Printer Friendly

Locative construction and positionals in Trumai.


In Trumai, three constructions can possibly be used as a reply to a where question: (i) a construction with a positional verb, such as 'sit', 'lie', 'stand'; (ii) a copular construction; (iii) a zero-copula construction. In the current study, I analyze the relationship among them, trying to determine what the Basic Locative Construction (BLC) of Trumai is. The main hypotheses are: (i) the zero-copula construction is just a subtype of the copular one; (ii) the system of Trumai is moving from one type of BLC (copular) to another type (positional verb), but the move is not completed yet, and this would be the reason why both constructions can function as replies to where questions. I analyze the relationship between these constructions and how they code spatial information. Another relevant issue explored in this article is the class of positional verbs that occur in the locative constructions. I describe the formal characteristics and semantics of such verbs, exploring the role that these verbs can have in classifying types of Figures.

1. General introduction

The aim of the present article is to analyze the Basic Locative Construction of Trumai, a Brazilian indigenous language. (1) First, I will present an introduction to the Trumai people and their language (Section 1.1), and to the theoretical issues concerning Basic Locative Constructions (Section 1.2). Next, the various locative constructions found in Trumai will be described in detail (Sections 2 and 3). Section 4 explores the class of verbs that occur in one of these locative constructions. In Section 5, the relationship between the different constructions is analyzed and discussed. Section 6 presents a general conclusion about the facts observed in this language.

1.1. The Trumai group

The Trumai people live in the Xingu reserve, a multi-ethnic indigenous site in the central region of Brazil. There are three Trumai villages in the area: Cristalina, Boa Esperanca, and Steinen. However, it is possible to find Trumai families living in other locations within the reserve and in a city in the vicinity of the Xingu area. In other words, the group is relatively dispersed, not living in one single site.

The community has more than 100 individuals; however, only 51 people still effectively speak the language. The other members of the group use Brazilian Portuguese and/or another language of the Xingu reserve--such as Kamayura, Suya, or Aweti--in their daily conversations. The individuals who speak Trumai also have knowledge of Portuguese, with different degrees of bilingualism. There are historical explanations for the current linguistic scenario found in the Trumai villages (cf. Guirardello 1999: preface).

The Trumai language is genetically isolated. It has four open classes: nouns, verbs, adjectives, and adverbs. Its case-marking system presents an Ergative-Absolutive alignment in morphology. There is also a strong Ergative-Absolutive alignment in syntax, although certain Nominative-Accusative patterns can be observed as well. The basic word order in transitive clauses is AOV, but variations of order are possible, depending on pragmatic factors. There are five verb classes in Trumai; the positional verbs that will be treated in this article belong to the class of Intransitive verbs.

1.2. Studies on Basic Locative Constructions (BLC)

Over several years the Language and Cognition research group of the Max Planck Institute for Psycholinguistics (Nijmegen, Netherlands) conducted a project investigating the nature of spatial parameters in natural languages. One of the topics explored in this project was the configuration of locative constructions in various languages of the world. The work was based on "the assumption, common in typological linguistics, that is possible to identify a Basic Locative Construction (BLC) in a diverse range of unrelated languages" (Kita and Walsh Dickey 1998: 56).

The Basic Locative Construction of a language would be the typical form of answer employed by its speakers when they reply to where questions. The focus of interest in studying this construction is to analyze the spatial information codified in it, and how such information is distributed across the various components of the construction (adpositions, verb, etc.). As stated in the "Annual report 2000" (Senft and Smits 2000: 73), "on the basis of European languages it has often been assumed that topological relators are usually adpositions (prepositions and postpositions). In wider crosslinguistic perspective, however, it is clear that this information may be encoded in relational nominals, adverbial nominals, locative predicates, case affixes, and constructional alternates, and in any one language it is often distributed across the clause".

The studies realized at the Max Planck Institute revealed that there are some variations across the languages under investigation with regard to the verbal element found in their BLCs. On the basis of these studies, a working typology has been proposed (Van Geenhoven and Warner 1999: 61):

Type 0: No verb in the BLC, as observed in Turkish;

Type I: Single locative verb

Type Ia: Copula (i.e., a dummy verb). This is the case of English;

Type Ib: Locative verb (a general locative verb). It is the scenario found in Japanese;

Type II: A wide set of dispositional verbs, as observed in Tzeltal. Dispositional verbs provide spatially rich information and are not semantically restricted to the posture domain. The sets are usually quite large--e.g., some Mayan languages have several dozen of these verbs, some have a much higher number (cf. the article by Bohnemeyer and Brown [this issue] about the Tzetal and Yukatek languages);

Type III: A small, contrastive set of posture or positional verbs. It is the case of Dutch, Arrernte, and other languages.

Taking into consideration the working typology presented by Van Geenhoven and Warner, I conducted research on the Trumai language, in order to identify the characteristics of its BLC.

As we will see in Sections 2 and 3 below, Trumai has three constructions that can possibly be used as a reply to a where question:

-- a copular construction

-- a zero-copula construction

-- a construction with positional verbs, such as 'lie', 'sit', 'stand', etc.

The zero-copula construction can be analyzed as a subtype of the copular one (basically, the copula is omitted under certain circumstances), but the positional construction is indeed of a different kind. There is no particular reason to state that one of the constructions is necessarily the BLC of Trumai, as I discuss in Section 5. Therefore, Trumai does not fit very well in the typology presented above, in which each language belongs to just one type. Within this framework, Trumai would belong to two different types: Ia and III.

In relation to the languages spoken in the Xingu reserve and neighboring areas, Trumai seems to be a special case. The languages of that region belong to either type 0 or Ia-Ib with regard to the kind of verb they employ in the locative construction. For instance:

-- Type 0 (no verb in the BLC): this is the case of Kuikuro (Franchetto 2006, p.c.) and Panara (Dourado 2001);

-- Type Ia (copula): as observed in Kamayura (Seki 2000) and Bakairi (Meira 2004, p.c.);

-- Type Ib (a general locative/existence verb): this is the case of the Aweti language (Drude p.c.).

To my knowledge, no other language in the Xingu region belongs to type III (positional verbs). The Bakairi language oscillates between employing the copula or not (in the present tense, the copula is often omitted), but no language oscillates between employing copula or a positional verb. Trumai is the only language in the Xingu area that presents this behavior. The scenario observed in Trumai evokes two relevant questions:

-- Why does a language present two different types of locative construction? What would motivate the existence of different simultaneous patterns?

-- If the speakers have two main locative constructions available (the positional-verb and the copular), how do they select one over the other?

Before we can address these questions, the Trumai locative constructions will be examined in detail.

2. BLC in Trumai: investigation

For the investigation of the BLC in Trumai, I elicited data from Trumai speakers using the "Picture Series for Positional Verbs"--designed by Ameka, De Witte and Wilkins--also exploring further distinctions suggested by the authors in the guidelines of the material (Ameka et al. 1999: 52-53). I used the "Topological Relations Picture Series" developed by Bowerman and Pederson (1993) as well. The work was conducted with three consultants, all native speakers of Trumai, who varied with regard to age (one was young, two were older) and degree of proficiency in Portuguese (two consultants spoke Portuguese very well). I also worked with data from texts and observation of natural uses of the language. In the elicited data, (2) the construction used by the consultants to answer a where question was:

Ground (Focus/Tense) Figure Positional

For example:

(1) hamuna in [katatyi]?

Where FOC bottle YI

'Where is the bottle?'

mesa natu-n ka_in [katat yi] aha'tsi.

table back/top-LOC FOC/TENS bottle YI sit

'The bottle sits on the table.'

The answers provided by the Trumai speakers exhibited an extensive use of positional verbs. The forms most frequently used were: la 'be standing', (3) aha'tsi 'be sitting', chumuchu 'be lying', and tsula 'be lying (not at the floor level)'. There were also a few instances of pila 'be in liquid medium' and mula 'be in a closed place'.

As we can see in the example above, the NP that indicates the Ground (4) comes in the first position of the clause, followed by the Focus/Tense particle ka_in (although this particle is not always present). This actually represents the typical way of answering information questions in Trumai: the requested (the new) information comes in the first position, followed by ka_in (or ch'i_in), a second-position particle that indicates focus with further information about tense. (5)

The NP-Ground receives -(V)n, which is the general locative marker of the language. Given that the locative case marker is so general in meaning, relational nouns can be employed (formally possessed by the noun that indicates the Ground) in order to further specify location:
(2) X natu-n 'on top of X'
X fax-on 'in X'
X real-an 'on the borders of X'
X xop-an 'on the borders of X'
X heni-n 'beside X, next to X'
X dacha-n 'behind X'
X wela-n 'in the middle of X'
X iuda-n 'behind X'
X (h)uk-an 'in front of X'
X apud-an 'under X'

Several of these relational nouns are clearly body part terms: natu 'back'; fax 'visceras, intestines'; mal 'lip', xop 'mouth'; heni 'side of the body'. In the case of iuda, huk, and apud, Trumai consultants are not able to explain the meaning of these words in isolation. Apparently, they only know the meaning of the whole combination, i.e., [word + Locative marker], which suggests that these combinations might be turning into postpositions.

The expression of 'behind X' in Trumai is a little peculiar: the word used in the expression is dacha 'spinal cord'. This is a curious fact because if an object is behind X, it is located not on the spinal cord of X, but rather in relation to it. Interestingly, some speakers employ the word dacha not with the Locative marker, but rather with the Ablative postposition (3b), maybe to better express the spatial information they want to convey:

(3) a. ha dach-an ka_in [pola yi] chumuchu.

1 spinal.cord-LOC FOC/TENS ball YI lie

'The ball lies behind me.'

b. ha dach lots' ka_in [kasoro yi] aha'tsi.

1 spinal.cord Ablat FOC/TENS dog YI sit

'The dog sits behind me.'

Besides the Locative case marker--(V)n, there are other elements in the language that can be used to indicate Ground: the Ablative postposition lots' (as in the example above), the Allative postposition ita (6) and the Dative markers--(V)s and -ki. (7) However, a semantic difference exists between the use of-(V)n and the use of these other markers: the Locative marker--(V)n indicates a static location on the Ground. With the Ablative and Allative postpositions and the Dative marker -ki, the idea of motion is evoked (i.e., the Ground would be the "target/recipient" of the motion event). Compare these two examples:

(4) a. pike-n ka_in [axos yi] mula.

house-LOC FOC/TENS child YI

'The child is in the house.'

b. pike-ki ka_in [axos yi] pumu.

house-DAT FOC/TENS child YI enter

'The child is entering the house.'

With the Locative marker--(V)n, the Figure is located on a specific area of the Ground. With the Dative marker--( V)s, the Figure is not in a single spot, but rather it is dispersed all over the Ground (as if it was spread over it). For instance:

(5) a. mesa natu-n ka_in [yakir yi] tsula.

table back/top-LOC FOC/TENS salt YI lie

'The salt is on the table.'

b. mesa natu-s ka_in [yakir yi] tsula.

table back/top-DAT FOC/TENS salt lie

'The salt is on the table, all over it.'

In summary, in the answers provided by the Trumai speakers for the stimulus kits, the Ground was often marked by--(V)n, but there were also instances of the combinations presented in (2).

Although the consultants used the construction with positional verbs extensively, this was not the only kind of answer provided. One of the consultants mentioned that two other constructions can also be used: one with a copula, one with zero copula, such as the examples below:

(6) mesa natu-n ka_in [ole] chi.

table back/top-LOC FOC/TENS manioc COP

'The manioc is on the table.'

(7) mesa natu-n ka_in [ole yi] O.

table back/top-LOC FOC/TENS manioc YI

'The manioc is on the table.'

And indeed, there were a few spontaneous instances of the zero copula construction in the elicited data. For example:

(8) ei-n ka_in [iwir yi] O.

stump-LOC FOC/TENS stick YI

'The stick is on the tree.'

The three locative constructions attested in the elicited data (i.e., positional verb; copula; zero copula) are also frequently observed in natural uses of the language. In Section 3, we will take a more careful look at them, and how they relate to the whole system of Trumai.

3. The locative constructions

3.1. The copular construction

This construction can actually have two possible word orders:

(i) Ground (Focus/Tense) Figure Copula

(ii) Figure Copula (Focus/Tense) Ground

Order (i) is the one that typically occurs in answers to where questions. As already mentioned, in Trumai new information comes in first position. Order (ii) is frequently used when speakers merely want to make a comment about the location of something. This order may occur in answers to where questions, but only occasionally so. As examples, we have:

(9) a. hikala-n [adis paine] chi.

village-LOC Indian group COP

'All the Indians are in the village.'

b. [adis paine]chi hilaka-n.

Indian group COP village-LOC

'All the Indians are in the village.'

(10) a. misu-n [ha] chi'.

river/water-LOC 1 COP

'I am in the fiver/water.'

b. [ha] chi misu-n.

1 COP river/water-LOC

'I am in the fiver/water.'

For the expression of negation, the particle tak is employed, modifying the Ground:

(11) a. misu-n tak [ha] chi'.

river/water-LOC NEG 1 COP

'I am not in the river/water.'

b. [ha] chi misu-n tak.

1 COP river/water-LOC NEG

'I am not in the fiver/water.'

With regard to tense, there are no special markers on the copula. For expressing the time of the location of a Figure in a particular place, Trumai speakers employ adverbs (such as 'yesterday'), or they change the Focus/ Tense particle, as in (12b).

(12) a. misu-n ka_in [ha] chi'.


'I am in the river/water.'

b. misu-n chi_in [ha] chi'.


'I was in the river/water.'

Compared to other kinds of constructions found in the language, we observe that the locative construction described here is very similar to nominal predicates. They have the following characteristics in common: (i) the presence of a copula in the clause; (ii) word order; (iii) the expression of "tense". However, when we take into consideration the negative modality, we observe similarities and differences:

--similarity: in nominal predicates, the element modified by the negator is not the copula, but rather the predicate itself;

-- difference: nominal predicates employ other kinds of negators: anuk and nikik.

Thus, although the locative copular construction is quite similar to nominal predicates, it has its own particularities. The table below presents a brief comparison between the two kinds of predicates:
Table 1. Comparison between copular locative and nominal predicates

 Locative Nominal predicate

ORDER Ground S Cop Pred S Cop
 S Cop Ground S Cop Pred

NEGATION Ground tak S Cop Pred {anuk}S Cop
 S Cop Ground tak {anuk}
 S Cop Pred nikik

3.2. The zero-copula construction

This construction is similar to the locative copular construction in all aspects (examples [13]-[15]), and it can actually be considered a subtype of it. The absence of the copula is linked to the presence of the morpheme yi in the NP that refers to the Figure. (8) Apparently, an incompatibility between this morpheme and the copula exists (16c).

Order and negation:

(13) a. mesa-n [k'ate yi] O.

table-LOC fish YI

'The fish is on the table.'

b. [k'ate yi] O mesa-n.

fish YI table-LOC

'The fish is on the table.'

(14) a. mesa-n tak [k'ate yi] O.

table-LOC NEG fish YI

'The fish is not on the table.'

b. [k'ate yi] O mesa-n tak.

fish YI table-LOC NEG

'The fish is not on the table.'

"Tense" manipulation:

(15) a. mesa natu-n ka_in [pola yi] O.

table back/top-LOC FOC/TENS ball YI

'The ball is on the table.'

b. mesa natu-n chi_in [pola yi] O.

table back/top-LOC FOC/TENS ball YI

'The ball was on the table.'

Incompatibility between Cop and yi:

(16) a. pike-n ka_in [kiki] chi'.


'The man is in the house.'

b. pike-n kain [kiki yi] O.

house-LOC FOC/TENSE man YI

'The man is in the house.'

c. * pike-n ka_in [kiki yi] chi'.


(The man is in the house)

The role of the morpheme yi is not easy to determine. Syntactically, it always occurs as the rightmost element of a 3rd person NP, and it can occur in an NP in any morphological case (i.e., Absolutive, Ergative, Dative). At first sight, one may think that yi is a definite article. However, this analysis does not work because: (i) although yi often occurs with definite nouns, there are instances of it with indefinite nouns as well; (ii) yi modifies not only nouns, but also 3rd person pronouns. A possible analysis for yi is to consider it a pragmatic marker, expressing information about the importance of the referent (Guirardello-Damian 2005), but this point needs to be further investigated. In any case, we observe the correlation between the presence of yi in the NP that refers to the Figure and the nonoccurrence of the copula. This holds for nominal predicates as well:

(17) a. t-adifle ka_in [di] chi.

3POSS-sister FOC/TENS woman COP

'The woman is his sister.'

b. t-adifle ka_in [di yi] O.

3POSS-sister FOC/TENS woman YI

'The woman is his sister.'

The conclusion is that the zero-copula construction is not a special type, but rather a subtype of the copular construction, which has a broader distribution (the copular construction occurs with all persons, while the zero copula is restricted to 3rd person NPs).

3.3. The positional-verb construction

This construction similarly presents variations of order. Again, order (i) is the one that typically occurs in answers to where questions, with the new information in first position. Order (ii) is used when speakers want to comment on the location of something.

(i) Ground (Focus/Tense) Figure Positional

(ii) Figure Positional (Focus/Tense) Ground

(18) a. yu'dut natu-n [murir] aha'tsi.

bench back/top-LOC basket sit

'The basket sits on the bench.'

b. [murir] aha'tsi yu'dut natu-n.

basket sit bench back/top-LOC

'The basket sits on the bench.'

For the expression of tense, there are no special markers on the positional verb. In Trumai, verbs do not receive tense-aspect-mood affixes. Aspectual differences can be expressed via the use of auxiliaries, while tense is expressed through the use of adverbs or the two Focus/Tense particles ka_in and chi_in (Example [19]). When a clause has neither adverbs nor the Focus/Tense particles, tense is understood from context.

(19) a. arukuru fax-on ka_in [ole yi] tsula.

basket viscera-LOC FOC/TENS cassava YI lie

'The cassava is (lit: lies) in the basket.'

b. arukuru fax-on chi_in [ole yi] tsula.

basket viscera-LOC FOC/TENS cassava YI lie

'The cassava was (lit: lay) in the basket.'

For the expression of negation, there is the use of the negator tak modifying the verb, as in (20), the same scenario observed in other verbal predicates. However, according to the consultants, the locative construction with positional verbs can have an alternative way of negation, not attested with ordinary verbal predicates: tak can modify the Ground, instead of the verb (21). This alternative way of negation is the only significant difference between the positional-verb locative predicate and other verbal predicates.

(20) yu'dut natu-n [pola] tsula tak.

bench back/top-LOC ball lie NEG

'The ball is not (lit: does not lie) on the bench.'

(21) yu'dut natu-n tak [pola] tsula.

bench back-Loc NEG ball lie

'The ball is not (lit: does not lie) on the bench.'

As a final remark, it should be pointed out that while the copular construction is sensitive to the occurrence of the morpheme yi in the NP that refers to the Figure (if yi, then zero copula), the positional-verb construction is not; i.e., a positional verb can occur with a NP that has the morpheme yi, as we can see in example (19) above.

3.4. The relationship between locative and existential predicates

The constructions presented in Sections 3.1 to 3.3 have a locative function (i.e., they are used when the speaker wants to point out the location of a Figure). However, they can also be used with a presentative function, i.e., they can be employed to introduce an entity by asserting its existence. In Trumai, affirmative existential clauses are not structurally different from locative ones. What makes a clause have an existential reading instead of a locative interpretation is the fact that the Figure being mentioned is still unknown to the listener as in (22). If the entity is known and identifiable, then the locative reading applies as in (23).

Entity is unknown, nonidentifiable: existential interpretation

(22) ni ka_in tahu chumuchu.

here FOC/TENS knife lie

'There is a knife here.'

Entity is known, identifiable: locative interpretation

(23) ni ka_in hai-kte tahu chumuchu.

here FOC/TENS 1-GEN knife lie

'My knife is here.'

Thus, the constructions presented in Sections 3.1, 3.2, and 3.3 can have either a locative or an existential interpretation depending on contextual information (as exemplified in [24]). The connection between locative and existential predication is attested in other languages of the world as well; for instance, in Manam (Lichtenberk 1983). Various authors have explored the relationship between these two kinds of predicates, such as Clark (1978) and Culioli (1995), among others.

Locative or Existential interpretation, depending on context

(24) pike-n ka_in yaw chi'.

house-LOC FOC/TENS human.being COP

'The people are in the house.'

[if 'people' are identifiable]

'There are people in the house.'

[if 'people' are unknown]

The scenario is a little different in the case of the negative modality. A negative copular or positional-verb construction can potentially have an existential sense if the Figure is considered nonidentifiable, as in (25)-(26). However, when Trumai speakers want to assert the nonexistence of an entity, they prefer to employ a different construction, which involves the use of the postposition nik 'without' as in (27)-(28). This construction has a negative existential interpretation only; it cannot have a locative sense.

(25) mesa-n tak ka_in k'ate chi.


'The fish is not on the table.'

'There is no fish on the table.'

[if Figure is nonidentifiable]

(26) tsula tak ka_in tar_hunhunke kama natu-n.

lie NEG FOC/TENS spider bed back/top-LOC

'The spider is not on the bed.'

'There is no spider on the bed.'

[if Figure is nonidentifiable]

(27) pola nik ka_in pola chi mesa natu-n.

ball without FOC/TENS ball COP table back/top-LOC

'There is no ball on the table.' (lit: ball is without ball on the table)

[this variant is less often attested]

(28) pola nik ka_in mesa natu-n.

ball without FOC/TENS table back/top-LOC

'There is no ball on the table.' (lit: (it) is without ball on the table)

[this variant is frequently used]

It should be pointed out here that the postposition nik also occurs in the construction used for the expression of negative predicative alienable possession (29). Thus, in Trumai a certain alignment exists between negative existentiality and negative alienable possession (although the two constructions are not exactly the same: the possessive construction has no Ground; and although [27] and [29] are quite parallel, note that in [29] the Possessor and the Possessed object are different entities). (9)

(29) sapaun nik ka_in ha chi.

soap without FOC/TENS 1 COP

'I do not have soap.' (lit: I am without soap)

In sum, it is interesting to observe that the system of Trumai is asymmetric with regard to the expression of existentiality. While the negative modality has a construction that has exclusively an existential sense, the affirmative modality does not, and the same constructions that perform a locative function can also cover a presentative one.

4. The class of positional verbs

Now that we have examined the various types of locative constructions found in Trumai, let us take a closer look at the class of verbs that occur in the positional-verb construction. The information presented in the next subsessions is relevant for the discussion in Section 5.

First, the formal characteristics of the positional verbs are described. Next, the semantics of each form is explored. The facts described here were also explored in another article (Guirardello-Damian 2002) which investigated the class of positional verbs in detail. The following sections merely present an overview of the Trumai system (for more information, cf. the mentioned article).

4.1. Formal characteristics

As outlined in Section 2, there are six positional forms that can occur in the locative construction. They can be divided into two subgroups:

(30) a. forms that provide information about orientation:

la be standing

aha'tsi sit/be sitting

chumuchu lie/be lying

b. forms that provide information about Ground and special conditions of support or visibility of the Figure:

tsula lie/be lying in a place that is not the floor

pila be in a liquid medium or be spatially like a liquid

mula be in a closed place or place where the visibility of the entity is affected

There are syntactic reasons to state that the positionals in (30) belong to a formal class: in general, these forms occur as the main verb of a clause, as illustrated in (31). However, they can also be found as an auxiliary modifying a main verb ([32]-[33]); (10) the auxiliary refers to the position/posture of the S or A argument of the clause. (11)


(31) ha tsula ka_in esak-en.

1 lie FOC/TENS hammock-LOC

'I am lying in the hammock.'

S V Aux

(32) hawal tsula.

1 sing lie

'I am singing while lying.'


(33) hai-ts oke kiti tsula Amati-tl.

1-ERG medicine give lie Amati-DAT

'I am giving medicine, while lying, to Amati.'

Ordinary verbs do not show these characteristics; they are found as the main verb of a clause only. Adjectival roots do not present such behavior either: they can be the predicate of a clause (34), but they cannot occur as an auxiliary modifying a verb (35). If an adjectival root modifies a main verb, it has an adverbial meaning, i.e., it refers to the quality of the event (e.g., She saw "beautiful" = She saw well). In order to refer to conditions or attributes of the S or A argument, a different construction is employed, with the adjectival root being modified by the postposition tam 'Comitative' (46).

(34) ha eni ka_in.

I dirty FOC/TENS

'I am dirty.'

(35) * hai-ts ha atle midoxos eni.

1-ERG 1 mother call dirty

(I called my mother, while I was dirty.)


(36) hai-ts ha atle midoxos ha eni tam.

1-ERG 1 mother call 1 dirty COM

'I called my mother, while I was dirty.'

In other words, the forms in (30) have syntactic properties that give them a special status in the system of Trumai. Another reason to group them together is the fact that some of the forms are phonologically similar (la, tsula, pila, and mula all end in the same syllable, which suggests that they do belong to the same formal class). (12)

It is important to point out that although the positionals can occur as an auxiliary, the situation is slightly different in the case of aha'tsi 'sit'. For the expression of the 'sitting' position, there is a form that is found exclusively as an auxiliary, katsi (38). According to one of my consultants, the use of aha'tsi as an auxiliary would be possible, but such use is not observed in spontaneous speech.

(37) ha aha'tsi ka_in.

1 sit FOC/TENS

'I sit down.' or 'I am sitting.'

(38) hai-ts Tata midoxos katsi.

1-ERG Tata call be.sitting

'I am calling, while sitting, Tata.'

Note in the examples above that while the verb aha'tsi can have both active and stative meanings ('sit' or 'be sitting'), the auxiliary katsi has a stative meaning only ('be sitting'). This is also true for the positionals chumuchu and tsula: their occurrence as verb can always have a double interpretation ('lie' or 'be lying'), but the auxiliary use has a stative reading only ('be lying'). For instance:

(39) kiki yi tsula chi_in.

man YI lie FOC/TENS

'The man lay down.' or 'The man was lying.'

[verb: active/stative sense]

(40) kiki yi otl tsula chi_in.

man YI sleep lie FOC/TENS

'The man was sleeping lying.'

(*The man lay down in order to sleep)

[aux: stative sense only]

In the case of la, the scenario is different: the verb la has a stative meaning only ('be standing'). For the expression of the active sense, another form is used, lakida:

(41) a. ha la ka_in.

1 be.standing FOC/TENS

'I am standing.'

b. wana la!

IMP be.standing

'Stay in a standing position!'

(42) a. ha lakida ka_in.

1 stand.up FOC/TENS

'I stood up.'

b. wana lakida!

IMP stand.up

'Stand up!'

Thus, both the verb la and the auxiliary la have stative interpretations, and it is not clear what the semantic difference between them is (the auxiliary does not seem to convey information on aspect, like in other languages where posture auxiliaries can express the durative/continuative aspect; cf. Keegan 1997). The same can be said with regard to the positionals pila and mula: the semantic difference between the verbal use and the auxiliary use is still unknown (in both uses, pila and mula have stative meanings).

4.2. The semantics of the forms

4.2.1. Positionals that provide information about posture/orientation. La, aha'tsi and chumuchu can be used with all kinds of NPs, i.e., animates and inanimates. Their occurrence with animals and objects depends on the physical characteristics of the Figure. The tables below present the positionals that can be used for each kind of animal or object.

As we can see, for unidimensional objects or animals whose bodies do not present radical variations in their body arrangement (such as turtles, crabs, lizards, snakes, spiders, cockroaches, frogs, etc), 'lying' is the default posture. As Newman (2002) points out, postures present differences with regard to the level of physical effort that they require to be maintained. 'Lying' is the easiest posture to hold; it is the one that can be assumed without any special efforts. This being so, 'lying' is probably the least marked of postures, and it might be that this is the reason why in Trumai 'lying' is the posture assigned to Figures whose body do not really change their configurations. (14)

For animals and objects that can have variations in their body configuration, and for human beings as well, la, aha'tsi and chumuchu are used according to the orientation that the Figure has at a certain moment.

La is the positional used when the Figure is projected in a perpendicular means in relation to the Ground: if the Ground is horizontal, the Figure is vertically oriented. If the Ground is vertical, the Figure is in horizontal orientation. Below, examples of situations for which la is used are presented. Note that for human beings and bidimensional objects, the perpendicular orientation of the whole Figure is the important factor (Ia, Ib, IIa-IIc), while for animals and pieces of furniture it is the orientation of the legs that matters (Ic, Id). In the case of Figures that are in a Ground that is against the force of gravity (e.g., a roof), la is used for the situations in which the Figure is fixed and stable (IIIa). If the Figure is not fixed and can move or oscillate, another form is used, tonela 'hang/ drape' (cf. Section 4.3 for more details). Finally, la can be used for Figures that do not move or change position, but whose configuration also has a perpendicular orientation in relation to the Ground (Ig, Ih).

I. Ground (the floor) is horizontal, Figure is vertical

a. a human being standing on the floor (i.e., the person is vertically straight and supported by the feet);

b. bidimensional objects, preferably with an elongated shape (e.g., a stick, pen): the object is vertically projected;

c. four-legged animals whose legs are all straight and vertically oriented;

d. birds whose legs are straight and vertical;

e. pieces of furniture that have four legs, like chairs and tables (but not beds): the legs of the furniture are vertically oriented;

f. butterfly on the floor, with its wings closed and vertically straight;

g. houses;

h. trees; plants that grow vertically, such as corn, sugarcane, etc.

II. Ground is vertical, Figure is horizontal

a. nail in a wall, with a big part of it standing out;

b. knife inserted into the trunk of a tree;

c. a cigarette in the mouth of a person;

d. a hammock that is tied to posts and ready to be used.

III. Ground is horizontal, Figure is vertical

a. a light bulb directly attached to a metal frame on the roof.

Thus, although the positional la in some contexts can be glossed as 'be standing' (or as 'estar em pe', in the case of Portuguese), its semantics is not exactly parallel to the English or Portuguese expressions. It would be interesting to analyze how the speakers of Trumai who are also bilingual in Brazilian Portuguese deal with the differences between these two languages (while in Trumai the scenes mentioned above are all categorized as being of the same type--they can all be described by la--in Portuguese the scenes in II and III would be considered of a different kind in relation to the scenes in Ia, Ic and Id. Furthermore, in Brazilian Portuguese one does not use positional verbs with inanimate entities as such use is considered odd).

The positional aha'tsi is linked to the 'sitting' posture. This is the form used to describe a human being or a four-legged animal in a relatively compact position whose buttocks are in contact with the Ground. (15) This positional can also be used for objects that have three dimensions and a base, such as a bowl, a pot or a bottle; aha'tsi is employed when the object is in its canonical position, i.e., upwards with the base touching the Ground. (16) In all cases (people, animals, objects), the Ground is horizontal and gives support to the Figure; aha'tsi is not found describing Figures on surfaces that are vertical (e.g., a spider on a wall) or against the force of gravity (e.g., a snail on a roof).

Chumuchu is the positional form used when the Figure is in a horizontal orientation. A substantial extension of the body of the Figure touches the Ground, which gives support to the Figure. Chumuchu is used for human beings and animals lying on a horizontal surface. It is also used for objects with two dimensions and an elongated shape whose longer axis is horizontally oriented (e.g., a stick lying on a table). In the case of objects with three dimensions and a base, the use of chumuchu does not exactly involve horizontal orientation, but rather noncanonical configuration: chumuchu is employed when the object is on its side (e.g., a bottle lying on a chair) or it is upside down.

Chumuchu can also be used for beds--a type of furniture whose longer axis is horizontally oriented--and for Figures that do not move or change position, but whose configuration is projected in a horizontal axis, such as villages and ground plants (e.g., watermelons, pumpkins, etc). Finally, chumuchu is the form that describes the position of unidimensional objects and animals with a rigid body. For these, the use of chumuchu is not linked to horizontality; it occurs because--as already mentioned--'lying' is the default body configuration assigned to these kinds of Figures.

4.2.2. Positionals that provide information about Ground. Like the positionals described in the previous section, tsula, pila, and mula can also be used with all types of NPs (i.e., animates and inanimates). In the case of mula and pila, the physical characteristics of the Figure do not matter; it is the kind of Ground on which the Figure is that determines their use. In the case of tsula, the characteristics of the Figure play a role, because for tsula the kind of contact between the Figure and the Ground is an important factor.

Pila is the form used to indicate that a Figure is in a liquid medium. This is the positional used to describe people, animals and objects in the river, when a good portion of their bodies is immersed in the water. (17) However, immersion in the water itself is not the crucial factor; what really counts is the fact that the water is the Ground giving support to the Figure. If an object is under water but at the bottom of the river, we have the use of chumuchu 'lie' instead of pila, because in this situation the Ground is not the water, but rather the bottom of the river. Rather than 'be immersed in a liquid', the proper semantics of pila is 'be suspended in a liquid'. Variations of posture presented by a Figure in a liquid medium are not distinguished: while immersed in water, even if the Figure varies its posture (i.e., be standing; be lying; be sitting), pila is always the positional used.

Pila can also be employed for cases in which the Figure itself is a liquid. Its occurrence indicates that the Ground is not a container and the liquid is loose in its form (43). Pila is also used to indicate the location of lakes and rivers (44). However, if the liquid is in a container, pila cannot be used. Instead, we have the positionals mula, if the container is a closed area (45a), or aha'tsi 'sit', if the container is open (45b). The occurrence of aha'tsi for liquids in containers probably is linked to the fact that the container itself "sits", since it is a three-dimensional object with a base. The liquid inside of a container assumes its shape; thus, if the container "sits", so does the liquid in it--in other words, we have a Figure/ Ground reversal. Such reversal is also attested in other languages of the world, such as Tzeltal (cf. Brown 1994).

(43) mesa natu-n kafe su pila.

table back/top-LOC coffee juice

'The coffee (liquid) is on the table, spread.'

(44) Anaria xuik ka_in Su_Suxdi_xu pila.

Anaria near FOC/TENSwater_milky_lake

'The "Milky Water" lake is near Anaria.'

(45) a. gahafatermika-n kafe su mula. bottle thermal-LOC coffee juice

'The coffee (liquid) is in the thermos.'

b. kopu faxo-n kafe su aha'tsi.

glass viscera-LOC coffee juice sit

'The coffee (liquid) is in the glass.'

The verb mula is a form that expresses not only information linked to Ground (be located in a closed area, as in 46), but other meanings as well: in some cases, mula means 'live with' (47); in other examples, its meaning is 'be in seclusion' (48).

(46) kaicha fax-on ka_in katat yi mula.

box viscera-LOC FOC/TENS bottle YI

'The bottle is in the box (which is closed).'

(47) ha pa-s tenuk hai tam hi mula.

1 marry-TEMP then 1 COM 2 live

'When I get married, you will live with me.'

(48) ha mula hen. 1 be.seclusion then

'Then I stayed in seclusion.'

[Trumai teenagers undergo a period of seclusion in preparation for adult life]

When mula is used in locative predicates, it indicates that the Figure is in a location that is closed, such as a house, a box, a bag, a wardrobe, etc. The configuration of the Figure inside of the closed place is not relevant; for example, in (46), it does not matter if the bottle is oriented upwards or on its side. What determines the use of mula is the fact that the Ground is a closed area, surrounding the Figure (e.g., when a foot is inside a sock, the foot is being "enveloped" by it). However, the Ground does not necessarily have to be a completely closed area. Sometimes there might be an opening in the place, but for some reasons the Figure inside of it cannot be seen or is only partially visible (e.g., an owl in a hole in a tree; only part of its body can be seen). Thus, nonvisibility (or low-visibility) is also an important factor in the occurrence of mula. This positional is employed when the Figure is inside a Ground that blocks vision (most likely a closed area, since closed places are very effective in blocking the vision of a Figure).

Finally, we have the positional tsula. Like pila and mula, the main role of tsula is to provide information about the kind of Ground where the Figure is. However, the semantics of this positional can also indirectly imply the 'lying' posture, for the reasons outlined below.

Tsula describes a Figure that is on a Ground that is not at the floor level; if the Figure is on the floor, the use of tsula is not allowed. The Ground can be in any orientation: horizontal (e.g., a pen on a table); vertical (e.g., a poster on a wall); horizontal but against the force of gravity (e.g., a spider on a roof). The Figure is parallel to the Ground and its whole body--or at least a substantial portion of it--is in contact with it. The parallelism of the Figure in relation to the Ground is what makes tsula different from the positional la.

Thus, if the Ground is horizontal, tsula indicates that the Figure is equally horizontal, since it is parallel to it. As a consequence, tsula can sometimes overlap with the positional chumuchu 'lie', which (as shown in Section 4.2.1) describes Figures in horizontal orientation. During the investigation of the BLC of Trumai, I observed that there are scenes for which both chumuchu and tsula can be used. For example:

(49) a. kama natu-n ka_in ha adifle chumuchu.

bed back/top-LOC FOC/TENS 1 sister lie

'My sister is lying on the bed.'

b. kama natu-n ka_in ha adifle

bed back/top-LOC FOC/TENS 1 sister tsula.

'My sister is lying on the bed.'

It seems that Trumai speakers choose one form over the other depending on which aspect of the scene they want to highlight: according to the consultants, if tsula is used, the focus is on the fact that the Figure is in a place that is not the floor. If chumuchu is chosen, the focus is now on the lying posture itself (i.e., the Figure is not standing or sitting, but lying). The semantic difference between the two positionals is more clearly illustrated by the examples below:

(50) a. wana chumuchu!

IMP lie

'Lie down (on the table)!'

[situation: the child is standing on the table, and I want her to change position]

b. wana tsula!

IMP lie

'Lie (on the table)!'

[situation: the child is on the floor, and I want her to climb on the table.]

It should be mentioned here that although tsula and chumuchu can overlap in some cases, this does not occur for all scenarios. There are situation in which the use of chumuchu is not possible, while tsula can be applied. For instance, a sock on a foot: the Figure here is not clearly in a vertical or horizontal orientation. Part of it is horizontal (foot), part of it is vertical (the shin and leg). For this situation, tsula is the positional used, but not chumuchu 'lie':

(51) ha pits'-an ka_in meya tsula.

1 foot-LOC FOC/TENS sock lie

'The sock is on my foot.' (18)

Figures 1 and 2 below present possible algorithms that a speaker would use for choosing a positional verb in Trumai. Figure 1 shows the criteria for choosing among the various positional verbs. Figure 2 shows the algorithm for choosing between the verbs chumuchu and tsula, whose uses partially overlap.

4.3. Other verbs

This final section is dedicated to some verbs that were also attested in the answers provided by the Trumai consultants for the stimulus kit "Picture Series for Positional Verbs", and which need to be further investigated.

4.3.1. Tonela. For some scenes of the stimulus kit, the verb tonela 'hang/drape' was employed. In principle, this verb would be a good candidate to be a member of the formal class described in Section 4.1: tonela exhibits formal similarity to some of the other positionals--its final syllable is la--and its semantics also involves information about spatial configuration. Note that in some languages of the world, 'hang' aligns with positional verbs; for example, in Goemai, a Nigerian language, 'hang' is member of a formal class that includes postural verbs such as 'sit', 'stand', and 'lie' (cf. Hellwig, this volume).

Apparently tonela cannot be used as an auxiliary, however. In spontaneous speech, such use is never attested (speakers use the construction with tam 'Comitative' instead, as in 52). When an example with tonela as an auxiliary was presented to a consultant, he found it strange and was not sure if the example would be acceptable (53). Thus, it seems that tonela does not belong to the class presented in Section 4--or, if it does, it would be a marginal member since it does not show all the properties of the other positionals.

(52) tonela-a tam chi_in ine-k tsi-tle

hang-3POSSCOM FOC/TENS 3-ERG 3POSS-mother midoxos.


'He, while hanging (from a tree), called his mother.'

(53) ?ine-k chi_in tsi-tle mixodos tonela.

3-ERG FOC/TENS 3POSS-mother call hang

(He, while hanging, called his mother.)

[not good, according to the consultant]



4.3.2. Pes. In certain scenes of the stimulus kit, the Figure consisted of a group of objects--for example, several cassava roots, several bottles. For these scenes, sometimes the verb pes was employed. Apparently, the meaning of this verb is 'be in a group', and there is no reference to the orientation of the Figure (i.e., if the objects are standing, lying, or sitting. Pes was used for all kinds of configurations). In the majority of the cases, the consultant offered pes as an alternative to one of the positionals presented in Section 4.1 (as in [54]). It seems that pes cannot be used as an auxiliary, therefore it cannot be considered a member of the class of positionals.

(54) a. First answer:

tehnene-n ka_in kuman chumuchu.

floor-LOC FOC/TENS beans lie

'The beans are lying on the floor.'

b. Second answer:

"Given that there are many beans here, you can say pes."

tehnene-n ka_in kuman pes.

floor-LOC FOC/TENS beans

'The beans are lying on the floor.' 4.3.3. Hybrid descriptions. Some answers provided by the Trumai speakers for particular scenes of the stimulus kit are peculiar: in the answer, a main predicate is modified by a positional auxiliary. While the verb refers to one part of the Figure, the auxiliary refers to another part of it and its relation to the Ground. For instance:

(55) Situation: a piece of clothing is on a table. Part of it is lying on the table; the other part is hanging.

Ground Figure

mesa natu-n kain [yaw.mut perew yi]

table back/top-LOC FOC/TENS clothing piece YI

V Aux

tonela tsula.

hang lie

'The piece of clothing is hanging lying on the table.'

This is a kind of hybrid description which points out two spatial configurations for the same scene. In the elicited data, the main predicate is always one of the following forms: tonela 'hang/drape'; mantsu 'be leaned'; uyar 'be touching', or homama 'be encircled'. There was no instance of two positional forms combined, i.e., a positional verb + a positional auxiliary (e.g., *chumuchu la, *aha'tsi chumuchu, etc).

It is interesting to observe that the hybrid descriptions occurred with Figures that have an elongated shape: a piece of cloth, a stick, and a piece of rope. Since the Figure is elongated, it can be subdivided into two parts (or two extremities), and each one can assume its own shape or spatial configuration. From this perspective, the use of a hybrid description actually makes sense.

It is necessary to investigate if these hybrid descriptions also occur in natural speech, or if they were just a special strategy used during the elicitation because the consultants wanted to provide very accurate answers to the where questions. Consider this explanation given by a consultant:

(56) Situation: a piece of clothing is dangling from a basket; part of the clothing is touching the floor.

a. First answer:

arukuru-n ka_in [yaw.mut perew yi] tsula.

basket-LOC FOC/TENS clothing piece YI lie

'The piece of clothing is lying on the basket.'

b. Second answer:

"Since it (the piece of clothing) is touching the basket and it is also on the floor, a more accurate answer is:" ni ka(_in) arukuru xop-an tonela chumuchu. here FOC/TENS basket mouth-LOC hang lie
 'It (the piece of clothing) is here, hanging lying on the border of
 the basket.' [i.e., the piece of clothing hangs on the basket, and
 lies on the floor]

5. Analysis: the interplay between the copular and the positional-verb constructions

5.1. Main discussion

Now that we have seen the Trumai locative constructions in a more detailed way, let us return to the issues that were raised earlier in the article. The first question that needs to be addressed is: why does Trumai present two different types of locative construction (i.e., the positional verb versus the copular construction)? What would motive the coexistence of these patterns?

In natural uses of the language, the occurrence of both constructions is attested, as in (57). However, when we look more carefully at the data, we recognize that the use of the positional-verb construction is highly frequent, almost obligatory, for the location or assertion of existence of Figures that have a fixed spatial configuration, such as houses, villages, trees, rivers. In my corpus of the Trumai language, there are numerous spontaneous examples of positional verbs in locative and existential clauses with Figures that have a permanent location (examples [58] and [59] come from texts); in contrast, the number of instances with the copular construction (or with its variant, the zero copula) is very small.

(57) hamuna in hi chi?

where FOC 2 COP

'Where are you?'

a. ni ka(_in) ha chi. [one possible answer]


'I am here.'

b. ni ka(_in) ha aha'tsi. [another possible answer]

here FOC/TENS 1 sit

'I am sitting here.'

(58) ina tsile de, [pike t'ox yi] la le.

there hearsay already house building YI stand hearsay

'So, people say that there was a house there.' (lit: a house stood there)

(59) kaina ka_de(_in) Trumai hilaka-t' chumuchu le.

there FOC/TENS Trumai village-NzrEx lie hearsay

'People say that the old Trumai village was there.' (lit: the village lay there)

The scenario is slightly different with regard to the location of Figures that can change location and spatial configuration, such as human beings, animals, and objects. For those, there is much more variation in the data: sometimes the verbal construction is used, sometimes the copula/zero-copula one. There is one situation in which the copula/zero copula is the form necessarily chosen: it is when the Figure is in the air, which is a special kind of Ground (it does not give support to the Figure; the "contact" between the Figure and the Ground is not obvious, since the Ground in question is not visible). It seems that there is something distinctive about the air that does not allow the use of the positionals; in this case, the copular/zero-copular construction is the only possible alternative. However, for other kinds of Grounds, both constructions are acceptable.

I believe that the use of positional verbs in the locative construction is something more recent in the history of the language than the use of the copula in this same construction. My hypothesis is that at some moment in the history of Trumai, the copula was used in the locative construction, but not the positional verbs. Their original function was merely descriptive, i.e., they were used exclusively to describe the configuration of an entity (standing, lying, etc). Eventually the use of positional verbs was expanded to cover the locative and presentative functions as well--as Newman (2002: 7) points out, in some languages positional/posture verbs "may be extended to help conceptualize the position of some entity". Thus, the positional verbs started being employed in locative and existential constructions as well. They began competing with the copular construction for the same functional domain, creating the current scenario observed in the language.

The positional-verb locative construction seems to be becoming more and more entrenched. It is already turning into the typical way of expressing the location or existence of Figures whose spatial configuration is fixed. If the evolution continues, it might be that in the future the positional-verb construction will become the canonical way of expressing location/existence of all sorts of Figures, replacing the copular locative construction completely, or leaving it only for very specific situations (like Figures in the air). If my hypothesis is correct, Trumai would be a language moving from type Ia (copular construction) to type III (positional verbs construction) in the typology scale presented in Section 1.2.

In the present stage, the language is in an intermediate position. On the one hand, Trumai cannot be considered a language of type Ia because positional verbs definitively can be used in answers to where questions (and not only that: they are also used in existential clauses, a fact that is very significant). On the other hand, Trumai cannot be considered a pure typeIII language because the positional-verb construction is not yet the canonical way of expressing location/existentiality; the locative copula/ zero-copula construction also exists in the language and it is used as well. Therefore, a more appropriate classification for Trumai would be to place it in an intermediate position across the various language types.

This leads us to another relevant question: if the current speakers of Trumai have two main locative constructions available, how do they select one over the other? When asked about the difference between them, one consultant said that he considers the positional-verb construction "more informative" than the copular one, and for this reason sometimes it would be more appropriate to use it. It is interesting to observe that Trumai speakers do present a high use of the positional-verb construction when they perform tasks that require more precision in the localization of a Figure (such as in providing answers to the "Picture Series for Positional Verbs"). (19) On the other hand, when a Trumai speaker wants to talk about a person but does not know his/her position, the speaker uses the copula instead of a positional verb:

(60) yaw chi' ka_in pike iuda-n.

human.being COP FOC/TENS house behind-LOC

'There is a person behind the house.'

Thus, the choice of one construction over the other is linked to the degree of information that the speaker wants to/can present to the listener, i.e., the choice of a particular construction reflects the degree of information available to the speaker.

It also has to be taken into consideration that Trumai has only one general locative marker for static location, the morpheme -(V)n. As explained previously, this marker is very generic in meaning, signaling simply a locative relation between Figure and Ground. In order to clarify this relationship, body parts terms can be employed, allowing the speaker to express other nuances (such as 'in', 'on', 'beside', 'behind', 'under', etc) in the locative relation. The use of positional verbs would be one step further, allowing the speaker to specify the relationship in greater detail.

The scenario found in Trumai is similar to what is observed in others languages of the world, such as Lipke (a Niger-Congo language from Ghana) and Tzeltal (a Mayan language from Mexico). Like Trumai, these languages employ various types of spatial configuration verbs in the locative construction, and have "highly general adpositions that do little more than signal that there is some type of locative relation, typically leaving it to the verbs to specify the nature of the relation" (Kita and Walsh Dickey 1998: 57). However, it should be pointed out that Lipke and Tzeltal are not of the same type as Trumai--they are actually type II in the working typology presented in Section 1.2. Thus, even though Trumai is not a type II language, it shares some features with them.

The fact that Trumai positional verbs help in the task of describing the Figure-Ground relationship lead us to another important issue: the role that these verbs can play in the system of the language.

5.2. The classificatory role of the positionals

As mentioned in Section 4.2.1, for the use of the positionals la 'stand', aha'tsi 'sit' and chumuchu 'lie', the physical characteristics of the Figure are a relevant issue. This being so, we could state that although the crucial semantic feature of these positionals is to present information about the orientation of the Figure, they also provide some information about the kind of Figure whose configuration is being described.

The use of la 'stand' with animals indicates that the body of the animal is flexible and can change its configuration. In the case of objects, the use of la signals that the Figure is not unidimensional and does not have a base. The occurrence of the positional aha'tsi with animals indicates that the animal has four legs. If aha'tsi is employed for an object, this means that the object has a base.

In the case of ehumuchu, the situation is slightly different: given that this positional occurs with all kinds of objects and animals, it is not so helpful to learn that a particular type of Figure accepts the occurrence of chumuchu. What is more informative is to learn that the Figure in question allows only the use of this positional. If this is the case, then we know that the figure is an animal whose body cannot present extreme changes in its configuration, or it is an object that has one dimension only (or it is perceived as being unidimensional). (20)

In other words, the use of the positionals that are linked to posture (la, aha'tsi, chumuchu) provide a way of classifying Figures in various types:

(i) Type 1: Figures that can assume all possible postures. These are human beings and four-legged animals with a malleable body. No object is included here;

(ii) Type 2: Figures that can assume two postures only, and one of the possible postures is 'lie':

a. Stand+ Lie: two-legged animals (birds); two-dimensional objects (objects that have an axis that can be projected perpendicularly from the Ground);

b. Sit + Lie: three-dimensional objects with a base;

(iii) Type 3: Figures that assume one posture only: 'lie'. These Figures are animals with a nonflexible body and unidimensional objects.

It is important to note here that the classification presented above applies to moveable Figures. Nonmoveable Figures, i.e., the ones that have a fixed location--such as houses, villages, trees, lakes, etc.--constitute a different case. Like unidimensional Figures, nonmoveable Figures have only one possible choice of positional, but this is not necessarily 'lie'. The positional will be selected according to the physical characteristics of the Figure: houses and trees "stand", since their major axis is vertically oriented in relation to the floor; villages "lie", because they are spread in a horizontal orientation. In other words, nonmoveable Figures are classified according to the same principles as movable Figures, but in a more permanent way.

The classification provided by the positionals la, aha'tsi, and chumuchu for moveable Figures is congruent with other ways of categorization found in the language. For example, the use of the pluralizers a 'Dual' and wan 'Plural' indicates that Trumai distinguishes between nouns that refer to animate entities (with a human and nonhuman subdivision) and inanimate entities (cf. Guirardello 1999: Ch. 2). The use of the Dative markers -(V)tl, -ki, and -(V)s also draws a distinction among nouns referring to humans, animates, and inanimates (cf. Guirardello 1999: Ch. 7). The contrast between animate and inanimate entities further manifests itself in the use of the Imperative particles wa versus waki (cf. Guirardello 1999: Ch. 3).

This same type of distinction is observed in the use of the positionals linked to orientation (although here what is classified is not the noun itself, but rather the type of Figure): in the scheme above, note that there is no inanimate Figure that allows the use of all three positionals. Among the animate ones, there is a subdivision between human and animals, because not all animals permit the occurrence of the three positionals, whereas humans do.

One interesting aspect about the classification provided by the positionals la, aha'tsi, and chumuchu is that it leads to a further subdivision among the inanimate Figures: one-posture Figures versus two-posture Figures. The other modes of categorization found in the language (pluralizers, Dative markers, Imperative particles) do not suggest additional divisions for the nouns that refer to inanimate entities. Maybe this is an indication that the system of classification provided by the positionals is indeed of a different type, categorizing not nouns, but rather nominal concepts. What this means is as follows: when analyzing Yeli Dnye, a language of Papua New Guinea that also employs positionals in the locative construction, Levinson formulates some hypotheses about languages that have a positional system. One of his hypotheses is that "these systems are classificatory, since they assign different subject nominals to different predicates under specific conditions" (Van Geenhoven and Warner 1999: 68). Then he raises a question: "what is actually classified, the referent, the noun, or the nominal concept?" In the case of Yeli Dnye, Levinson concludes that it is the nominal concept that is classified, and provides one example: for the same array of six balls on the ground, one uses the verb 'sit' if talking about them as separated entities ("the balls are sitting on the ground"), but one can also use the verb 'stand' if the balls are seen as forming a pile ("a small pile of balls is standing on the ground").

In the case of Trumai, we can also state that it is the nominal concept that is classified by the use of positionals, and not the referent or the noun itself. For example, mut perew means 'piece of cloth'. With regard to the use of pluralizers, mut perew is always classified as an 'inanimate' noun. However, with regard to the use of positionals, mut perew can have more than one possible classification, depending on whether it is conceived as a unidimensional Figure (e.g., when the cloth is rolled up) or as a Figure with an elongated shape that can 'stand' in relation to the Ground (e.g., when the cloth is unfolded). The referent is always the same (a piece of cloth) and so is the noun (rout perew), but the nominal concept changes depending on how the object is perceived by the speaker--and the positional changes accordingly.

Levinson raises another relevant hypothesis with regard to positional systems: "systems of this kind always have a residual category" (Van Geenhoven and Warner 1999: 68), that is, a posture that is assigned when the Figure does not meet the conditions to be designated to the other postures. In the case of Yeli Dnye, the residual category is 'sit'. In the case of Trumai, the residual category is 'lie', as we can see in the algorithm presented in Figures 1 and 2: when the entity does not fulfill the criteria to be assigned to the other postures, it falls into the 'lie' category.

6. General conclusion

As we have seen in the present study, in Trumai there are two main types of constructions that can be used to reply to where questions: a copular construction (with a subtype, the zero copula), and a positional-verb construction. These two constructions coexist and both are frequently attested in the data of the language. Although the positional-verb locative construction cannot be considered yet the Basic Locative Construction of Trumai, it seems to be evolving in this direction. The high occurrence of positional verbs in locative and existential clauses with nonmoveable Figures suggests that the use of the positional-verb construction is becoming more entrenched and may turn into the canonical way of expressing existentiality/location for such kinds of Figure. In the current stage of the language, both constructions are employable and acceptable. For this reason, it seems more appropriate to say that Trumai is not exactly a type III language in the working typology proposed by Van Geenhoven and Warner (1999), but rather holds an intermediate position between type Ia and III.

With regard to the interplay between the copular and the positional-verb constructions, they express different degrees of information about the spatial relationship between Figure and Ground. The copular construction is less specific as the copula does not provide spatial information (its function is simply to link the subject to the predicate). The information is codified by the general locative marker that appears on the Ground. However, like Tzeltal and other languages of the world, the locative marker of Trumai is very generic in its meaning, merely indicating that some kind of locative relationship exists. To detail the spatial information further, one needs to employ body part nominals to mark the Ground as well. In the case of the positional-verb construction, the spatial information is codified not only by the locative marker/body part nominals, but by the verb as well, which results in a more precise view of the relationship between Figure and Ground. The uses of these various elements (locative marker, body part nominals, positional verbs) allow the speaker to refine the spatial information s/he wants to convey. Finally, regarding the six positional forms that occur in the locative construction, they constitute a formal class, they have quite interesting semantic contents, and the use of the positionals linked to 'posture' (la, aha'tsi, and chumuchu) allows the possibility of classifying Figures under different types, depending on the number of posture forms that a Figure can take. The categorization that results from their use is compatible with other means of classification attested in the language (in the sense that it also distinguishes animates from inanimates), but at the same time has its own dynamics--as in other languages that have a positional system (e.g., Yeli Dnye), what is classified in Trumai is not the noun, but rather the nominal concept. Thus, the facts found in Trumai confirm some of the hypotheses about languages of this type.

Appendix. Abbreviations used in glosses

ablat = ablative; abs = absolutive; allat = alative; com = comitative; cop = copula; dat = dative; erg = ergative; foc = focus; foe/tens--particle of focus + tense; imp = imperative; gen = genitive; loc = locative; neg = negation; nzr = nominalizer; poss = possessive; temp = temporal subordinate clause marker.

Received 8 June 2005

Revised version received

9 January 2007

University of the West of England

Max Planck Institute for Psycholinguistics


Ameka, Felix; De Witte, Carlien; and Wilkins, David (1999). Picture series for positional verbs: eliciting the verbal component in locative descriptions. In Field Manual 1999, David Wilkins (ed.), 48-56. Nijmegen: Max Planck Institute for Psycholinguistics.

Bowerman, Melissa and Pederson, Eric (1993). Topological relations picture series. In Manual for the Space Stimuli Kit 1.2, Eve Danziger and Deborah Hill (eds.), 40-50. Nijmegen: Max Planck Institute for Psycholinguistics.

Brown, Penelope (1994). The INs and ONs of Tzeltal locative expressions: the semantics of static descriptions of location. Linguistics 32(4/5), 743-790.

Clark, Eve V. (1978). Locationals: existential, locative and possessive constructions. In Universals of Human Language, vol. 4, Joseph Greenberg (ed.), 85-126. Stanford: Stanford University Press.

Culioli, Antoine (1995). Cognition and Representation in Linguistic Theory. Amsterdam and Philadelphia: John Benjamins.

Dourado, Luciana (2001). Aspectos morfossintaticos da lingua Panara (Je). Unpublished doctoral dissertation, Universidade Estadual de Campinas (Unicamp), Brazil.

Franchetto, Bruna (2006). Are Kuikuro roots lexical categories? In Lexical Categories and Root Classes in Amerindian Languages, Valentina Vapnarsky and Ximena Lois (eds.), 33-68. Bern: Peter Lang.

Guirardello, Raquel (1999). A reference grammar of Trumai. Unpublished doctoral dissertation, Rice University.

Guirardello-Damian, Raquel (2002). The syntax and semantics of posture forms in Trumai. In The Linguistics of Sitting, Standing, and Lying, John Newman (ed.), 141-177. Amsterdam and Philadelphia: John Benjamins.

--(2005). Fonologia, classes de palavras e tipos de predicado em Trumai. Boletim do Museu Paraense Emilio Goeldi, Serie Ciencias Humanas 2005 1(2), 193-306.

Keegan, John M. (1997). A Reference Grammar of Mbay. Munich: Lincom Europa.

Kita, Sotaro and Walsh Dickey, Laura (eds.) (1998). Annual report 1998, Max Planck Institute for Psycholinguistics.

Lichtenberk, Frantisek (1983). A Grammar of Wardaman: A Language of the Northern Territory of Australia. Berlin and New York: Mouton de Gruyter.

Meira, Sergio (2004). O linguista e a ortografia indigena: o caso da lingua Bakairi. Revista de Estudos e Pesquisas 1(2), 73-99.

Newman, John (2002). A cross-linguistic overview of the posture verbs 'sit', 'stand', and 'lie'. In The Linguistics of Sitting, Standing, and Lying, John Newman (ed.), 1-24. Amsterdam and Philadelphia: John Benjamins.

Seki, Lucy (2000). Gramatica do Kamaiura, lingua Tupi-Guarani do Alto Xingu. Campinas: Editora da Universidade de Campinas.

Senft, Gunter and Smits, Roel (eds.) (2000). Annual report 2000, Max Planck Institute for Psycholinguistics.

Talmy, Leonard (1985). Lexicalization patterns: semantic structure in lexical forms. In Language Typology and Syntactic Description 3: Grammatical Categories and the Lexicon, Timothy Shopen (ed.), 57-147. Cambridge: Cambridge University Press.

Van Geenhoven, Veerle and Warner, Natasha (eds.) (1999). Annual report 1999, Max Planck Institute for Psycholinguistics.


(1.) I would like to present my thanks to Felix Ameka for comments and suggestions; to Stephen Levinson for providing support for my research; to the Max Planck Institute for Psycholinguistics for financial and technical assistance; to Bruna Franchetto, Sebastian Drnde, and Sergio Meira for providing information on the Kuikuro, Aweti and Bakairi languages, respectively; and to the Trumai consultants who participated in the present study. Correspondence address: School of Languages, Linguistics and Area Studies (LLAS), Faculty of Humanities, Languages and Social Sciences, University of the West of England, Frenchay Campus--Coldharbour Lane, Bristol BS16 1QY, United Kingdom. E-mail:

(2.) For the data, I am adopting the orthography employed by the Trumai people for educational purposes. The special symbols are:

t (IPA: t); t (IPA: t); ' (IPA: ?); ch (IPA: [integral]); tl (IPA: [??]); y (IPA: j); i (IPA: i).

(3.) For practical reasons, la is glossed as 'be standing/stand' throughout this article, but in reality the semantics of this form is slightly more complex. See Section 4.2.1.

(4.) In this article, the terms Figure and Ground- which have been used by various authors, such as Talmy (1985)--will be employed. Figure refers to the entity whose location is being pointed out; Ground refers to the place where the Figure is.

(5.) The Focus/Tense particle ka_in is internally composed of two morphemes (ka + in), but it behaves as a unit. Sometimes ka_in can be reduced to ka without any obvious change in either its syntactic status or its semantics. It can also exhibit some variations in its form, combining with the adverbs de 'already' and huk'an 'still', which appear in the "middle" of the particle: ka_de_in, ka (h)uk'an in. In order to capture the nature of this particle, I represent it with an underscore. There is another Focus/Tense particle in the language, chi-in, which presents similar characteristics. For more details, cf. Guirardello (1999: Ch. 5, Sect. 5.1.1).

(6.) The Allative postposition ita indicates motion towards a Ground, but not necessarily reaching it. Compare these examples:

(i) a. huyat piwda hilaka-ki. Seagull fly village-DAT 'The seagull flew to the village.'

b. huyat piwda hilaka ita. seagull fly village ALLAT 'The seagull flew towards the village (but it did not reach it).'

(7.) Trumai has three Dative markers: -(V)tl, -ki, and -(V)s. The marker is selected depending on the characteristics of the head of the NP that receives it (i.e., pronoun or noun; if noun: human, animal, inanimate). Number also plays a role.

(8.) In my corpus, there are a few instances of the zero-copula construction without the presence of yi. However, when consultants are presented with such examples, they say that yi should be used in them. This being so, I believe that there is a correlation between the absence of the copula and the presence of yi.

(9.) For a more detailed analysis of the configuration of negative existential clauses in Trumai, see Guirardello (1999: Ch. 5, Sect. 5.2.5).

(10.) In Guirardello-Damian (2002), I discuss why the combination [[V.sub.main] Positional] is considered a [V Aux] sequence, rather than a serial verb construction; (cf. Section 3.2 of the mentioned article).

(11.) In order to refer to the position/posture of the O or DAT argument of the clause, a different kind of construction is used:

A O V temporal clause

(i) hai-ts Tata midoxos, tsula-n-es.

1-Erg Tata call lie-3Abs-TEMP

'I am calling Tata, who is lying.'

A O V DAT temporal clause

(ii) hai-ts oke kiti Amati-tl, tsula-n-es.

l-Ergmedicine give Amati-Dat lie-3Abs-TEMP

'I am giving medicine to Amati, who is lying.'

(12.) One could wonder if the forms have evolved from la 'be standing', and if the 'standing' posture would have a special status in the Trumai language (would it be the default posture for some kinds of Figures?). Although the positional la does have some special characteristics, such as the possibility of receiving the middle-voice marker wa- (cf. Guirardello-Damian 2002, Section 3.1), there is no particular reason or clear historical evidence to claim that tsula, pila, and mula evolved from it. And even if they had evolved, it is not obvious why la would be the basis for building these forms (as we will see, la expresses perpendicular orientation, i.e., a Figure-Ground relation. In which way would la contribute to the composition of pila, which is a positional linked to liquid Grounds?). Thus, even though we can notice the similarity in the phonological forms of the positionals mentioned above, it has not been possible to find other kinds of links among them, such as a historical relationship.

(13.) Exceptions to this table: (i) butterflies, which 'stand' when their wings are not open (the wings stay vertically when closed); (ii) parrots sitting on a tree branch: according to some consultants, a parrot on a tree branch should be described as 'sitting' (since its "buttocks" touch the ground); other consultants describe it as 'standing' (given that its body is vertically oriented).

(14.) Actually, some animals with a rigid body can have a variation in their body configurations: their bellies can be upwards. However, this variation is also classified as being the 'lying' posture (i.e., a subtype of it):

(i) te-natu-k ka_in tsul yi chumuchu.

by.means-back-NZR FOC/TENS river.turtle YI lie

'The river turtle is lying on its back.'

(15.) For birds, the use of aha'tsi is not possible (except for parrots, as mentioned in Note 13). Chumuchu 'lie' is the preferred form, probably because when a bird is "seated", an extensive part of its body makes contact with the Ground.

(16.) Aha'tsi seems to be the default position for three-dimensional objects with a base: when there are several three-dimensional objects of the same kind, but in different orientations, aha'tsi is used to describe the general scene (ia). If the speaker wants to present a more accurate description, each object can be mentioned separately (ib):

(i) a. mesa natu-n ka_in katat yi aha'tsi.

table back/top-LOC FOC/TENS bottle YI sit

'The bottles sit on the table.' (3 are in canonical position, 3 are on their sides)

b. inde yi aha'tsi katsi, inde yi chumuchu.

that YI sit be.sitting that YI lie

'Those ones are sitting (i.e., canonical position), those ones are lying.'

(17.) Interestingly, another Brazilian language, Bakairi, also has a grammatical element that codifies information about liquid Ground. Bakairi does not have a positional verb like the Trumai pila, but it has a special postposition used when the Ground is water (Meira p.c.).

(18.) Here, the sock is the Figure, and the foot is the Ground. If the reverse scenario is expressed (i.e., foot is the Figure, sock is the Ground), the positional used is mula:

(i) meya-n ka_in ha pits' mula.

sock-LOC FOC/TENS 1 foot

'My foot is in the sock.'

(19.) In the answers to the picture series, one of the consultants sometimes employed the zero-copula construction. It is interesting to observe that first she used the zero-copula, but afterwards she tried to provide more information about the scene, and then she used a positional verb.

(20.) The positional tsula 'be lying on a high place' also provides information about the Figure, although in a more indirect way: given the semantic overlap between tsula and chumuchu, the Figures that allow the occurrence of only chumuchu can actually accept tsula too. However, this is the only possible alternative for them. Thus, if a specific type of animal or object only allows tsula besides the use of chumuchu, we can easily tell which type of animal or object the Figure is: an animal with a rigid body or a unidimensional object.
Table 2. Animals (13)

Animals with flexible bodies 4 legs

 2 legs (birds)

Animals with rigid bodies
Reptiles chumuchu 'lie'

Animals with flexible bodies la 'stand'
 aha'tsi 'sit'
 chumuchu 'lie'
 la 'stand'
 chumuchu 'lie'
Animals with rigid bodies

Table 3. Ohjects

1 dimension (e.g., ball) 2 dimensions, having an
 axis that can be projected
 from the ground (e.g., stick)

chumuchu 'lie' la 'stand'
 chumuchu 'lie'

1 dimension (e.g., ball) 3 dimensions with a base
 (e.g., bottle, pan)

chumuchu 'lie' aha'tsi 'sit'
 chumuchu 'lie'
COPYRIGHT 2007 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Guirardello-Damian, Raquel
Publication:Linguistics: an interdisciplinary journal of the language sciences
Article Type:Report
Geographic Code:3BRAZ
Date:Sep 1, 2007
Previous Article:'To sit face down'--location and position in Goemai (1).
Next Article:Grounding objects in space and place: locative constructions in Tidore (1).

Related Articles
Locative trigrams in Northern Sotho, preceded by analyses of formative bigrams *.
A Grammar of Mina.
The typology and semantics of locative predicates: posturals, positionals, and other beasts.
Four languages from the lower end of the typology of locative predication *.
'To sit face down'--location and position in Goemai (1).
Grounding objects in space and place: locative constructions in Tidore (1).
Why a folder lies in the basket although it is not lying: the semantics and use of German positional verbs with inanimate Figures (1).
Laz positional verbs: semantics and use with inanimate Figures *.
The coding of topological relations in verbs: the case of Likpe (Sekpele).
Standing divided: dispositionals and locative predications in two Mayan languages *.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters