A Typology of Clausal Complementation in Oceanic Languages
1
Projet Frantisek Lichtenberk pour demande de Poste Rouge au Lacito en 2011
A typology of clausal complementation in Oceanic languages
- Background
Following Noonan (2007: 52) (clausal) complementation can be defined in functional terms as a “syntactic situation that arises when a notional sentence or predication is an argument of a predicate”. Complementation has been studied extensively from diverse theoretical perspectives in a considerable variety of languages. And there are also cross-linguistic, functionally and typologically oriented studies of complementation, such as Noonan (2007), Givón (1984, 1990), Dixon (1995, 2006) and Cristofaro (2003). In fact, for the purposes of a typological study a functional definition of a syntactic phenomenon is usually the only workable type of definition, rather than one that is grounded morphosyntactically. What can be thought of as one functional domain may be realized by quite different morphosyntactic means in different languages, or even within one and the same language. What unites such various constructions is precisely the fact that they serve what can broadly be characterized as a single function. Thus, most likely all languages have morphosyntactic means whereby a “notional sentence or predication [can be made] an argument of a predicate” (see the quote from Noonan (2007) above), but the formal strategies with which this is achieved vary from language to language, and typically a single language has more than one such strategy. Noonan’s definition will be adopted as a starting point, although some adjustments will have to be made to fit the Oceanic languages.
While there are studies of complementation in individual languages, and while there are cross-linguistic, typological studies of complementation, what is much less common is studies of complementation within language families. As Dixon (1995: 183) puts it: “Little systematic study has as yet been attempted of the genetic and areal distribution of the various kinds of clause linking.” The research project proposed here is intended as a step in the direction of rectifying this. The aim of the project is to provide a comprehensive typology of complementation strategies in the Oceanic language family. The Oceanic languages form a subgroup within the Austronesian family. There are anywhere between 400 and 500 languages in the Oceanic group. (Austronesian as a whole has upwards of 1,000 languages, which makes it one of the two largest language families of the world.) The Oceanic languages are spoken in Papua New Guinea, Island Melanesia, Micronesia and Polynesia, and they are typologically quite diverse. There are verb-medial, verb-final and verb-initial Oceanic languages. While most of the Oceanic languages are of the accusative type, there are also some ergative languages. And while some languages are relatively isolating, others exhibit a fair amount of morphological complexity of words.
A morphosyntactic topic that has attracted considerable attention in the study of Oceanic languages is transitivity. In many Oceanic languages the distinction between intransitive and transitive verbs is overtly marked by morphology: transitive verbs are derived from intransitive ones by affixation, and typically several classes of transitive verbs can be distinguished on the basis of their morphology. The proposed study of complementation is clearly directly relevant to the phenomenon of transitivity. However, while there has been a number of cross-linguistic studies of Oceanic transitivity from a morphological perspective (for example, Clark 1973, Pawley 1973, 2001, Ross 1998 and Evans 2001), much less has been said about the syntactic aspects of transitivity. (A notable exception is the debate about ergativity in Polynesian; see Moyse-Faurie 2003 and Pawley 2001 for a recent discussion.) And while complementation, specifically cases where the complement clause functions as a direct object, has direct relevance to transitivity, there are no comprehensive studies of complementation in Oceanic. (In a recent wide-ranging study of grammatical structures in Oceanic languages (Lynch et al. 2002) complementation is treated only cursorily.) The proposed research will remedy this. The topics that will be investigated are discussed in the next section.
2. Topics to be investigated
2.1. Complementation strategies found in the Oceanic languages. Noonan (2007) distinguishes the following main types of complementation strategies: sentence-like (S-like) complements, paratactic complements, infinitive complements, nominalized complements and participial complements. According to Noonan, all languages around the world have S-like complements and at least one other type. On the other hand, Dixon (1995: 213) says that “[t]here are many … languages that lack complement-clause constructions” and stresses that “[c]areful grammatical description of such languages should have a high priority in future research”. It should be borne in mind though that Dixon has a quite narrow definition of complement clauses and makes a distinction between complement clauses and complementation strategies (see also Dixon 2006.) Available data suggest that as a rule Oceanic languages have at least S-like complements and nominalized complements, as in Xârâcùù (Moyse-Faurie 1995) and in Nêlêmwa (Bril 2002). The purpose of this part of the research project will be to determine whether or not S-like complementation and nominalized complementation are universal in Oceanic, and what other types are found. For example, Manam and Tamambo have paratactic complementation (Lichtenberk 1983, Jauncey 1997).
Besides complementizers (see sections 2.3 and 2.4 below), there may be other morphosyntactic strategies involved in interclausal dependencies, such as anaphoric demonstratives, clause delimiters and boundary morphemes. The project will investigate the types of clause-linking that have to do with discourse strategies, factors such as focussing and topicalization, information flow, cohesion, referential hierarchy, and definiteness or saliency (foreground/background) effects.
In some cases, however, interclausal dependency has no dedicated morphosyntax; rather, it may be signalled solely by prosody, or it may be only pragmatically based. Prosodic and pragmatic factors in complementation will also be investigated.
2.2. Finite vs. non-finite complement constructions and their syntactic distribution. In the Oceanic context, finite complementation includes S-like and paratactic complements, and non-finite complementation includes bare verbal forms and nominalizations. Many languages around the world exhibit restrictions on the positioning of finite complements. For example, languages may freely allow finite complements in object position, but not so freely in subject position. English does allow finite complements both in object position and in subject position, but while the former type of case is common, the latter is much less so, especially in informal speech. Furthermore, while in English, subject finite complements can occur in declarative sentences (where they are sentence-initial), they cannot occur in interrogative sentences (where they would be sentence-medial). On the other hand, nominalized complements typically have fewer, if any, conditions on their occurrence in different syntactic positions. The purpose of this part of the project will be to investigate the presence or absence of syntactic restrictions on the occurrence of finite and non-finite complements.
2.3. The presence and absence of complementizers. Finite complement clauses may require a complementizer, or a complementizer may be optional, or there may not be any complementizer possible. An Oceanic language may have a relatively large number of complementizers, it may have just one, or it may have none. For example, Boumaa (a variety of Fijian) has four complementizers, and there are structural and semantic differences among them: one is used only in interrogative sentences, another one is used when there is uncertainty about the truth status of the proposition, another one is used to express the fact that an event “should” happen, and the last one is used elsewhere (Dixon 1988). Similarly, Tîrî (Osumi 1995) and Nêlêmwa (Bril 2002) have several complementizers. On the other hand, Toqabaqita has only one complementizer, which is used only rarely (Lichtenberk 2008). And Manam has no complementizer at all; finite complement clauses occur in parataxis with their main clauses (Lichtenberk 1983). A typology of complementizers and their uses in Oceanic will be developed as part of the project.
2.4. Historical origin of complementizers and other functions of the etyma. Complementizers develop historically, via grammaticalization processes, from elements with other functions, such as demonstratives, conjunctions, case markers and verbs (Dixon 2006, Noonan 2007). Among verbs, it is those with the meaning ‘say’ that often develop into complementizers (see, for example, Schachter and Shopen 2007: 52). This is also the case in Oceanic, for example in Lolovoli (Hyslop 2001), Erromangan (Crowley 1998) and in Touho (Rivierre 1980). The forms that function as complementizers not infrequently have other grammatical functions, such grammatical polyfunctionality also being the result of grammaticalization processes. For example, in Toqabaqita the element na introduces complement clauses and relative clauses and it also functions as a focus marker (Lichtenberk 2008). In Nêlêmwa, xe is a topic marker, occurs in relative clauses introducing new information and also functions as a complementizer with verbs of perception and cognition (Bril 2002). In Mwotlap, an element that functions as a deictic clitic also serves as a topic and focus marker and as a clause union marker (François 2000). And in a number of Polynesian languages, Fagauvea being one of them, there are complementizers of deictic origin. The questions of the historical origin of complementizers and of their polyfunctionality are intimately related and will be investigated jointly.
2.5. Nominalized complements. Nominalizations are clearly a wide-spread complementation strategy in Oceanic, and most grammars of Oceanic languages contain some discussion of nominalization. (Tallerman (1998) mentions nominalization as a widespread complementation strategy in Austronesian.) Many Oceanic languages possess more than one way of deriving nominals from verbs, and a distinction is usually made between proposition/activity/state nominalizations on the one hand and deverbal nouns on the other. For example, in Toqabaqita both types can be derived from one and the same verb: based on the verb foqa ‘pray’, there is a proposition/activity/state nominalization foqa-laa ‘(activity/event of) praying’ and a deverbal noun foqa-a ‘prayer’. It is the proposition/activity/state nominalizations that are relevant here. It is well known from the study of nominalizations in various languages of the world that they often exhibit both nominal and verbal properties, and the same applies to Oceanic languages, although no systematic study has been done to date. For example, in Toqabaqita, nominalizations exhibit both nominal and verbal properties. Like nouns they can function as the head of a possessive noun phrase:
(1)Nguu-la-naqeleqa.
sing-nmlz-3sg:pers3sg:nfutbe.nice
‘His singing is nice.’
Compare:
(2)thaina-na
mother-3sg:pers
‘his mother’
They can cooccur with quantifiers and classifiers:
(3)Fitafariki-laan=oriki-aqana
how.many?clfsee-nmlzfoc=2sg:nfutsee-3:objgenp
wanebaa?
manthat
‘How many times have you seen that man?’ (Lit.: ‘How many seeings that you have seen of the man?’)
Compare:
(4)fitafaniu?
how.many?clfcoconut
‘how many coconuts?’
Like nouns they can be modified by a demonstrative:
(5)Thuuma-laabaaitaraqena.
reconcile-nmlzthatloctoday
‘The (time of) reconciliation is today.’
Compare the use of the demonstrative baa with a non-derived noun in (3) above. Noun phrases headed by nominalizations can function as arguments of verbs or complements of prepositions. Thus, in (1) above, the nominalization occurs in subject position, and in (6) below it occurs as the complement of a preposition.
On the other hand, like verbs, nominalizations can occur with at least some of the particles that occur with verbs in a “verb complex”. In (6) the nominalization occurs with the ventive directional:
(6)…kakwaimangosiqanalae-laamai.
3sg:seqpreventgenpgo-nmlzvent
‘… and it prevented (him) from coming.’
Compare:
(7)Laemai!
govent
‘Come here!’
And in (8) the nominalization occurs with the perfect-tense/aspect marker:
(8)Tawanekaloleqemae-laanaqa, ....
oneman3sg:seqpretenddie-nmlzprf
‘One of the men pretended to have died ….’
Compare:
(9)Niaemaenaqa.
3sg3sg:nfutdieprf
‘It [a possum] has died.’
In his discussion of “clausal NPs” (i.e. nominalized complements) in Boumaa Fijian, Dixon (1995) comments that “they have the internal structure of a clause”, but also “a structure very similar to that of a plain NP” (p. 193), and then goes on to say: “The question of whether ‘clausal NPs’ in Fijian should be regarded as a type of NP, or simply as a type of clause, is a tricky one, which could be long debated.”; and he concludes, “… the question is scarcely important” (p. 194). There is no need to decide whether nominalized complements have the structure of a clause or of an NP. Clearly such complements partake both of nominal and of verbal properties, and it is precisely their mixed nature that makes them interesting and worth studying, and so their properties are, in fact, highly important.
2.6. Complement-taking verbs/predicates. There exist various kinds of classification of verbs/predicates that take complements, with various degrees of elaboration. For example, Givón (1990) postulates only three major classes of complement-taking verbs: modality, manipulative and cognition-utterance. Dixon (2006) distinguishes four primary verb types: attention, thinking, liking and speaking; and several secondary types: negating, modal, beginning (i.e. phasal), trying, wanting and causing. On the other hand, Noonan (2007: 126–145) provides a much more detailed nomenclature of complement-taking verbs, or more accurately complement-taking predicates: utterance, propositional attitude, pretence, commentative (factives), knowledge and acquisition of knowledge, fearing, desiderative, manipulative, modal, achievement, phasal (aspectuals), immediate perception, negative and conjunctive. The purpose of this part of the project will be to investigate what types of complement-taking predicates Oceanic languages tend to have, and what kinds of complements those types tend to take, for example S-like complements vs. nominalized complements, and full complements vs. reduced complements. (For reduced complements, see section 2.10 further below). Of special interest will be utterance verbs and Dixon’s secondary types (see sections 2.7, 2.8 and 2.9).
2.7. Utterance complement-taking predicates. Here the main focus of the study will be the extent to which languages use direct and indirect speech (including direct and indirect interrogatives). In the studies of some languages around the world, claims have been made that those languages do not have indirect-speech constructions. Such claims have been questioned by Palmer (1986), who suggests that those languages do, in fact, have indirect speech. However, Dixon (1995: 211) explicitly disagrees with Palmer, saying that “[i]n common with many other languages in the world, [endnote omitted] Dyirbal has no grammatical construction of ‘indirect speech’”; and “Some languages lack this grammatical technique [i.e. indirect speech] and simply employ ‘direct speech’ ….” (Dixon 2006: 10). The extent to which indirect speech is used in Oceanic languages varies: it appears quite common in Boumaa Fijian (Dixon 1995) and in Nêlêmwa (Bril 2002), while it is relatively rare in Manam, where direct speech is preferred (Lichtenberk 1983). Another aspect of utterance complementation that will be studied is the fact that in some languages the predicates used to introduce direct speech exhibit special properties. Thus, in Toqabaqita, direct speech is most commonly introduced by a predicate that is internally a noun phrase. The noun phrase is headed by the noun ‘manner’, which is modified by a cataphoric deictic. However, this noun phrase takes a subject-tense marker, as if it were a verbal construction:
(10)Thaaribaakaqunaqeri,"Qeleqaba-na,
girlthat3sg:seqmannerthis3sg:nfutbe.goodlim-3sg:pers
kokifanga.”
1du(incl):futeat
‘The girl said, “That’s good; let’s eat.”’
It is not unusual for expressions of direct speech to exhibit idiosyncratic syntactic properties. For example, in Toqabaqita, clauses of direct speech are complements in the sense that the higher predicate requires an expression of direct speech but they are not arguments of the higher predicate. Thus, in (10) the predicate quna qeri requires an expression of direct speech, which, however, is not its argument (subject or object). Accordingly, this part of the project will also investigate syntactic properties of direct-speech constructions.
2.8. Manipulative complement-taking predicates. Here the focus of the study will be causative and permissive constructions. In some languages causative and permissive relations between events are encoded by means of serial verb constructions, but serial verb constructions will not be specifically considered here. Neither will morphological causatives. The focus will be on periphrastic causative and permissive constructions, with a higher manipulative verb. While in such constructions the complements (which encode the caused events) may be clause-like, they often exhibit various kinds of syntactic reduction. (For reduced complements see section 2.10 further below). For example, in Toqabaqita in a causative sentence the causee is encoded both in the higher clause and in the complement. However, while in the higher clause it can be encoded by a noun phrase that functions as the direct object of the ‘cause’ verb, in the complement clause it can only be encoded by means of a subject-tense marker, not by means of a noun phrase. In example (11), the causee noun phrase ‘the children’ is the direct object of the cause verb ‘do’, and as such is indexed on the verb by the object suffix. In the complement clause the children can only be referred to by means of the sequential subject marker.
(11)Fangaqeade-awelaki,kekamataqi.
food3sg:nfutdo-3:objchildpl3pl:seqbe.sick
‘The food made the children sick.’
In this part of the project the focus will be on the morphosyntactic properties of complements under verbs of manipulation.
2.9. Modal predicates. In many Oceanic languages various kinds of epistemic and agent-oriented modalities are encoded by higher predicates that take as their complement clauses whose predications are characterized by the modality. Modal higher verbs are found in many Polynesian languages, for example in Tuvaluan (Besnier 2000: 55). In (12) the complement clause (in square brackets) functions as the subject of the modal verb.
(12)Eailogalaakoe[ofanoitevakanei].
npastunlikelythen2sgcompgoontheshipthis
‘It’s unlikely that you’ll be leaving on this boat trip.’
Modal verbs are also found in other Oceanic languages, for example in Boumaa Fijian and in Toqabaqita, as illustrated in (13) and (16), respectively. In Boumaa the complement clause is the subject of the modal verb (Dixon 1988):
(13)Erawa[nila’ooJone].
3sgcancompgoartJohn
‘John can go.’
Such constructions are also found in Nêlêmwa (14), besides other constructions with a semantically full subject (15):