* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download WORD ORDER AND CONSTITUENT STRUCTURE IN
Lithuanian grammar wikipedia , lookup
Arabic grammar wikipedia , lookup
American Sign Language grammar wikipedia , lookup
Macedonian grammar wikipedia , lookup
Old Irish grammar wikipedia , lookup
Ojibwe grammar wikipedia , lookup
Swedish grammar wikipedia , lookup
Zulu grammar wikipedia , lookup
French grammar wikipedia , lookup
Udmurt grammar wikipedia , lookup
Modern Hebrew grammar wikipedia , lookup
Ancient Greek grammar wikipedia , lookup
Kannada grammar wikipedia , lookup
Old English grammar wikipedia , lookup
Esperanto grammar wikipedia , lookup
Scottish Gaelic grammar wikipedia , lookup
Polish grammar wikipedia , lookup
Navajo grammar wikipedia , lookup
Malay grammar wikipedia , lookup
Portuguese grammar wikipedia , lookup
Yiddish grammar wikipedia , lookup
Chinese grammar wikipedia , lookup
Turkish grammar wikipedia , lookup
Georgian grammar wikipedia , lookup
Lexical semantics wikipedia , lookup
Icelandic grammar wikipedia , lookup
Latin syntax wikipedia , lookup
Serbo-Croatian grammar wikipedia , lookup
English clause syntax wikipedia , lookup
WORD ORDER AND CONSTITUENT * STRUCTURE IN MEKENS Ana Vilacy GALUCIO Ministério da Ciência e Tecnologia/Museu Paraense Emílio Goeldi RESUMO Este artigo procura determinar se existe uma ordem básica de constituintes em Mekens. Usando freqüência textual e outros critérios relevantes, argumenta que SOV deve ser a ordem básica na língua, e analisa outras variantes como sendo o resultado marcado de processos morfossintáticos e/ ou pragmáticos, como cliticização e focalização. ABSTRACT This paper tries to determine whether there is a basic word order in Mekens. Using text frequency and other relevant criteria, it argues that SOV is the basic word order in the language and analyzes other variants as the marked result of morphosyntactic and/or pragmatic processes such as cliticization and focus. PALAVRAS-CHAVE Estrutura constituinte, ordem de palavras, línguas indígenas, Mekens KEY-WORDS Constituent structure, word order, Indigenous languages, Mekens Introduction This paper presents evidence for determining what the basic word order is in Mekens. On the basis of text frequency and other relevant constituent structure tests, I argue that SOV is the basic unmarked word order in Mekens and that other variants found in the language Revista da ABRALIN, vol. 1, no 2, p. 51-73, dezembro de 2002. WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS result from morphosyntactic and/or pragmatic processes such as cliticization and focus. Mekens or SakÆrabiat1 is one of the five surviving languages from the Tupari subfamily (Tupi family), spoken by approximately 23 people in the state of Rondônia, in Northwestern Brazil. It figures among the native Brazilian languages that are in great danger of disappearing (Rodrigues 1993, Moore, forthcoming), since it has fewer than fifty speakers and is, generally, not being learned by the younger generations. The study of the languages of the Tupari family is still very much in its beginnings. There are only a few partial descriptions (Alves 1991, 2002, Braga 1992, 2001, Galucio 1994, 1996, 2001, 2002, Seki 2002). Aside from an article about the phonemic system of Mekens (Hanke et al. 1958), the only other available materials for this language are the outcome of my own research (Galucio ibid). The data for this paper were collected by the author in the period between 1994 and 1996, during four trips to the Área Indígena Rio Mekens in Rondônia. 1 Morphosyntax Summary There is no case marking in Mekens. The noun phrase (NP) grammatical relations of S, A, and O2 are marked through a fixed order of the noun phrases in relation to their respective predicates. The nominal arguments precede the verb, which amounts to saying that SOV is the preferred order of constituents; however, there are grammatical and pragmatic processes that may affect this ordering. In the case of transitive verbs, since both arguments (A and O) precede the verb, their linear order functions as the main criterion in distinguishing them. Thus, in a transitive clause with two full noun phrases, the first NP refers to the subject and the second to the object. Note in the examples (1a-b) below that changing the word order results in changes in the meaning of the clause. 52 ANA VILACY GALUCIO (1) a. Ameko aose so-a-t jaguar/dog man see-Theme-past ‘The jaguar saw the man’ b. aose ameko so-a-t man jaguar/dog see- Theme-past ‘The man saw the jaguar’ There are three types of verb phrases (VP) in Mekens: intransitive, transitive, and non-inflectible or particle verbs. Intransitive VPs are composed of an intransitive verb stem and a person prefix that cross-references number and person of the clause subject. (2) a. o-ib-a-t sekwa 1s-return-Theme-past soon ‘I returned quickly’ ikãõ Dem b. Roque se-er-a i-toop Roque 3c-sleep-Theme 3s-aux+lying+non.past ‘Roque is sleeping’ Transitive VPs are composed of a transitive verb stem and an object. The object may be a noun phrase or a personal prefix, whereas the subject is realized as a full NP (noun or pronoun): (3) a. ameko e-so-a-t dog/jaguar 2s-see-Theme-past ‘The dog saw you’ b. e-iape seisa-t 2s-drink bring+Theme3-past ‘I brought your drink’ 53 õt I WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS Note in the above examples that whereas in the intransitive VPs the personal prefix co-occurs with the subject noun phrase, in the transitive VPs the personal prefix is in complementary distribution with the object NP. In addition, there is a person hierarchy regarding pronominal subjects in transitive clauses with respect to non-deletability. If the subject of a transitive verb is either a first or a second person, the subject pronoun must be present. If it is a third person, the pronoun can be optionally omitted. Therefore, a zero subject marking in a transitive clause is interpreted as a third person singular subject. In the case of example (3b) above, if the pronoun õt ‘I’ were not present, the clause would have an interpretation with a third person singular subject: ‘He/she brought your drink’. This ranking regarding the possibility of deleting a subject marking in transitive clauses does not extend to plural subjects. A few three-argument (ditransitive) verbs occur in Mekens. They show the same inflection paradigm of transitive verbs for subject and direct object. The indirect object is realized in an oblique postpositional phrase. The third type of verb phrase, the uninflected or particle VP, is composed of a verb which does not inflect for person and/or number of its arguments. This class of VP is largely characterized by onomatopoeic verbs but also includes lexical verbs. Uninflected verbs subcategorizing for one argument have only a subject, realized as an NP (pronoun or noun) as in (4a). In the rare cases where uninflected verbs have an explicit second argument, like a notional object, this is either realized in an oblique phrase or omitted, but is never realized as a prefix. The default rule for this type of VP is to have one or both of their arguments omitted under identity with previously mentioned NPs. Third person subjects are also generally omitted. The uninflected VPs are underlined in the examples below: 54 ANA VILACY GALUCIO (4) a. te arop s koboy-koboy ebapita kwep Foc then deer dive-Red emerge climb ‘Then the deer dove, emerged, and came up (again)’ b. pˆp pe=i-pap throw Obl=3s-dead ‘(They) threw away the dead (animal)’ Semantically, this class can be organized into three subgroups: (i) verbs that express activity, including concepts generally expressed by onomatopoeia, (ii) utterance verbs, which signal direct speech in the discourse, and pro-sentence verbs, and (iii) a closed set of stance verbs, such as standing, sitting, and lying. There is a series of arguments internal to the language that distinguishes uninflected verbs from ideophones. The occurrence of this type of verb is a characteristic of other Tupian languages, consistent with data presented in Moore (2002). Mekens has a series of auxiliary verbs used in the progressive aspect which mark person, number, and positional posture of the clause subject through a series of personal prefixes (Galucio 1996b). In intransitive clauses, the personal prefix in the lexical verb cross-references the person and number of the subject marked in the auxiliary. In transitive clauses, the personal prefix in the auxiliary is the only reference to the sentence subject, when it is pronominal. This is clearly shown in the examples below. The pattern of auxiliaries with intransitive (one-argument) verbs and their inflectional paradigm reveals a nominative-accusative pattern since it always marks the clause subject, regardless of the valence of the lexical verb. (5) a. Ameko se-er-a i-toop dog/jaguar 3c-sleep-Theme 3s-aux+lying+non-past. ‘The dog is sleeping’ 55 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS b. terro mõkwa chicha make ‘I am making chicha’ o-koop 1s-aux+mov+non-past. On the other hand, pronominal marking with transitive and intransitive verbs shows an ergative pattern. This pattern is revealed in a complementary distribution between free pronouns and personal prefixes. The prefixes mark the verb’s S and O arguments, as shown in (6a-b), while the pronouns mark the A argument (7a). That distribution becomes more complex once the auxiliary paradigm and other sentence and functional types are considered, but this discussion is beyond the scope of this paper. (6) a. e-er-a-t 2s-sleep-Theme-past ‘(you) slept’ b. e-so-a-t 2s-see-Theme-past ‘He/she saw you’ (7) a. i-so-a-t 3s-see-Theme-past ‘You saw him/her/it’ (e)t) 4 you (sete) he/she/it e) t you 2 SOV as the basic order of constituents Some of the criteria frequently used in the literature to determine the basic order of constituents are textual frequency, descriptive simplicity, morphological marking, and pragmatic value (Mithun 1992). On the basis of these criteria, I analyzed 113 simple and complex sentences extracted from texts in order to determine the basic constituent order in Mekens. The corpus included 62 transi56 ANA VILACY GALUCIO tive clauses, 44 intransitive clauses, and 07 nominal clauses. Among the transitive clauses, only 21 (34%) had both subject and object overtly expressed with an NP. The general pattern for Mekens is to have either one of the core arguments expressed by a verbal prefix or left unmarked, as in the case of a third person singular A argument. This tendency explains the low percentage of overt NPs for A and O. As pointed out by Siewierska (1988:10) ‘transitive clauses containing full nominal participants are uncommon in many languages in which the subject and object are either optionally or obligatorily indicated in the verb.’ In simple transitive sentences with both arguments expressed with an overt NP, the predominant constituent order is Subject-ObjectVerb (SOV) when the subject is nominal and OVS when it is expressed with an independent pronoun. In intransitive clauses, SV is the most frequent order with nominal subjects, while VS is the most common with pronominal subjects. The order SOV/SV was found in 70% of the analyzed clauses with nominal subjects, but only in 29% of the clauses with pronominal subjects; so the corpus of pronominal subject clauses showed 71% of clause final subject (OVS/VS). I had initially entertained the hypothesis of two basic orders in Mekens, one with full subject NPs, and the other with pronominal subjects. The former would be subject initial, and the latter, subject final. This type of distinction has been described for various languages of the world according to Givon (1984:217-18). Nonetheless, a closer observation of all the clause final pronominal subjects revealed that more than 90% of them occur with first and second singular person pronouns–õt and e)t. The other pronouns had SOV/SV as the common order. Thus, there is not a split between nominal and pronominal subjects, but rather a tendency for first and second singular person pronouns to occur in final position following the verb. It is interesting to note that these two pronouns are oscillating between the two orders given above. In all the textual examples that were considered, simple clauses with first and second singular person pro57 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS nouns had final subject, as shown in (8a), whereas in the elicited examples the order varies between SOV/SV and OVS/VS, though the latter is generally preferred. Changing the subject position does not seem to entail any pragmatic or functional difference, though more research is needed on this topic. Examples (9a-b) are elicited clauses. (8) a. eba e-iape seisa-t evid. 2s-drink bring+Theme-past ‘Look, I brought your drink’ (9) a. kwaay õpa-a-t tapir kill-Theme-past ‘I killed the tapir’ õt I õt I b. õt kwaay õpa-a-t I tapir kill-Theme-past ‘I killed the tapir’ The behavior of these pronouns in two independent morphophonological processes presents evidence for analyzing them as clitics as opposed to free pronouns. The first process involves sonorization of the voiceless stops /p, t, k/ when followed by a vowelinitial morpheme: /p, t, k/ become /b, r, g/, respectively. The second process involves nasalization in morpheme boundaries: nasalization may spread from a vowel to a morpheme final voiced stop, causing an alternation between the voiced stop and its nasal counterpart. The spreading can be left or rightwards, and, both morpheme internally or intermorphemically (10a-b), but it does not span word boundaries (10c-d). (10) a./kp + e)po/ tree + ?? ‘tropical creeper’ [kbe)mpo] or 58 [kme)mpo] ANA VILACY GALUCIO b. /o+ane)p+asi/ [oane)mbasi] or 1s-head-pain ‘I have a headache’ c. /opoep # ãkãy)/ [opoebãkãy)] but not 1s-possession burning.coal ‘My burning coal’ d. /Nwe)re)p # k/ [Nwe)re)mb k] but not [oane)masi] *[opoemãkãy)] *[Nweremk] finish water ‘Finish the water’ As we can see in (11a-b), the first person pronouns, õt and e)t, may cause nasalization of the final consonant on a preceding word. Therefore, they do not have the status of a phonologically free word, since nasalization does not occur between word boundaries, as shown above. These clause-final pronouns are a syntactic word, but not a phonological word, that is, they function as postverbal clitics in the language. (11) a. /o+ebasi-r-e)p=õt/ 1s-alive-rel-really=I ‘I am alive’ [webasire)bõn] b. [o+akop=õt] [o-akob-õn] 1s-hot/heat=I ‘I am hot’5 (lit: my heat I (have)) or [webasire)mõn] or [o-akom-õn] The fluctuation in the order of constituents shown in (9a) and (9b) indicates that the reanalysis of õt and e)t as clitics is not yet completed, but it is currently taking place in the language, thus explaining the current alternation between subject initial SOV/SV and subject final OVS/ VS word orders. As is well known, cross-linguistically clitic pronouns are often subject to special positioning rules, which are not necessarily related to their grammatical relations (Comrie 1989:89). 59 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS Therefore, we can maintain that the basic order of constituents in Mekens is subject initial. The subject-final variants that were found in the corpus come either from phonological processes, such as the cliticization of pronouns at the end of verbs, as seen above, or from morphosyntactic and/or pragmatic processes, such as focus and deverbalization. Moreover, there is strong evidence that Mekens is not just a subject-initial, but also a verb-final language. That is, both subject and object precede the verb, resulting in SOV as the basic order. As it was previously mentioned, since there is neither case marking in the nouns nor agreement on the verb, when both NPs precede the verb, the linear order of constituents functions as a decisive criterion in identifying the grammatical function of the NPs. 2.1 The Object (O) + Verb (V) sequence form a constituent In addition to the linear precedence of the object in relation to the verb, the data show that these two constituents form a syntactic phrase in the language. There have been a number of contentions in the typological literature arguing in favor of a cross-linguistic tendency for verb and object to form a syntactic unit (Comrie 1989, Croft 1990). The data from Mekens confirms that tendency by revealing a well-established VP unit. The existence of a verb phrase in Mekens is justified on different syntactic grounds. Evidence for object and verb forming a syntactic unit arises from tests involving the possibility of movement, coordinated structures, and from the fact that nothing may intervene between object and verb. a) Nothing may intervene between the verb and its object. Even when the object is extracted from the position immediately preceding the verb, the OV sequence is maintained through affixation of the 60 ANA VILACY GALUCIO object marker prefix (i-, s-) to the verb, which, then, functions as the grammatical argument. Note in the following examples that the freely ordered particle eba may occur anywhere in the clause, except between object and verb. (12) a. Eba e-iape seisa-t=õt evid 2s-drink bring+Theme-past=I ‘Look, I brought your drink’ b. e-iape seisa-t=õt eba c. *e-iape eba seisa+Theme-t=õt (b) Some processes require the object and verb to move as a constituent to the left of the subject at the beginning of the clause. One such process is focus, as shown in (13a) below: (13) a. kpe siit õ-a-t=te tabisarã te-bõ big.knife small. give-Theme-past=Foc chief he/she-Dat ‘The chief gave a knife to him’ (lit. Give a knife to him is what the chief did) (c) The distribution of adverbial phrases in the established pattern of constituent order is consistent with the hypothesis of object and verb forming a constituent. Four types of adverbial phrases formally marked according to their syntactic function have been identified: locative (N-ese), ablative (N-eri), dative (N-bõ) and associative (N-ese p ). These adverbial phrases are composed of a noun (phrase) and a postposition and function grammatically as obliques. Adverbial phrases, including the time adverbs, may occur in basically any position within the clause, except between verb and object. Structures (14a-c) below are possible, but (14d) is not. 61 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS (14) a. NP S (14b) AdvP VP Subj O Pedro pp s AdvP NP VP V Subj O pp miat Pedro ontemveado matou (14c) S NP VP S pedro s V miat ontem Pedro veado matou *(14d) S AdvP NP O AdvP Subj O V Adv Subj Obj Adv pedro s miat pp Pedros pp V V miat Textual frequency is another strong argument favoring OV as the basic order. Only two of the morphologically unmarked simple clauses of the analyzed corpus did not have the object preceding the verb. The only clauses that can have a notional object following the verb are those in which the object has been syntactically and pragmatically demoted from its argument function and marked as an oblique. At the pragmatic level, the object demotion deemphasizes the role of the object NP, functioning as a backgrounding antipassive, according to the definition given in Foley and Van Valin (1985). Syntactically, it ceases to be the verbal internal argument (object), and patterns with the oblique NPs, marked by an adposition. In such cases, the prefix i- (object marker) appears on the verb, and functions as the O argument. A more comprehensive account of this construction in Mekens is given in Galucio (2002). se-yãã te pe-k (15) a.Poret i-õ-a-t then OM-give-Theme-past 3c-aux+sitting+nonpast foc obl-water ‘Then he gave water (to him)’ 62 ANA VILACY GALUCIO As shown above, the sequence of object and verb forms a syntactic unit. Furthermore, there is an asymmetric relationship between subject and object in relation to the verb. Therefore, instead of a flat structure having both NPs and the verb at the same level (16a), we propose a branching structure (16b) for Mekens, having the subject NP at the same level of the VP, and above the object NP. First Approximation of a phrase structure for Mekens: (16) a. NP *S NP (16b) V S NP Subj Obj VP Subj NP V Obj 2.2 Nominal Clauses confirm SOV as the basic word order The ordering pattern shown in nominal clauses confirm SOV as the basic order of constituents in the language. There is no copula verb in Mekens. Predicative nominal clauses are formed through the juxtaposition of two nominal phrases, as illustrated in (17a) below. (17) a.o-top kwãmõã 1s-father shaman “My father is a shaman” In the analyzed corpus, there were seven nominal sentences, five of which had the predicative NP following the subject NP. That is, they presented the same ordering pattern found in verbal clauses, as illustrated by (18a) below. . The only two examples that showed the reverse order —subject NP following the predicative NP—are transcribed in (19a-b) below. The clause in (19a) has the clitic õt as sub63 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS ject, and, as it has already been shown, this pronoun tends to occur post verbally. The clause in (19b) is a yes/no question, and as a property of this type of clause, the predicative NP, which bears the focus of the question, always occurs in initial position, thus preceding the subject NP. taba i-paba eteet kaat (18) a.(...) paay palm.tree leaf 3s-dry really say ‘(...) the palm tree leaf was really dry, (he) said’ (19) a. Are)p are)bõ o-tag-e)pit kpkat poor-apõõt=õt õt=te then then-really 1s-son-?? tree old-neg+1s=I I=emph ‘Then (he said): As for me, I was not an old tree, my grandson’ b. Are)p sete i-so-a pase pe=õt eyat-same eyat-kera then he/she OM-see-Theme all obl=I 2pl-good/well 2pl-non.assert ‘Then she looked at me, at every one (and asked) ‘Are you okay?’ 2.3 Cases of postverbal NPs The (O)VS ordering was found in 34% of the clauses that had an explicit subject NP different from first and second person pronouns. In general, all these cases can be explained through interaction with other grammatical or pragmatic processes. In the following sections we will consider two of these processes. 2.3.1 Deverbal construction with -pit Mekens has a type of construction in which the focus of the clause is placed on the situation described by the predicate, and the subjectundergoer is not pragmatically relevant or is unknown. This construction is employed, for instance, when the speaker does not know who the undergoer is. This is not to be confused with an anticausative construction, however, for the agentive argument might not be known, but the causative component is clearly implied to be present. Anti64 ANA VILACY GALUCIO causative constructions in Mekens involve a co-referential (reflexive) affix in the inchoative clause, similar to the constructions found in French or Portuguese. The Mekens deverbal construction is characterized as follows: Firstly, a deverbal suffix -pit appears on the verb, deriving into an adjective stem. Secondly, the NP corresponding to the subject in the respective active clause is either omitted or treated as oblique with respect to its distribution within the clause. Thirdly, the verbal phrase may be divided, that is, the object may be separated from the verb, and no additional morphological marking appears. This indicates at least that the object NP of the active clause no longer forms a constituent with the verb. Fourthly, the object NP of the active clause is not marked as an oblique, even when occurring outside the VP. These two latter points indicate that the verb marked with -pit does not subcategorize as a transitive verb. Examples (20a-b) below illustrate this type of construction. (20) a. korakora o-pãrãre pt-ka-a-t chicken 1s-sieve hole-verbz-Theme-past ‘The chicken pierced my sieve’ b. oe)p pt-ka-pit o-pãrãre already hole-verbz-part. 1s-sieve ‘My sieve was pierced’/’My sieve has been pierced’ In previous versions of this paper, I had termed constructions like those in (20b) passive-like constructions motivated by the fact that the subject NP of a corresponding active clause, such as in (20a), presents the behavior of an oblique NP in the clause, being either omitted or freely ordered in the clause. Such properties are typical of oblique NPs, not of core arguments. However, in spite of the fact that the object NP of a corresponding active clause does not form a syntactic unit with the verb in sentences like (20b), 65 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS as explained above, it does not show the morphosyntactic properties of a subject either. These characteristics have motivated my adopting the term passive-like construction, in an attempt to acknowledge both the similarities and differences between this construction and real passives. However, to avoid further misinterpretation, I will simply refer to constructions, such as those in (20b) and (21a), as deverbal constructions with –pit. One of the sentences from my corpus that had the subject NP following the verb (21a) fits the given characteristics of deverbal constructions with -pit in Mekens. The OVS ordering in this case is explained grammatically, by the oblique status conferred to the agentive argument in these constructions; therefore, it is not a counterexample for the analysis of SOV as the basic order. (21) a. kare)p kare)p aisi õ-ã-pit tabisarã then then wife give-Theme-part. chief ‘Then a wife was given by the chief ’/’Then the chief gave (him) a wife’ Regardless of translation, constructions such as those in (20b) and (21a) involve deverbalization of the transitive verb, giving rise to an adjective stem, and freeing the object NP of its VP internal position. A more literal translation of a sentence like (20b) would be ‘my by now pierced (by someone) sieve’, which acknowledges the structure of the deverbal adjective and the notional NP object. Nonetheless, the AN order found in (20b) remains to be explained. The pattern of the language is NA, as seen in (21a), but that question is out of the scope of the present paper and will be discussed in future works. 2.3.2 Focus The term focus has been generally used in the literature to include nonpresupposed asserted new information as well as contrastive information (Payne 1992). Dahlstrom (forthcoming) gives 66 ANA VILACY GALUCIO the following definition of focus found in Lambrecht (1994): “typically new information asserted against the background of a presupposed proposition”. This is the focus definition that we are adopting here. One example of a construction that is generally assumed crosslinguistically to involve focus is the question-word questions in which the information in the proposition is presupposed, except for the object of the interrogative word. Thus, the question word normally bears the focus. Focus constructions in Mekens are syntactically and morphologically marked. First, the focused element occurs at the beginning of the clause. Secondly, the clitic te appears to the right of the constituent in focus position, and before the rest of the clause. In the following sentence, the focus constituent is underlined. (22) a.apo=te who=foc o-tek 1s-house poga-b-eg-a open-Nmlzr-house-Theme ke say i-koop ikãõ tete 3s-aux+mov that time ?? ‘“Who opened my house?” She was saying at that time’ All question-word questions in Mekens are realized in this manner, that is, the interrogative word occurs in focus position at the beginning of the clause, preceding the focus particle. In the same way, in an answer to such questions, the non-presupposed new information generally comes in a focus construction. Any constituent in a Mekens clause can bear the focus feature: subject, object, verb, verbal phrase, and oblique NPs. Therefore, focus constructions may give rise to alterations in the basic order of constituents, since any focus element will necessarily occur at the beginning of the clause. The following example illustrates this point. 67 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS (23) a. serek=te kpmãkãy kweõp =te split=foc floor climb+neg=foc ‘The floor split and the deer sank’ s deer Note that, in the two clauses in (23a), the subject occurs after the verb, which is clause initial and followed by the clitic te. This particular example is extracted from a text in which the mythological hero, Pasiare, is preparing himself to cause a flood. He uses a deer to verify whether or not he is ready for the task. He throws the deer into the water to see the depth, but since there is not much water at that point, the deer struggled for a short moment and rose back easily. This is said in the regular order SV, as shown in (24a). Then, the second time, after having ingested a great quantity of water, Pasiare experiments with his powers again, by expelling a lot of water and throwing the deer into it. This time, the water level is really high causing the floor to split into two sides and the deer to sink completely. It is this contrast between the first and the second time the deer is thrown into water that is expressed by the focus construction in (23a) above. (24) a. Arop s koboy-koboy ebapi-ta kwep then/that deer dive-red front-?(emerge) climb+up “Then the deer dove dove (into the water) and surfaced back” It is worth mentioning that 85% of all the corpus examples of reverse word order—subject NP following the verbal phrase (VP)— are cases in which the VP is clause initial because it bears the focus feature. Thus, focus constructions in Mekens are evidence for the SOV basic word order precisely because they deviate from it, giving rise to an OVS order. Therefore, placing a constituent in focus gives rise to a syntactically marked structure, which deviates from the SOV basic word order, as in (23). Nonetheless, there is still the question of how the focus element 68 ANA VILACY GALUCIO fits the phrase structure of the sentence. There are at least two possible hypotheses: the first is shown in (25a) where the focus element is a daughter of the S node and stands in a symmetrical relation to the others elements in the remainder of the clause. The second is (25b) where the focus is a sister of S. The former structure would imply that the non-focus elements do not form a constituent, whereas the latter has the opposite implication. (25a) *S Foc SN (25b) S’ SV Foc SN S SV Since there are coordinate sentences in Mekens in which a sole element in focus position refers simultaneously to anaphoric positions in the two S’s, the second structure (25b) seems to be the right one. Considering that in this structure, the remainder of the sentence to the right of the focus constituent forms a syntactic unit, the existence of coordinated structures that share exactly the focus element is predictable, since there is nothing in the phrase structure blocking such constructions. On the other hand, in a symmetric structure like (25a), in which the elements to the right of the focus do not form a constituent, such coordinate structures would not be allowed. Note in the example below that the focus element s is coreferential to the object of both verbs. (26) a. s tõe=te Chico i-mi Rosalina i-perop-ka-t deer non-afirm=foc Chico OM-kill Rosalina OM-food-TR-past ‘It looks like it was a deer that Chico killed and Rosalina cooked’ 3 Final Remarks This paper has attempted to show that SOV is the basic unmarked order of constituents in Mekens and that other processes of the lan69 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS guage work to produce the alternative orderings identified in the corpus. These processes include the on-going cliticization phenomenon of first and second person pronouns, denominalization of the verb associated with de-emphasis on the role of the subject (deverbal constructions with –pit), and placement of the direct object in the initial focus position with subsequent fronting of the constituent in question. The analysis proposed in this paper is a contribution to the typology of basic word order and possible pragmatic permutations on the syntax of clauses in the languages of the Tupari family. Abbreviations 1s: first person singular; 2s: second person singular; 3c: third person coreferential; 3s:third person singular; 2pl: second person plural; aux: auxiliary; aux+mov: auxiliary+moving; Dat: dative; Dem: demonstrative; Emph: emphasis; Evid: evidential; Foc: focus; Neg: negation; Nmlzr: nominalizer; Non.-Afor,: non-afirmative; Non.Assert: nonassertive; NP: noun phrase; Obl: oblique; OM: object marker; Part: participle; Red: reduplication; TR: transitivizer; Verbz: verbalizer; VP: verbal phrase. Recebido em References ALVES, Poliana M.. 1991. Análise fonológica preliminar da língua Tupari. Dissertação de mestrado, Universidade de Brasília, DF. ____. 2001. Flexão Relacional em Tupari e em Tupi-Guarani. In Cabral e Rodrigues (org.) Línguas brasileiras: fonologia, gramática e história, Atas do I Encontro Internacional do Grupo de Trabalho sobre línguas indígenas da ANPOLL, tomo I :269-273, editora universitária UFPA:Belém, 2002. 70 ANA VILACY GALUCIO BRAGA, A. 1992. A fonologia segmental e aspectos morfofonológicos da língua Makurap (Tupi). Dissertação de mestrado, Universidade Estadual de Campinas, SP. ____. 2001. A Estrutura Oracional, o Tópico e a Ordem dos Constituintes em Makurap. Trabalho apresentado no I Encontro Internacional do GTLI da ANPOLL/UFPA, Belém, October, 2001. COMRIE, Bernard. 1989. Language Universals and Linguistic Typology. nd 2 edition. Chicago: University of Chicago Press. CROFT, William. 1990. Typology and Universals. Cambridge: Cambridge University Press. DAHLSTROM, Amy. (no prelo) Morphology and Syntax of the Fox (Mesquakie) Language. DIXON, Robert M.W. 1994. Ergativity. Cambridge University Press. FOLEY and R.VAN VALIN JR. 1985. Information packing in the clause. In T. Shopen (ed.), Language typology and syntactic description, vol 1:282-364. Cambridge: Cambridge University Press. GALUCIO, A. Vilacy. 1994. Fonologia Segmental da Língua Mekens. In Anais do IX Encontro Nacional da Associação Nacional de Pós-Graduação em Letras e Lingüística. Caxambu, MG, vol 1, p. 988-997. _____. 1996a. Mekens Phonology. University of Chicago, IL. (ms.) _____. 1996b. Mekens Syntax: a preliminary survey. Dissertação de Mestrado. University of Chicago, IL. _____. 2001. The Morphosyntax of Mekens (Tupi). Tese de Doutorado, University of Chicago, IL. _____. 2002. O prefixo i- em Tupi: morfema antipassivo vs. marcador pronominal incorporado. In Cabral e Rodrigues (org.) línguas brasileiras: fonologia, gramática e história, Atas do I Encontro Internacional do Grupo de Trabalho sobre línguas indígenas da ANPOLL, tomo I :274-287, editora universitária UFPA:Belém, 2002. GIVON, T. 1984. Syntax: A functional typological introduction. Amsterdam:John Benjamins. HANKE, W., M. Swadesh, and A. Rodrigues. 1958. Notas da fonologia Mekens. In Miscellanea Paul Rivet Octogenario. Dicata, vol. 2: 71 WORD ORDER AND CONSTITUENT STRUCTURE IN MEKENS 187-217. México. LAMBRECHT, Knud. (1994). Information Structure and Sentence Form: Topic, Focus, and the Mental Representation of Discourse Referents. Cambridge:Cambridge University Press. MITHUN, M. (1992). “Is basic word order universal?” In Payne, ed. Pragmatics of word order flexibility. 15-61. MOORE, D. 2002. Verbos sem flexão. In Cabral e Rodrigues (org.) Línguas brasileiras: fonologia, gramática e história, Atas do I Encontro Internacional do Grupo de Trabalho sobre línguas indígenas da ANPOLL, tomo I :139-150, editora universitária UFPA:Belém, 2002. ________.(no prelo). Endangered Languages of Lowland Tropical South America. Linguistic Diversity Endangered, ed. por Matthias Brenzinger, Mouton. PAYNE, D. 1992. “Nonidenfiable information and pragmatic order rules in ‘O’odham”. In Payne, ed. Pragmatics of word order flexibility. 137-166. SEKI, L. 2002. Aspectos da Morfossintaxe do Nome em Tuparí. In Cabral e Rodrigues (org.) Línguas brasileiras: fonologia, gramática e história, Atas do I Encontro Internacional do Grupo de Trabalho sobre línguas indígenas da ANPOLL, tomo I :274-287, editora universitária UFPA:Belém, 2002. SIEWIERSLA, A. 1988. Word order rules. London: Croom Helm. Endnotes * A preliminary version of this paper was presented at the 1996 Conference on American Indian Languages, sponsored by the Society for the Study of the Indigenous Languages of the Americas (SSILA) at the 95th Annual Meeting of the American Anthropological Association, in San Francisco, California, November, 1996. I would like to thank Amy Dahlstrom, Dennis Moore and two anonymous reviewers for comments and suggestions on earlier versions of this paper, and Chris Corcoran for helping with editing and proofreading. 1 Mekens is the name generally found in the literature, but the group has chosen Sakrabiat as the self-identifying name. 72 ANA VILACY GALUCIO 2 S, O, and A are used here in the sense of Dixon (1994). S is the subject of an intransitive verb, A and O are, respectively, subject and object of a transitive verb. 3 In this example, the final vowel of the verb stem –a fuses with the theme vowel –a. 4 The parenthesis in 6a indicates that a pronoun may optionally co-occur with the personal prefix on the intransitive verb. In these cases, the pronoun may be omitted, whereas the prefix may not. In 6b the parenthesis indicates the possibility of deletion of a third person subject in transitive clauses, regulated by the person hierarchy explained in section 1 above. 5 The syntactic structure of clauses like those in (11a-b) is that of a predicative clause formed by two NP’s — an adjective stem modified by a personal prefix and a subject pronoun. Although they are on the surface similar to an intransitive clause with pronoun doubling, such as those shown in (6a-b) above, they have different morphological and syntactic structures (Galucio 2001). Those predicative clauses based on an adjective stem are formed in that manner. 73