Insights into the syntax of Iatmul, a language of Papua New Guinea

University of Regensburg 27 September 2011 Insights into the syntax of Iatmul, a language of Papua New Guinea Gerd Jendraschek University of Regensburg CONTENTS 1 The language ................................................................................................................ 2 2 Switch reference........................................................................................................... 2 2.1 Form and functions of switch-reference marked clauses ................................................... 2 2.2 Tail-head linkage ............................................................................................................... 3 2.2.1 2.2.2 Tail-head linkage as a referent-tracking strategy ...............................................................3 Tail-head linkage as a coordination strategy......................................................................4 2.3 Non-canonical switch reference in clause-chains .............................................................. 5 2.3.1 2.3.2 Clause skipping..................................................................................................................5 Centre-embedded clauses ..................................................................................................6 2.4 From switch reference chains to complex predicates ........................................................ 6 2.5 Lexicalization of non-finite forms ..................................................................................... 7 3 Focus constructions...................................................................................................... 7 3.1 3.2 3.3 3.4 3.5 Subject and non-subject focus declaratives........................................................................ 7 Origin of focus constructions............................................................................................. 8 Obligatory focus in content questions................................................................................ 9 Semantics and pragmatics.................................................................................................. 9 Focus relative clauses....................................................................................................... 10 3.5.1 3.5.2 3.5.3 4 From desubordination to resubordination ........................................................................10 Structural properties of focus relative clauses .................................................................11 Semantics and pragmatics................................................................................................12 Valency and transitivity ............................................................................................ 13 4.1 Morphological transitivity................................................................................................ 13 4.2 Ambitransitivity ............................................................................................................... 14 4.3 Multi-verb predicates ....................................................................................................... 14 1 The language • Iatmul is a Papuan language of the Ndu family, spoken in the East Sepik Province of Papua New Guinea. • Approx. 40,000 speakers; 30 villages along the Sepik River, half of the population has migrated to towns, shift towards Tok Pisin with people under the age of 25. • The morphosyntax of Iatmul is moderately synthetic and mildly fusional, with bound morphology being predominantly suffixing. • Basic constituent order is AOV/SV, and subjects (S/A) appear unmarked for case and are cross-referenced on the verb. • Data from my own fieldwork of about 14 months, distributed over four trips taking place between May 2005 and December 2008; I spent most of my time in Koloku village (also spelled Korogo or Korogu), where Western Iatmul is spoken, with some influence from the neighbouring central dialect. Figure 1. Papua New Guinea Figure 2. Sepik River 2 Switch reference 2.1 Form and functions of switch-reference marked clauses • Switch-reference systems ‘indicate coreference or disjoint reference between the subjects of adjacent, syntactically related clauses’ (Stirling 2001: 14). • When an S/R-clause and the superordinate (=main clause) clause share their subject referent, a non-finite verb form must be used in the subordinate clause. (1) ki’ki’da kuk-ka yaki ki’-li’-li’1 [food do-DEP] [tobacco eat-IPFV-3SG.F] ‘She was preparing food and smoking’ ‘She was smoking while preparing food’ • Person-marking in both clauses (plus the absence of tense-marking and non-final intonation) indicates disjoint reference. The subject-switch is not marked by a dedicated morpheme (2). 1 Iatmul data are presented in the orthography used in Jendraschek (2008). The major differences from the International Phonetic Alphabet are as follows (allophones in complementary distribution are separated by a comma, those in free variation by a tilde): a [ ], [ɑ]~[ɒ]; a’ [ ʔ]~[ ]̰ ~[ ˑ]; a’i [ ̰ ̯]~[ ̯]; aa [ ː]; b [mb]; d [nd]; g [ ɡ]; i’ [ɨ]; j [nd͡ʒ]; k’ [k ]; l [l]~[ɾ]; n’ [ ]; ny [ɲ]; tt [t]; v [ ]. 2 (2) ki’ki’da kut-ti’-li’ yaki ki’-li’-li’ [food do-IPFV-3SG.F] [tobacco eat- IPFV-3SG.F] ‘(While) shei was preparing food, she*i/j was smoking’ • The lack of tense is not obvious in a realis context, as realis is zero-marked. In (3), however, the subordinate clause has irrealis reference, but tense is only marked on the superordinate clause. (3) Pius avla but-di’ wuk-kiya-mi’n [Pius self tell-3SG.M] [hear-IRR-2SG.M] ‘(When) Pius himself tells you, you’ll hear it’ • Lexicalization path from S/R verb forms to adverbs: In (4), the verb pi’li’ ‘run’ yields the adverb pi’li’ka ‘fast’; the lexicalization can be seen from the fact that the sentence would normally not be interpreted as ‘don’t talk while running a lot’. (4) ni’man pi’li’-ka wana gabu-m-ay-a big run-DEP SHOULD.NOT speak-2SG.M-IRR-SR ‘don’t speak too fast’ 2.2 Tail-head linkage 2.2.1 Tail-head linkage as a referent-tracking strategy • Tail-head linkage (THL) is ‘a way to connect clause chains in which the last clause of a chain is partially or completely repeated in the first clause of the next chain’ (de Vries 2005: 363). (5) 1.a 1.b 1.c 2.a 2.b 2.c Di’-kat yi-ka 3SG.M-DAT go-DEP (The dog) going to him SUBJECT = SUBJECT = ‘dog’ ‘boy’ SS ta’bak yi-ka hand go-DEP touching (him) SS di’n-a gusa-maan-ba kut-di’. 3SG.M-GEN paddle-leg-LOC touch-3SG.M FIN he touched the back of his leg. Kut-di’ touch-3SG.M He (=the dog) touched TAILHEAD LINKAGE DS di’-kak vi’-di’ 3SG.M-DAT see-3SG.M he (=the boy) saw him (=the dog) DS avayabi’ pi’li’-ka wakwai-ka-di’. path run-DEP advance-PRS-3SG.M he (=the dog) advanced running to the path SS-FIN 3 2.2.2 Tail-head linkage as a coordination strategy • In Iatmul, there is no native class of clause-linking conjunctions. • The observation that Iatmul has borrowed the Tok Pisin conjunctions o ‘or’ and tasol ‘but’ indicates that speakers perceived the impossibility to coordinate independent clauses as a gap in Iatmul grammar. • The absence of coordinating devices has been reported for other clause-chaining languages (see e.g. Genetti 2005: 51; Coupe 2007: 416). • The recapitulation clauses wasodi’and wadi’ are reminiscent of switch-reference markers linking independent sentences, as they appear as the final constituent in a. and d., where they are ‘attached’ to the finite predicate, rather than as an initial element of b. and e. (6) a. wuba wa-so-ka-di’ ↗wa-so-di’ (1.4) [D3:LOC say-UP-PRS-3SG.M] [say-UP-3SG.M] ‘then he shouted up, (when) he shouted up, b. ↘ana wuk-ka-di. (.8) [NEG hear-PRS-3PL] they didn’t hear c. Ki’gi ak (.3) nak li’vi’ wa-so-ka-di’: [D1.NR D2:F other time say-UP-PRS-3SG.M] he shouted up again d Uuu! wa-a-di’ ↗wa-di’ (.6) [uuu say-PRS-3SG.M] [say-3SG.M] uuu, he said, (when) he said, e. ↘ana wuk-ka-di [NEG hear-PRS-3PL] they didn’t hear’ ‘He shouted up to them, but they didn’t hear him. So he shouted up again, uuu, he shouted, but they didn’t hear him.’ • Figure 3 compares linkage with THL and the coordinating conjunction ‘but’. Figure 3. THL and conjunction compared CLAUSE 1 LINKING DEVICE: THL/CONJUNCTION CLAUSE 2 wa-so-ka-di’ wa-so-di’ ana wuk-ka-di [say-UP-PRS-3SG.M] [say-UP-3SG.M] [NEG hear-PRS-3PL] ‘he shouted up but they didn’t hear’ 4 2.3 Non-canonical switch reference in clause-chains 2.3.1 Clause skipping • A subordinate clause can split up a clause-chain or even a clause. (7) a. si’lak-ka kwakla-laa, break-DEP leave-CONSEC2 having finished breaking it b. ki’ta vali’-laa one bite-CONSEC having bitten into one c. sava-li’-li’ watch-IPFV-3SG.F (while) she was watching <-- skipped clause d. ki’-laa, eat-CONSEC having eaten e. nak ki’ta vali’-laa other one bite-CONSEC having bitten into another one f. yi-ka go-DEP going g. kwi-ka-di’ li’l-ak give-PRS-3SG.M 3SG.F-ALL he gives it to her a. ‘When he had broken it (= a bamboo filled with a bandicoot), b. he bit into one piece, c. and as she was watching, d. ate it, e. then bit into another piece, f. and went g. and gave it to her’ • Clause c presents background information and is therefore not part of the sequential chain. • For the purposes of reference-tracking, this clause is ‘skipped’ (cf. Huang 2000: 292). 2 kwakla ‘leave’ is here used as a phasal verb in a complex predicate 5 2.3.2 Centre-embedded clauses • A subordinate clause presenting background information can further be inserted between clause-initial noun phrases and the controlling verb. • In (8), clause a ‘you and me arrive’ is split up by clause b: The subject is separated from the verb. (8) a. mi’n-okwi wun-okwi [2SG.M-COM 1SG-COM you and me b. wun-kat kup-ba [1SG-DAT behind-LOC you followed me ya-mi’n come-2SG.M] <-- centre-embedded clause a. ya-a-li come-PRS-1DU] we arrive ‘you and me have arrived, with you following me’ 2.4 From switch reference chains to complex predicates • In (9), the construction -taa li’ ‘-CONSEC stay’ expresses resultative aspect. (9) […] di’ wan Waji’mauk kla-da-laa maan-ba 3SG.M D3.SG.M water.spirit get-down-CONSEC foot-LOC ‘the Water Spirit grabbed her leg and pulled her down.’ Kut-taa li’-di’ hold-CONSEC stay-3SG.M yi-ka go-DEP kut-di’. hold-3SG.M nyaamun tat-ta brother be.first-CONSEC vali’kai-laa ti’-di’. cross-CONSEC stay-3SG.M ‘He had grabbed it, [but] her brother had crossed first.’ • The construction -ka li’ ‘-DEP stay’ serves to express progressive aspect. (10) ni’man yalavi’k-ka di-kak big think-DEP 3PL-DAT ‘I’m thinking a lot about them’ ti’-ka-wun stay-PRS-1SG When li’ ‘stay’ is directly bound to the stem as in (11), it is fully grammaticalized and has lost the stance implications it has in (12). (11) gabu-li’-ni’n speak-IPFV-1PL ‘we were talking’ (12) gabu-ka li’-ni’n speak-DEP stay-1PL ‘we stayed (there) talking’ ‘we were staying (there) and talking’ 6 2.5 Lexicalization of non-finite forms • A clear example of lexicalization is the fusion of the non-finite forms of the directional verbs in (13) with the verb yi ‘go’. (13) wak-ka + yi > wakkai ‘ascend’ ascend-DEP go da-a fall-DEP + yi > daai ‘descend’ go wula + yi > wulaai ‘enter’ enter(DEP) go waku-ka exit-DEP + yi > wakwai ‘come forward, come out’ go 3 Focus constructions 3.1 Subject and non-subject focus declaratives • Comparison of neutral, subject focus, and non-subject focus declarative. Neutral declarative (14) Joachim yuwisi’k kut-di’ Joachim rice do-3SG.M ‘Joachim has cooked rice’ • In (15), the focus marker is the suffix -a. The verb form contains no cross-reference marker, but is also suffixed by a marker -a. Subject focus declarative (15) Joachim-a yuwisi’k kuk’-a Joachim-FOC rice do-FOC ‘Joachim has cooked rice’ • • A subject focus utterance as in (15) would be used, for example, as an answer to the question ‘Who cooked the rice?’, where the question word is in focus. The third construction to be discussed here involves focusing of a non-subject noun phrase. The non-subject focus counterpart of (14) and (15) is presented in (16). Non-subject focus declarative (16) Joachim yuwisi’k’-a kut-d-a Joachim rice-FOC do-3SG.M-FOC ‘Joachim has cooked rice’ • The verb form has a cross-reference marker for the subject, followed again by a focus suffix -a. A typical discourse context for (16) would be as an answer to the question ‘What did Joachim cook?’. 7 3.2 Origin of focus constructions • • • The focus constructions have arisen from clefts. In (15) and (16), the focused noun phrases have third-person singular reference, which explains their uniform marking with -a. In contrast, the focused noun phrase in (17) has second-person singular feminine reference, which is expressed by the pronominal marker -nyi’n on the question word kada ‘who’. (17) kada-nyi’n daai-a who-2SG.F descend-FOC ‘Who (are you who) went down?’ • • (17) could be used in the situation where the speaker knows that a female person has just left the house (e.g. because there were only women in the house), but does not know the identity or name of that person. The focused noun phrase kadanyi’n is formally identical to an independent clause, meaning ‘who are you?’, as suggested in the English translation. Subject relative clause (18) yuwisi’k kuk’-a nyaan [rice do-SR] person ‘the person who has cooked rice’ Non-subject relative clause (19) a. Joachim kut-d-a yuwisi’k [Joachim do-3SG.M-SR] rice ‘the rice that Joachim has cooked’ b. kut-d-a yuwisi’k [do-3SG.M-SR] rice ‘the rice that he has cooked’ • • The verb forms used in subject focus clauses are thus the same as those used in subject relative clauses, and those of non-subject focus clauses are identical with those of non-subject relative clauses. The second line in (20) reflects the functions in the target construction, while the line below it reflects its biclausal origin as a cleft. Subject focus question (20) 1. Joachim-a 2. Joachim-FOC 3. [Joachim-3SG.M] 4. ‘[It is Joachim] 5. [focus main clause] 6. [clefted constituent] yuwisi’k kuk’-a rice do-FOC [rice do-SR] [that has cooked rice].’ [extrafocal dependent clause] [content clause] 8 3.3 Obligatory focus in content questions • Whereas the focus constructions are pragmatically conditioned in the declaratives (15) and (16), the corresponding content questions involving the question words kada ‘who’ and mi’da ‘what’ in the relations of subject and direct object functions do not have a neutral alternative. Subject focus question (21) kada-na yuwisi’k kuk’-a? who-3SG.M rice do-SR ‘Who has cooked rice?’ Non-subject focus question (22) Joachim mi’da-na kut-d-a? Joachim what-FOC do-3SG.M-FOC ‘What has Joachim done?’ • • Unlike common nouns, which take the endings -a (masculine) and -ak (feminine), the corresponding endings on kada are -na and -lak; since mi’da ‘what’ refers only to inanimate entities, its focused form is always mi’dana. These allomorphs are most likely the diachronic result of reinforcement. addition of simple markers reduction of identical vowels reinforcement segmental simplification neutralization in coda ‘who(M.)?’ ‘who(F.)?’ kada-an kada-at kadan kadat kadan-an kadat-at kadalat kadana – kadalak Table 1. From simple to double focus marking 3.4 Semantics and pragmatics • • As the cleft undergoes grammaticalization, ‘the difference between focus and presupposition can be smoothed out’ (Lehmann 2008: 213). Since the subject focus clause consists of only two main parts, the subject and the verb phrase, there are only two possible interpretations: either the subject is in focus, as exemplified by the contrastive context of (23), or the utterance is thetic (24). Focus marked on the subject: argument scope (23) gul-aya wun-a da kuk’-a 2PL-3SG 1SG-GEN thing do/hold-SR o ki’di’ki ni’ba-di wun-a da kuk’-a or D1.PL:NR people-3PL 1SG-GEN thing do/hold-SR ‘Have you taken my things or have those people taken my things?’ Focus marked on the subject: clausal scope (24) vaala-di ya-a canoe-3PL come-PRS:SR ‘There are canoes coming.’ (said upon seeing canoes in the distance) 9 • For a habitual reading, the neutral counterpart in (25) has to be used. Neutral construction (25) vaala ya-li’-ka-di canoe come-IPFV-PRS-3PL ‘Canoes are coming.’ (could be used to express that the place can be reached by canoe) • • The focus constructions with non-argument scope are thus used for utterances particularly relevant for the speech situation. In (26), an argument focus interpretation is pragmatically implausible, as the washing in or with water is not contrasted to washing in or with something else. (26) gu-a yaaku-li’-m-a? water-3SG wash-IPFV-2SG.M-SR ‘Are you bathing?’ awa, gu-a yaaku-li’-w-a. yes water-3SG wash-IPFV-1SG-SR ‘Yes, I’m bathing.’ 3.5 Focus relative clauses 3.5.1 From desubordination to resubordination • The desubordinated predicate in the focus construction of Iatmul can be ‘resubordinated’. This occurs when the cleft is transformed into a new type of relative clause. • The resulting option to have focus-marking in relative clauses is cross-linguistically unusual; in English, one cannot obtain a construction like *That’s the rat that it’s the cat that chased (Schachter 1973: 21). • Example (27) illustrates the source construction, with the underlying cleft structure, and its two possible translations into English. Independent clause with subject focus (27) Joachim-a yuwisi’k kuk’-a [Joachim-3SG]FOC [rice do-SR]EXTRAFOCAL CLAUSE ‘Joachim has cooked rice’ ‘It is Joachim who has cooked (the) rice’ • In (28), the direct object noun phrase yuwisi’k ‘rice’ has been ‘taken out’ of the extrafocal clause and becomes the head noun in a focus relative construction. Non-subject relative clause with subject focus: (28) Joachim kuk’-a yuwisi’k ki’-li’-ka-wun [[Joachim do-SR]REL [rice]HEAD]NP [eat-IPFV-PRS-1SG]PRED] ‘I’m eating the rice that Joachim has cooked’ 10 3.5.2 Structural properties of focus relative clauses • There are two important structural differences between the focus relative clause in (28) and its neutral counterpart in (29): o First, in the focus relative clause, there is no person-number marking on the verb (kuk’-a ‘do-SR’ vs. kut-d-a ‘do-3SG.M-SR’). o Second, an overt subject noun phrase is required in the focus relative clause, whereas it is optional in the neutral relative clause. Non-subject neutral relative clause (29) (Joachim) kut-d-a yuwisi’k Joachim do-3SG.M-SR rice ‘the rice Joachim/he has cooked’ • The reanalysis from (27) to (28) further involves the following morphosyntactic changes: o The marking on the focused noun phrase is lost (Joachim instead of Joachim-a), except on pronouns. The former main clause of the construction becomes indistinguishable from an ordinary noun phrase. o The erstwhile relative clause of the cleft comes to modify a following noun phrase. The relative clause verb form, which had been desubordinated in the (largely) monoclausal focus declarative sentence, is thereby resubordinated as the predicate of a new type of subordinate (modifying) relative clause. o The referential head of the relative clause is not the preceding focused noun phrase (Joachim), but the following noun phrase (yuwisi’k). This ‘resubordination’ of the verb form kuk’a is the last stage in the diachronic unification process of the clefted constituent with the extrafocal clause. • Pronouns do retain the marking they have in the independent focus clause. • Having lost its status as a predicate marker, the marking on the pronouns is a morphological fossil. Independent focus clause (30) din-aya ya-li’-ka 3PL-3SG come-IPFV-PRS:SR ‘That’s them coming.’ Focus relative clause (31) din-aya ya-li’-ka kava ana-di’ 3PL-3SG come-IPFV-PRS:SR place NEG-3SG.M ‘This is not a place for them to come.’ • In the new construction, the verb is the head and governs the preceding constituent as its subject. It further specifies that this subject has to be an overt noun phrase, and finally, if the subject is pronominal, that it must appear in a particular form, ending in -aya. • The verb plus its subject together constitute a minimal focus relative clause, which modifies a following constituent, the head of the relative construction. 11 3.5.3 Semantics and pragmatics • The focus relative clause is chosen if its subject is in contrastive focus. • However, there is often no focus involved at all. The focus construction is then equivalent to a neutral relative clause, as long as the subject referent, i.e. the actor, is high on the nominal hierarchy, and has control over the situation. • This is the case in (32) vs. (33), where the subject can either be expressed on the verb (neutral relative clause) or as a pronoun (focus relative clause). Neutral relative clause (32) ki’-w-a yuwisi’k apman’-a eat-1SG-SR rice good-3SG ‘the rice I ate was good’ Focus relative clause (33) wun-aya k-a yuwisi’k 1SG-3SG eat-SR rice ‘the rice I ate was good’ apman’-a good-3SG • The situation is different with subject referents low on the nominal hierarchy, where crossreferencing on the verb is dispreferred if the subject noun phrase is overt. • This means that a non-human actor is either expressed on the verb (34), or by a noun phrase (35), but not both (36): (34) Wan kabai baai-a. Vaali’-d-a du laba kiya-di’. D3.SG.M snake death.adder. bite-3SG.M-SR man already die-3SG.M ‘The snake was a death adder. Then man it had bitten has already died.’ (35) Kabai vaal-a du kiya-di’ snake bite-SR man die-3SG.M ‘The man that was bitten by a/the snake has died.’ (36) ?? Kabai vaali’-d-a du laba kiya-di’. snake bite-3SG.M-SR man already die-3SG.M ‘The man which a snake had bitten has already died.’ • Note that the rejection of (36) is not categorical, but only a matter of preferring (35) over (36). This shows that the choice of one of the two constructions is determined by pragmatic and semantic factors. • (37) has a human actor controlling the situation, whereas (38) has an inanimate force as its subject referent, which lacks volition. Again, (39), which lacks cross-reference, is preferred over (38). (37) Joachim viya-d-a maatnyan gla-a li’-ka-di’ Joachim hit-3SG.M-SR child cry-DEP stay-PRS-3SG.M ‘the child Joachim has beaten is crying’ (38) ?? mi viya-d-a maatnyan gla-a li’-ka-di’ wood hit-3SG.M-SR child cry-DEP stay-PRS-3SG.M ‘the child the tree hit is crying.’ 12 (39) mi viya maatnyan gla-a li’-ka-di’ wood hit:SR child cry-DEP stay-PRS-3SG.M ‘the child which was hit by a tree is crying.’ • When I asked what was wrong with (38), one consultant replied that a tree did not have hands, so how could it hit anyone? • However, the corresponding main clause the tree hit the child was accepted without problems. • Likewise with other examples I tested, cross-referencing in the relative clause was usually more acceptable with volitional agents, whereas the construction in (39) is used for backgrounding the actor, and is therefore preferred with non-human agents. 4 Valency and transitivity 4.1 Morphological transitivity • There are only a few Iatmul verbs where the morphological make-up correlates with transitivity. These verbs contain fossilized manner prefixes. prefix kVvaisi’vi’tV-~lV- instrument hand foot long object by hitting by itself origin ?kuk ‘hold, touch’ va’i ‘step on’ si’ ‘shoot, stab, poke’ ?vi’k ‘cut’; ?viya ‘hit’ ti’~li’ ‘be, stay’ Table 2. Manner affixes: meaning and origin • What is relevant here is that the first four prefixes yield transitive stems, while only the last one yields intransitive stems. This means that manner affixes can be used to form intransitive and transitive versions of the same verb root. The root itself may be a free form or always bound. intransitive: (40) a. balaku b. li’balaku ‘roll’ ‘turn around’ transitive: (41) a. kubalaku ‘turn around (sth flat, e.g. fish); translate’ b. si’balaku ‘turn around (sth round, e.g. a log)’ c. vi’balaku ‘roll’ (tr.) (42) li’-balaku-ka kwa-a-li’ lie-PRS-3SG.F ‘she turns around in her sleep’ BY.ITSELF-roll-DEP 13 • In the case of tamak vs. kemak ‘stop’, the root -(a)mak cannot appear by itself and therefore has only etymological relevance. The fused prefixes have no semantic content, so that we have here an ideal matching of morphological make-up with transitivity, tamak being the intransitive counterpart of transitive kemak. (43) mi’n-kak kemat-j-ay-a-n, tamak-kiya-mi’n 2SG.M-DAT stop:TR-3PL-IRR-SR-NR stop:INTR-IRR-2SG.M ‘if they stop you, you will stop’ 4.2 Ambitransitivity • Iatmul has a class of S=O ambitransitive verbs. Here, a participant role that would be the O when the verb is used transitively appears as the S when the verb is used intransitively. transitive: tau ‘set up’ (44) jula yi-ka tau-wun [net]O go-DEP put.upright-1SGA ‘I went to set the net’ intransitive: tau ‘stand’ (45) John wun-a abukaidaan-ba to-la li’-ka-di’ [John]S 1SG-GEN left.side-LOC stand-CONSEC stay-PRS-3SG.M ‘John is standing to my left’ transitive: swaak ‘pour’ (46) wun-a gu taba swaap-mi’n? [1SG-GEN water]O already pour-2SG.M ‘have you already poured my water?’ intransitive: swaak ‘overflow’ (47) gu swaak-ka-di’ [water]S overflow-PRS-3SG.M ‘the water is overflowing’ 4.3 Multi-verb predicates • A given argument can have more than one syntactic role in a sentence. (48) [[[ saava taba vli] kwi-di’] si’-di] copy hand two give-3SG.M shoot-3PL ‘he gave [them] ten copies [of a crocodile], [and] they speared [them]. • At first glance, we might favour an analysis whereby we have two clauses, saava taba vli kwidi’ and si’di, which are juxtaposed, with the NP being omitted in the second clause. • However, this analysis seems biased by the availability of convenient English translations. An alternative analysis is to claim that instead of two clauses, only the two verbs – kwidi’ and si’di – are linked, so that the first binary division would be between saava taba vli and a complex predicate kwidi’ si’di. 14 (49) [[ saava taba vli] [ kwi-di’ si’-di]] copy hand two give-3SG.M shoot-3PL ‘he gave [them] [and] they speared ten copies’ • While this creates an awkward translation in English, this second analysis acknowledges that verbs with different argument frames can be combined in a subclausal linkage. • We thus obtain a multi-verb predicate with sharing of arguments. • The verbs in a multi-verb predicate can either share the same subject referent, or have different subjects. (50) O tr. verb intr. verb di-kak yalavi’k-ka li’-ka-wun 3PL-DAT think-DEP stay-PRS-1SG ‘I am (staying here) thinking about them’ (51) S=O tr. verb intr. verb da’mage laavwi-di li’-ka-di’ door open-3PL stay-PRS-3SG.M ‘they left the door open’ lit.: ‘they opened the door (and it) stays (open)’ (52) S=O tr. verb intr. verb vaala ada-ba kawi-di kwa-a-di’ canoe which-LOC park-3PL lie-PRS-3SG.M ‘Where is the canoe lying (after) they parked it’ • The first construction bears similarity to periphrastic tense-aspect expressions, and can therefore be translated into English as a progressive. • The different-subject complex predicate has no equivalent in European languages. • The optionality of direct object marking in (53) and (54) shows that these constituents can be encoded as a dependent of either verb, either as the O of the first, transitive, verb, as in the a.-versions; or as the S of the second, intransitive verb, as in the b.-versions. (53) (54) a. Dadagi’mai-kak Waji’mauk [Dadagi’mai-DAT]O water.spirit kla-da-di’ get-descend -3SG.M ti’kali’ stay-PRS-3SG.F b. Dadagi’mai Waji’mauk kla-da-di’ [Dadagi’mai]S water.spirit get-descend -3SG.M ‘Waji’mauk has pulled down Dadagi’mai’ ti’kali’ stay-PRS-3SG.F a. li’-kak [3SG.F-DAT]O kla-da-di’ get-descend -3SG.M ti’-ka-li’ stay-PRS-3SG.F b. li’ Waji’mauk kla-da-di’ [3SG.F]S water.spirit get-descend -3SG.M ‘Waji’mauk has pulled her down’ ti’-ka-li’ stay-PRS-3SG.F Waji’mauk water.spirit 15 • A promising candidate for neutralization of transitivity involves directional verbs. (55) jaabi’-ba wak-ka table-LOC ascend-DEP ‘put it on the table’ laaka put • The subject of the second, transitive, verb is not the subject of the first, directional, verb. The speaker does not ask the addressee to climb up the table and then put the object on the table, but rather that they should put it in such a way that it ‘ascends’ to the table. • In (56), it is the object that is supposed to enter the tent, not the addressee. (56) nyiga-ba wulai-ka laaka leaf-LOC enter:go-DEP put ‘put it inside the tent’ • In these constructions the converb forms wakka and wulaika have lost their argument frame and behave like directional adverbials. • It is interesting that this construction is very similar to Tok Pisin putim i go insait ‘put [it goes] inside’, so we cannot exclude that language contact plays a role here. ABBREVIATIONS A ALL COM CONSEC D1 D2 D3 DAT DEP DS DU F FIN FOC GEN actor~transitive subject allative comitative consecutive demonstrative proximate demonstrative distal demonstrative anaphoric dative dependent (same-subject converb) different subject dual feminine final (verb) focus genitive INTR IPFV IRR LOC M NEG NR O PL PRS SG SR SS TR intransitive imperfective irrealis locative masculine negation nominalizer undergoer~direct object plural present singular subordinator same subject transitive REFERENCES Coupe, A., 2007. A grammar of Mongsen Ao (MGL 39). Mouton, Berlin/New York. de Vries, L., 2005. Towards a typology of tail-head linkage in Papuan languages. Studies in Language 29 (2), 363–384. Genetti, C., 2005. The participial construction of Dolakh Newar. Studies in Language 29 (1), 35–87. Huang, Y. 2000. Anaphora. A cross-linguistic approach. Oxford University Press, Oxford. Jendraschek, Gerd. 2008. Iatmul stories. Gepmakudiba buchelija wapuchapuk (1st edition). Bundoora: La Trobe University. Lehmann, Christian. 2008. Information structure and grammaticalization. In Elena Seoane and María José López-Couso, eds. Theoretical and empirical issues in grammaticalization. Amsterdam & Philadelphia: John Benjamins, 207-229 Stirling, L., 1993. Switch-reference and discourse representation. Cambridge University Press, Cambridge. Stirling, L., 2001. The multifunctionality of anaphoric expressions. A typological perspective. Australian Journal of Linguistics 21.1, 7–23. Schachter, Paul. 1973. Focus and relativization. Language 49: 19-46. 16