as AM and on their exact meanings. The table lists 7 suffixes which I classify as AM markers based on my own interpretation of the data found in the two studies cited above. If we compare the forms with those of Cavineña and Ese Ejja, no markers seem to be expressing subsequent motion and motion of the object argument.

Table 5 tentatively lists the AM markers of Tacana, the language that gave its name to the whole family and to the sub-branch in which it is classified within the family. The forms and the analysis are based on fieldwork data that I collect-ed through approximately two and a half months of fieldwork between 2009 and 2012. The data gathered reveals that Tacana has possibly 13 distinct AM markers, which would make it the language with the highest number of AM distinctions in the Tacanan family, and to my present knowledge in any family of South American

Table 3. Ese Ejja system of associated motion, Portachuelo dialect (Vuillermet 2012, 2013)

Deixis Motion of S/A Motion of O

prior concurrent subsequent

andative ‑kigo&do ‑jyago(O)

venitive ‑wa come&do (1) ‑xekicome(O)‑nyaki come&do (2)

unspecified ‑nya arrive&do ‑pokido.going ‑nado&leave‑jebedo.returning ‑nana do&return‑’aekido.here&there

Table 4. Araona system of associated motion (Pitman 1980; Emkow 2006)

Deixis Motion of S/A Motion of O

prior concurrent subsequent

andative ‑tigo&do e-…-yoado.ipfv.going (1)‑jajoarrive.there&do e-…-bodo.ipfv.going (2)

venitive ‑shao/‑jao come&do‑(ja)inya arrive.here&do

unspecified -shana do.pfv.going

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

138 Antoine Guillaume

languages (see Section 4). The Tacana AM system illustrates again most of the rel-evant distinctions, apart from subsequent motion.

Finally, Table 6 provides the inventory of the AM markers of Reyesano. This language is moribund, with probably less then 12 ‘good’ speakers, all above 60 years old, who no longer use the language. The only data available stems from a documentation project that I conducted between 2004 and 2008. In the corpus collected, I could find only 2 AM markers. It is quite likely that the language used to have a more elaborate AM system in the past which cannot be elicited any-more, due to language attrition.

4. Associated motion systems in neighboring languages: Areal perspective

Before moving on to the reconstruction, I will briefly provide some evidence for my earlier claim that the AM category is prevalent in many languages of the re-gion where Tacanan languages are spoken.

Table 7 summarizes the results of a survey that I recently conducted on AM systems in 35 languages that surround the Tacanan languages, an area roughly corresponding to the northern and eastern lowlands of Bolivia, the adjacent south-eastern lowlands of Peru and the adjacent highlands of Bolivia and Southern Peru. These 35 languages belong to 15 distinct genetic groupings and isolates. I com-pared these systems using the same analytical framework, looking for grammatical

Table 5. Tacana system of associated motion (Guillaume field notes 2009–2012)

Deixis Motion of S/A Motion of O

prior concurrent subsequent

andative ‑tigo&do ‑nado.ipfv.going (1) ‑usego(O)‑seu arrive.there&do e-…-udo.ipfv.going (2)

e-…-buyudo.ipfv.going (3)

venitive ‑jeu come&do ‑bedo.ipfv.coming (1) ‑tsu come(O)‑senaarrive.here&do e-…-siudo.ipfv.coming (2)

unspecified ‑wanana do.pfv.going‑niu do.here&there

Table 6. Reyesano system of associated motion (Guillaume 2012, field notes 2004–2008)

Deixis Motion of S/A Motion of O

prior concurrent subsequent

andative ‑tigo&do e-…-bu do.ipfv.going

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 139

Table 7. Survey of grammatical markers with associated motion meanings in 35 South American languages of Bolivia and Peru (Guillaume 2011, slightly revised, with new data and analysis for Tacana)

Family Language Motion of S/A Motion of O Total

prior concurrent subsequent

Arawak Nanti 4 (1) 1 – (6)Machiguenga 4 1 1 1 7Nomatsiguenga 3 ? 1 1 5Caquinte 3 ? 1 1 5Ashéninca 1 ? 1 1 3Asháninca 4 ? 1 1 6Apurinã – (2) – – (2)Baure 2 – 1 – 3Mojeño 1 1 1 – 3Paresi 1 – – – 1Terêna 1 ? ? ? 1

Aymaran Aymara 2 2 – – 4Cayuvava (isolate) – – – – –Chiquitano (isolate) – – – – –Harakmbut (isolate) (1) (3) ? ? (4)Itonama (isolate) – 1 – – 1Leko (isolate) 2 – – – 2Mosetén (isolate) 2 4 – – 6Movima (isolate) – – – – –Panoan Amahuaca 1 2 – – 3

Kashinawa 1 4 – – 5Yaminahua 3 4 – – 7Chácobo 1 2 – – 3

Quechua Quechua 2 – – – 2Tacanan Araona 4 3 – – 7

Reyesano 1 1 – – 2Tacana 4 7 ? 2 13Cavineña 4 5 1 2 12Ese Ejja 4 3 2 2 11

Tupi-Guarani Chiriguano – – – – –Siriono – – – – –Yuki – – – – –

Uru-Chipaya Chipaya 1 – – – 1Uchumataqu – 1 – – 1

Yurakaré (isolate) – – – – –

15 families 35 languages 25 18 10 8 28

Key: – not attested; ( ) unsure figure; ? insufficient data / information unavailable. languages with 3 AM markers or more;

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

140 Antoine Guillaume

markers expressing prior motion, concurrent motion and subsequent motion, and markers that specifically encode the motion of the object argument.

80% of the languages investigated (28 out of 35 languages) have grammati-cal markers in the form of verbal affixes for expressing some associated motion meanings. Within these languages, 68% have complex AM systems with at least 3 AM markers (19 out of 28 languages). These languages correspond to 6 distinct language families (Arawak, Aymaran, Harakmbut, Mosetén, Panoan and Tacan-an). Finally, 33% have very complex systems with 6 or more markers (9 out of 28 languages). These correspond to 4 different language families (Arawak, Mosetén, Panoan and Tacanan).

If I am correct in stating that complex and well-grammaticalized systems of AM are extremely rare cross-linguistically and if such systems are found in so many languages in this part of South America across so many unrelated genetic groupings, therefore diffusion must have played a major role, a fact that should be taken into account when reconstructing AM systems in languages from that area. It is worth mentioning that the category of AM is said to have diffused extensively among the aboriginal languages spoken in and around Central Australia (Austin 1989; Wilkins 1991; Dixon 2002: 201–202; Koch 2007). Koch (2011) even argues that this category has even been replicated in Central Australian Aboriginal Eng-lish through the use of English motion verbs and adverbs.

5. Reconstructing the history of AM systems in Tacanan languages

The first part of this section tries to reconstruct four AM markers in the proto-language (§5.1). The second part looks at the forms that cannot be reconstructed and argues that at least some of them have originated from independent verbs of motion (§5.2).

5.1 Forms possibly reconstructible to Proto-Tacanan or to an intermediate level of the family tree

In this section I will argue that 4 AM forms can be reconstructed for Proto- Tacanan (or an intermediate level of the family tree), namely the suffixes *-ti, *-na and *‑be, and the circumfix *e-…-bu. None of these forms were reconstructed in the previous comparative works mentioned in the introduction (Key 1968; Girard 1971). The cognate sets that I will use as diagnostics for these reconstructed forms are listed in Table 8. As we will see, only *‑ti is unambiguously reconstructible. For

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 141

the remaining 3 forms, there are plausible arguments in favor of their reconstruc-tion, but the formal and semantic matches within the sets still present enough irregularities to be able to state with full certainty that we are dealing with real cognates. In addition, it is worth remembering that detailed and reliable informa-tion on the AM systems of most Tacanan languages is still lacking, which means that the present reconstruction cannot pretend to be more than a first attempt to reconstruct the forms.

I will now discuss each reconstructed form in turn. The clearest case is *‑ti. As we can see in the 1st line of Table 8, it is present in all the languages, with forms and meanings that match perfectly. Note that the change from Proto-Tacanan *t to k in the Portachuelo variant of Ese Ejja, which gives the form ‑ki, is regular (Girard 1971: 32).8 Therefore, there is little doubt that Proto-Tacanan had a suffix *‑ti and that this suffix probably expressed the semantics of prior andative mo-tion, as is still the case in all the Tacanan languages today.

The second postulated proto-form, *‑na, based on the 2nd line, is found with the same segmental make-up in 4 Tacanan languages and with AM meanings in 3 of these languages, each pertaining to a different sub-branch. These facts sug-gest that it can be reconstructed to Proto-Tacanan, although the meaning of the reconstructed *‑na is somehow problematic, considering that the meanings of the reflexes in the different languages show striking opposite values, as summarized in Table 9.

8. The Portachuelo dialect falls into Girard’s “Chama complex”.

Table 8. Possible cognate sets related to the AM category in the different Tacanan languages

Kavinik Chamik Takanik

Cavineña Ese ejja (Portachuelo) Araona Tacana Reyesano

-tigo.temp&do

-ki go&do

-ti go&do

-ti go&do

-ti go&do

‑nacome.temp&dodo.pfv.coming.tempincho

‑na do&leave

-(a)naincept

‑nado.ipfv.going

-na incho

‑bedo.ipfv.coming.temp

‑jebedo.returning

‑bedo.ipfv.coming

‑bebe fast

e‑…‑u potential

e-…-bo do.ipfv.going

e-…-udo.ipfv.going

e-…-bu do.ipfv.going

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

142 Antoine Guillaume

If we take into account the fact that Araona, Cavineña and Reyesano also have cognate suffixes ‑na (or ‑(a)nain the case of Araona) with inchoative/incep-tive meanings (Table 8),9 we might suggest that the original meaning of Proto-Tacanan *‑na was prior motion (rather than concurrent or subsequent motion). This is supported by the fact that the development of motion morphemes (verbs or affixes) in prior motion constructions into change-of-state kinds of aspectual markers is well attested cross-linguistically (Heine & Kuteva 2002: 74–75, 156–157; Voisin 2010), as with English ‘come’ or ‘go’ in expressions such as ‘come true’ and ‘go mad’ respectively.

The third postulated proto-form, *‑be, based on the 3rd line, is also plausibly reconstructable since an AM marker ‑be is found with the same (concurrent mo-tion) meanings, ‘do.ipfv.coming’, in two languages from two distinct branches, Cavineña (Kavinik) and Tacana (Takanik). If this hypothesis is correct, it is likely that the AM suffixes ‑jebe of Ese Ejja and ‑bebe of Reyesano contain the same marker. In Ese Ejja, *‑be would have fused with a preceding morpheme ‑je whose origin is unknown, but which could have been a verbal suffix meaning ‘back’, so as to yield a concurrent AM form ‘do.returning’. In Reyesano, *‑be would have been reduplicated, and develop into a manner marker meaning ‘fast’, which is not incompatible with meanings generally expressed by reduplication.10

The fourth postulated proto-form, the circumfix *e-…-bu, based on the 4th line, is at least clearly reconstructable for the Proto-Takanik level of the family tree, as it has unambiguous reflexes with essentially the same discontinuous/cir-cumfix forms and (apparently) the same meanings in all three languages from that branch, Araona, Tacana and Reyesano. It happens that the consonant b is

9. The inchoative/inceptive suffixes ‑na in these three different languages do not have the same distribution. In Cavineña, the inchoative ‑na can only attach to an adjectival root and turn it into a verbal stem. In the other two languages the inchoative/inceptive ‑na (Reyesano) or ‑(a)na (Araona) only attach to a verbal root/stem and does not change its word class.

10. The suffix ‑bebe in Reyesano is to be taken with a lot of caution since I only have two ex-amples of it in the whole corpus, once with the verb pue ‘come’ and once with the verb puayu ‘come back’.

Table 9. Semantics of *-na reflexes in different Tacanan languages

Kavinik Chamik Takanik

Cavineña Ese Ejja Tacana

temporal relation prior or concurrent subsequent concurrentaspectual realization perfective perfective imperfectivedeixis venitive unspecified andative

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 143

lacking in the Tacana reflex e‑…‑u, but the loss of *b before *u is attested in at least one other Tacana suffix, ‑ute ‘downwards’ (corresponding to Reyesano ‑bute ‘downward’ and Araona ‑bote ‘downwards’; Emkow 2006: 530). Whether we can reconstruct *e-…-bu to a higher level of the family tree depends on the plausibil-ity of recognizing the circumfix e‑…‑u of Cavineña as a cognate, considering that it lacks the consonant b (as in Tacana) and that its meaning (potential modal-ity) is quite distinct from the AM meanings in the other languages. Regarding the loss of b, unlike in Tacana, I do not have other examples in which this has occurred in Cavineña. Regarding the meaning of the Cavineña marker e‑…‑u (potential modality), it could well have originated in a motion marker, consider-ing that e‑…‑u is actually only used in the future (Guillaume 2008: 178–179), and we know that motion verbs / markers are a very common source for expressing future meanings (Heine & Kuteva 2002: 161–163, 308–309), as happened with English ‘be going to’.

A case could be made for reconstructing a fifth AM marker, *‑diru, based on the form-meaning resemblances between the Cavineña AM suffix ‑diruand the aspectual markers ‑’yo of Ese Ejja and ‑yuof Reyesano and Tacana, shown in Table 10.

Formally, postulating a cognate relationship between these forms could make sense if we consider the fact that the proto-phoneme *r, which has a reflex in pres-ent-day Cavineña ‑diru, is regularly lost in all the non-Kavinik languages (Girard 1971: 36). The application of this phonological change to the putative proto-form *‑diru would yield the intermediate form *‑diu, which could have turned into ‑’yo in Ese Ejja or ‑yu in Reyesano and Tacana as follows:

– In Ese Ejja, the proto-phoneme *d regularly turns into a glottal stop when occurring between vowels (Girard 1971: 36). If we add that the o phoneme in this language is cognate to the uof the other languages – there is no uvs.o phonological distinction in any of the Tacanan languages – we obtain the form ‑’io, which is a very small step to ‑’yo, in particular knowing that diachronically i is normally palatalized before a back vowel in this languages (this is at least clearly attested before the vowel a; Vuillermet 2012: 181ff.);

Table 10. Form-meaning resemblances between Cavineña AM -diru and aspectual markers -’yo of Ese Ejja and -yu in Reyesano and Tacana

Kavinik Chamik Takanik

Cavineña Ese ejja (Portachuelo) Araona Tacana Reyesano

‑dirugo.perm&do

-’yo telic

? -yu ipfv

-yu ipfv

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

144 Antoine Guillaume

– In Tacana and Reyesano, I do not have examples of morphemes where a sequence yu would have resulted from an earlier sequence diu. However, a change from diu to yu is not inconceivable, considering that (1) similarly to Ese Ejja, i is regularly palatalized preceding u in these two languages (see Guillaume 2012 for Reyesano; Wynen & Wynen 1962 and Guillaume field notes 2009–2012 for Tacana) and that (2) d and y are phonetically rather simi-lar, sharing the same place of articulation and voicing specification.

Turning to the aspectual meanings of ‑’yo or ‑yuin these three languages – telicity in Ese Ejja and reiterativity in Tacana and Reyesano – it is not unthinkable that they represent advanced stages of grammaticalization of an AM marker *‑diru if this marker had meanings similar to those expressed by present day Cavineña ‑diru, namely ‘permanent motion’. Indeed this marker can express, among other meanings, a motion returning to home base, which involves the idea of reitera-tion. Nevertheless, there is a major argument against the reconstructibility of *‑diru in proto-Tacanan, which is that Cavineña has a formally and semantically corresponding independent verb of motion diru ‘go permanently’ which cannot be reconstructed for Proto-Tacanan. This fact strongly suggests that the Cavineña AM suffix ‑diru corresponds to a recent development out of this verb which is specific to Cavineña, rather than the remnant of a Proto-Tacanan suffix *‑diru. One might want to use the same argument to reject the reconstruction of *‑beand *‑bu (as part of *e-…-bu), considering that Cavineña also has formally and semantically corresponding verbs be ‘bring O’ and bute‘go down’, respectively. In theses cases, however, the argument is less cogent, for the reason that these two verbs can be reconstructed in Proto-Tacanan (see following section). This allows for the possibility that Proto-Tacanan had both the suffixes *‑beand *‑bu(as part of *e-…-bu) and the (historically related) independent verbs *be ‘bring O’ and *bute ‘go down’.

5.1 Forms not reconstructable to Proto-Tacanan or to an intermediate level of the family tree

The preceding section has made clear that we cannot reconstruct an AM system in Proto-Tacanan as elaborate as those found in its present-day daughter lan-guages. This brings up the question of the origin of the AM forms that are not reconstructible, which must have developed in more recent times. In this section, I argue that at least some of these forms (as well as some of the forms that are pos-sibly reconstructible) come from independent verbs for motion.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 145

In Table 11, I have listed all the possible correspondences (perfect or imper-fect) that I could find between AM markers in the different Tacanan languages and formally and/or semantically related independent verbs in the same (and/or

Table 11. Correspondences between AM suffixes, independent verbs and their reconstructed form in Proto-Tacanan

AM suffixes Independent verbs Reconstructed verbs*

‑diru ‘go.perm&do’ (C)** diru ‘go perm.’ (C) –‑aje ‘do.ipfv.going’ (C) aje ‘walk’ (C, A) *ase

ase ‘walk’ (T)adhe ‘walk’ (R)

‑be ‘do.ipfv.coming.temp’ (C) be ‘bring O’ (C, R) *be‑jebe ‘do.returning’ (E) beitu ‘send O’ (T)‑be ‘do.ipfv.coming’ (T) beicho ‘send O’ (A)‑dadi ‘go(o)’ (C) dadi ‘find O’ (C) –‑jeu ‘come&do’ (T) jeu ‘come and get O (T) –

jeo come and get O (A)‑tsu ‘come(o)’ (T)** tsu ‘meet, encounter’ (T) *tsuru

tso ‘meet, encounter’ (A)tsunetia ‘meet, encounter’ (R)tsuru ‘meet, encounter’ (C)

‑xeki ‘come(o)’ (E) xeki ‘go and get O’ (E) *ketijeti ‘go and get O’ (A, T, R)keti ‘go and get O’ (C)

‑jya ‘go(o)’ (E) jya ‘abandon/throw O’ (E) *se‑eti*** ‘come.perm&do’ (C) sia ‘throw O’ (T)

dhapua ‘throw O’ (R)je ‘come temp.’ (C)jeti ‘come perm.’ (C)

‑poki ‘do.going’ (E) poki ‘go’ (E) *putipoti ‘go’ (A)puti ‘go’ (T, R)

e‑…‑bu ‘potential’ (C) bute ‘go down’ (C) *butee‑…‑bu ‘do.ipfv.going’ (R) buteti ‘go down’ (R)e‑…‑u ‘do.ipfv.going’ (T) bute ‘go down’ (T)e‑…‑bo ‘do.ipfv.going’ (A) bote ‘go down’ (A)

* All the verbal forms were reconstructed by Girard and are confirmed by my own reconstruction.

** Language codes: A = Araona, C = Cavineña, E = Ese Ejja, R = Reyesano, T = Tacana.

*** The Cavineña suffix ‑etibe ‘do.ipfv.coming.perm’ seems to have developed out of the combination of the two suffixes -eti and ‑be.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

146 Antoine Guillaume

in other Tacanan) languages, and I have included the forms of these verbs when they can be reconstructed in Proto-Tacanan.

As we can see, there is a reasonable number of convincing correspondences between AM markers and independent verbs of motion synchronically attested in Tacanan languages, whether the verbs are reconstructible or not in Proto-Tacanan. It is therefore quite likely that at least a certain number of the AM markers that are not reconstructible for Proto-Tacanan or an intermediate level of the Tacanan family tree correspond to independent verbs of motion that have recently become affixes. This raises the question as to how these independent verbal forms hap-pened to become verb suffixes in these particular languages. The most obvious scenario would be to postulate a process of auxiliation and morphologization, which is a cross-linguistically well-attested path of grammaticalization, whereby a biclausal construction becomes monoclausal and finally monoverbal. Such a path has been proposed for the genesis of complex AM systems in three Central Australian languages, Adnyamathanha (Tunbridge 1988), Mparntwe Arrernte (Wilkins 1991) and Kaytetye (Koch 2007), and in the Mesoamerican language Olutec, from the Mixe-Zoquean family (Zavala 2000). In all these languages, (at least some of) the AM forms are internally complex, consisting of two parts. The first part can be traced back to a subordinating marker while the second to a mo-tion verb. In other words, these AM constructions result from the conflation of a subordinate clause and a following main clause containing a verb of motion. In the AM forms of Tacanan languages, however, there are no traces of subordinat-ing morphemes, so that it seems quite unlikely that the grammaticalization path proposed for the Australian and Mesoamerican languages took place in the pres-ent-day Tacanan languages. One could think of other constructions, such as verb serialization or light verb constructions, which would involve motion verbs but no subordinating markers, but there is no evidence that these have occurred in pres-ent-day Tacanan languages.11 Therefore it is tempting to suggest that the recently created AM suffixes stem from the spontaneous adhoc incorporation / copying of particular verbal roots into already well established morphological paradigms of AM markers. Although such a scenario is not among the commonly recognized and abundantly attested processes of language change, it has been proposed by Wilkins (1991: 245ff.) for several AM marker in Mparntwe Arrernte, in which he was unable to find any traces of a subordinating morpheme. Among the possible motivations for a process of ad hoc copying of verb roots, one could suggest the need to renew particular semantic contrasts that would have been lost through time, and/or the pressure to create new (or renew disappearing) contrasts so as to

11. See Guillaume (2013) for a discussion of these possible paths of morphologization in the case of Cavineña.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 14�

reinforce similarity with a neighboring language (from the same or from a differ-ent family). Several arguments plead in favor of ad hoc copying of verb roots in the case of Tacanan languages:

1. Since we can reconstruct several AM markers in the proto-language, we know that the grammatical category of AM is an old feature of Tacanan languages;

2. As argued above, it is very likely that diffusion of AM systems has taken place between different languages and languages families in the region where Ta-canan languages are spoken;

3. Tacanan languages are polysynthetic with a very ‘loose’ agglutinative mor-phological structure, in the sense that they have very transparent morpheme boundaries and very little allomorphic variation. This characteristic would allow for easy incorporation of ‘foreign’ material;

4. Not only AM markers have corresponding independent lexical items (mostly verbs) but many other verb suffixes as well, having to do with aspect, man-ner, posture, time of the day, etc. (see full list of such correspondences in Cavineña in Guillaume 2008: 252). Crucially, as far as I can say, there is again no evidence that these suffixes are the result of a gradual grammaticaliza-tion process of the corresponding verbs out of their use in analytic predicate constructions.

It remains quite possible that a grammaticalization process of auxiliation and morphologization did take place in the genesis of the AM systems of Tacanan lan-guages. However, if this happened, it must have taken place in pre-Proto-Tacan-an. Alternatively, such a process could have generated the grammatical category of AM in a neighboring non-Tacanan language, and the AM category been repli-cated later in pre-Proto-Tacanan or Proto-Tacanan (and in other languages).

6. Conclusions

In this paper, I have gathered and compared for the first time the information available on the AM systems of the five Tacanan languages of the Amazonian lowlands of Bolivia and Peru. AM systems consist of grammatical markers that attach to non-motion verbs and indicate that the verb action happens against the background of a motion event with a specific orientation in space.

This study has first shown that all but one of the Tacanan languages (Reye-sano) have very complex AM systems, ranging from 7 markers (Araona) to 13 markers (Tacana). I have made the comment that such types of systems are very rare cross-linguistically although not regionally, since they are attested in a fair

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

148 Antoine Guillaume

number of neighboring languages that belong to several distinct linguistic fami-lies. Secondly, the comparison of the different Tacanan systems has revealed that at least four AM markers can arguably be reconstructed in Proto-Tacanan, which can be added to the list of morphemes reconstructed by Key (1968) and Girard (1971). Thirdly, I have argued that for some of the markers that cannot be recon-structed, there is clear indication that they originated in independent verbs of motion. In addition I have put forward the hypothesis that they could have origi-nated in a phenomenon of adhoc copying of these verbs, rather than in a more standard in a phenomenon of adhoc copying of these verbs, rather than in a more standard process of gradual grammaticalization/auxiliation. This phenomenon could have been motivated by language contact.

The history of the AM systems in the different Tacanan languages seems therefore to be fairly heterogeneous, with at least some markers inherited from the proto-language and others developed more recently from independent verbs of motion. I have not yet investigated the possibility that some markers could have been borrowed from other non-Tacanan languages but this is an option that needs to be considered in future work.

It is important to recall that the results of this study are, in large part, tenta-tive. This is so because the information available on the AM systems of certain Tacanan languages, in particular Araona, Reyesano and Tacana, is very scanty. It is also so because to date there are very few detailed synchronic, and basically no diachronic, studies on AM systems in South America, in spite of the fact that they are very common in that part of the world.

References

Austin, Peter. 1989. “Verb Compounding in Central Australian Languages”. LaTrobeUniversityWorkingPapersinLinguistics 2.1–31.

Camp, Elizabeth L. 1982. “Referentes de movimiento y ubicación en el discurso narrativo en cavineña”. RevistaLatinoamericanadeEstudiosEtnolingüísticos 2.81–122.

Camp, Elizabeth L. & Millicent R. Liccardi. 1989. Diccionario Cavineña‑Castellano Castel‑lano‑Cavineña con Bosquejo de la Gramática Cavineña. Dallas: Summer Institute of Linguistics.

Craig, Colette. 1994. “Jakaltek directionals: their meaning and discourse function”. LanguagesoftheWorld 7.23–36.

Creissels, Denis, Gerrit J. Dimmendaal, Zygmunt Frajzyngier & Christa König. 2007. “Africa as a Morphosyntactic Area”. ALinguisticGeographyofAfrica ed. by Bernd Heine & Derek Nurse, 86–150. Cambridge: Cambridge University Press.

Crevels, Mily & Pieter Muysken. 2009. “Lenguas de Bolivia: presentación y antecedentes”. Len‑guasdeBolivia,vol.1,Ambitoandino ed. by Mily Crevels & Pieter Muysken, 13–26. La Paz: Plural Editores.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 149

Dixon, R. M. W. 2002. Australian Languages: Their Nature and Development. Cambridge: Cambridge University Press.

Emkow, Carola. 2006. AGrammarofAraona. Ph.D. dissertation, La Trobe University.Foley, William A. 1986. ThePapuanLanguagesofNewGuinea. Cambrigde: Cambridge Uni-

versity Press.Girard, Victor. 1971. Proto‑TakananPhonology. (= UniversityofCaliforniaPublicationsinLin‑

guistics, 70.) Berkeley & Los Angeles: University of California Press.Guillaume, Antoine. 1999. “Directionals Versus Associated Motions in Cavineña”. Paper pre-

sented at the 26th conference of the Linguistic Association of Canada and the United States (LACUS XXVI), Edmonton, August 3–7, 1999.

Guillaume, Antoine. 2000a. “Affixes of Motion and Direction in Cavineña”. Paper presented at the annual conference of the Society for the Study of the Indigenous Languages of the Americas (SSILA), Chicago, January 7–8, 2000.

Guillaume, Antoine. 2000b. “Directionals, Motion-with-Purpose and Associated Motion in Cavineña”. Paper presented at the Seminar of Amerindian Linguistics, University of Leiden, February 8, 2000.

Guillaume, Antoine. 2000c. “Directionals Versus Associated Motions in Cavineña”. The lex‑icon: Proceedings of the 26th conference of the Linguistic Association of Canada and theUnitedStates(LACUSXXVI),Edmonton,3–7August1999 ed. by Alan K. Melby & Arle R. Lommel, 395–401. Fullerton: Linguistic Association of Canada and the United States.

Guillaume, Antoine. 2006. “La catégorie du ‘mouvement associé’ en cavineña: apport à une typologie de l’encodage du mouvement et de la trajectoire”. BulletindelaSociétédeLinguis‑tiquedeParis 101:1.415–436.

Guillaume, Antoine. 2008. AGrammarofCavineña. (= MoutonGrammarLibrary,44.) Berlin & New York: Mouton de Gruyter.

Guillaume, Antoine. 2009a. “Hierarchical Agreement and Split Intransitivity in Reyesano”. In‑ternationalJournalofAmericanLinguistics 75:1.29–48.

Guillaume, Antoine. 2009b. “Les suffixes verbaux de mouvement associé en cavineña”. FaitsdeLangues:LesCahiers 1.181–204.

Guillaume, Antoine. 2009c. “Cavineña ‘Associated Motion’ Suffixes: Their Meanings and Dis-course Function”. Paper presented at the Transalpine Typology Meeting, Bern, January 22–24, 2009.

Guillaume, Antoine. 2010. “Los morfemas verbales de movimiento asociado en cavineña: ¿Ver-bos compuestos, verbos seriales, u otros?”. Paper presented at the Simposium “Expresión de nociones espaciales” of the International Conference Amazonicas III: Phonology and Syntax, Bogota, April 19–24, 2010.

Guillaume, Antoine. 2011. “Grammaticalization of Motion and Language Contact: The Cat-egory of ‘Associated Motion’ in Southwestern Amazonian Languages”. Paper presented at the 4th International Conference of the Association Française de Linguistique Cognitive (AFLiCo IV), Lyon, May 24–27, 2011.

Guillaume, Antoine. 2012. “Maropa (reyesano)”. LenguasdeBolivia,vol.2,Amazonía ed. by Mily Crevels & Pieter Muysken, 191–229. La Paz: Plural Editores.

Guillaume, Antoine. 2013. “Algunas reflexiones sobre las construcciones de verbos seriales ‘contiguas e incorporantes’ y el sistema de movimiento asociado en cavineña”. Expresióndenociones espaciales en lenguasamazónicas ed. by Ana María Ospina, 17–37. Bogota: Universidad Nacional de Colombia & Instituto Caro y Cuervo.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

150 Antoine Guillaume

Haviland, John B. 1991. “The Grammaticalization of Motion (and Time) in Tzotzil”. WorkingPaper, No.2,CognitiveAnthropologicalResearchGroup. Nijmegen: Max Planck Institute For Psycholinguistics.

Haviland, John B. 1993. “The Syntax of Tzotzil Auxiliaries and Directionals: The Grammaticali-zation of ‘Motion’”. ProceedingsoftheNineteenthAnnualMeetingoftheBerkeleyLinguisticsSociety:SpecialSessiononSyntacticIssuesinNativeAmericanLanguages, 35–49.

Heine, Bernd & Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge: Cambridge University Press.

Kerke, Simon van de & Pieter Muysken. 1990. “Quechua mu and the Perspective of the Speak-er”. UnityinDiversity:PapersPresentedtoSimonC.DikonHis50thBirthday ed. by Harm Pinkster & Inge Genee, 151–163. Dordrecht: Foris.

Key, Mary Ritchie. 1968. ComparativeTacananPhonologywithCavineñaPhonologyandNotesonPano‑Tacanan. The Hague: Mouton.

Koch, Harold. 1984. “The Category of ‘Associated Motion’ in Kaytej”. LanguageinCentralAus‑tralia 1.23–34.

Koch, Harold. 2006. “Kaytetye”. The Encyclopedia of Language and Linguistics, Vol. 6 ed. by Keith Brown, 170–172. Oxford: Elsevier.

Koch, Harold. 2007. “Language Contact and Tthe Grammaticalisation of Motion: A Case Study from Central Australia”. Paper presented at the 18th International Conference of Historical Linguistics (ICHL XXVIII), Montreal, August 6–11, 2007.

Koch, Harold. 2011. “The Influence of Arandic Languages on Central Australian Aboriginal Eng-lish”. Creoles,TheirSubstrates,andLanguageTypology(=TypologicalStudiesinLanguage,95) ed. by Claire Lefebvre, 437–460. Amsterdam and Philadelphia: John Benjamins.

Mithun, Marianne. 1999. TheLanguagesofNativeNorthAmerica. Cambridge: Cambridge Uni-versity Press.

Pitman, Donald. 1980. Bosquejode lagramáticaaraona. (= NotasLingüísticasdeBolivia,9.) Riberalta: Instituto Lingüístico de Verano.

Pitman, Donald & Mary Pitman. 1970. Lajerarquíafonológicadelaraona.(= NotasLingüísticasdeBolivia, 8.) Riberalta: Instituto Lingüístico de Verano.

Rousseau, André, ed. 1995. LesPréverbesdansleslanguesd’Europe.Introductionàl’étudedelapréverbation. Lille: Presses Universitaires du Septentrion.

Senft, Gunter, ed. 1997. Referring to Space: Studies in Austronesian and Papuan Languages. Oxford: Clarendon Press.

Senft, Gunter, ed. 2004. Deixis and Demonstratives in Oceanic Languages. Canberra: Pacific Linguistics.

Talmy, Leonard. 1985. “Lexicalization Patterns: Semantic Structure in Lexical Forms”. LanguageTypologyandSyntacticDescription,Vol.3ed. by Timothy Shopen, 57–148. Cambridge: Cambridge University Press.

Tunbridge, Dorothy. 1988. “Affixes of Motion and Direction in Adnyamathanha”. ComplexSen‑tencesConstructionsinAustralianLanguages (= TypologicalStudiesinLanguage, 15) ed. by Peter K. Austin, 267–283. Amsterdam/Philadelphia: John Benjamins.

Voisin, Sylvie. 2010. “Les morphèmes ‑i et -si en wolof ”. Sciences et Techniques du Langage, [Revue du Centre de Linguistique Appliquée de Dakar] 10.21–34.

Vuillermet, Marine. 2012. AGrammarofEseEjja,aTakananLanguageofAmazonianBolivia. Ph.D. dissertation. Université Lumière Lyon 2.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny

Reconstructing the category of “associated motion” in Tacanan languages 151

Vuillermet, Marine. 2013. “Dónde, cuando, y con quién ocurren acciones: El movimento asociado en Ese Ejja. Expresióndenocionesespacialesenlenguasamazónicas, ed. by Ana María Ospina, 39–59. Bogota: Universidad Nacional de Colombia e Instituto Caro y Cuervo.

Wilkins, David P. 1991. “The Semantics, Pragmatics and Diachronic Development of ‘Associ-ated Motion’ in Mparntwe Arrernte”. BuffaloPapersinLinguistics 1.207–257.

Wynen, Donald van & Mabel Van Wynen. 1962. Fonemas tacanaymodelosdeacentuación. (= NotaslingüísticasdeBolivia, 6.) Cochabamba: Instituto Lingüístico de Verano.

Zavala, Roberto. 2000. “Olutec Motion Verbs: Grammaticalization under Mayan Contact”. Pro‑ceedingsoftheTwenty‑sixthAnnualMeetingoftheBerkeleyLinguisticsSociety 139–151.

Unc

orre

cted

pro

ofs

- J

ohn

Ben

jam

ins

Publ

ishi

ng C

ompa

ny