Split Intransitivity and Saweru

Abstract

The phenomenon of split intransitivity is discussed in a variety of languages, emphasizing the contrast between two-way and three-way split intransitivity. The agreement system of Saweru, a Papuan language of West Papua, is examined, and there follows a discussion of where Saweru fits into a typology of split intransitivity.

1. Intransitivity Splits and Alignment.

The simple, yet effective, diagnostic of alignment (whether morphological or syntactic) in a language relies on the examination of the treatment afforded to the single argument of an intransitive verb, and a comparison of this with the treatment of the arguments of primary transitive verbs. Using the labels A, S, and O to represent the syntactic roles found in these clauses (Heath 1975, Dixon 1979, 1994-the labels are defined as per Andrews 1985:68, with S corresponding to an intransitive predicate's sole argument, and A and O representing, respectively, the most and least agentive arguments in a transitive verb's subcategorization frame), we can plot the more common variations in treatment according to groupings of these arguments, and the labels that are used to describe them, as shown in figure 1.

When we are discussing morphological alignment, the morphology that encodes these groupings can be nominal case marking, or verbal agreement, or both; additionally, the most common system has one category being overtly marked morphologically, and the other not. Examples of morphological groupings leading to alignment classification can be found overtly in Japanese and Nias. In Japanese, the alignment is determined by examining the case marking, which utilizes the particles ga and o, marking S and A groupings and an O category respectively, thus leading to the conclusion that there is a nominative-accusative alignment.

Figure 1.

Alignment

Japanese:

1. 1. [_S  Ano  hito   ga]    ik-ta.
       that   person NOM   go-PAST
   'That person went.'

2. 2. [_A  Ano  hito   ga]    [_O    tori o]    mi-ta.
       that   person NOM        bird   ACC   see-PAST
   'That person saw a bird.'

3. 3. Japanese case marking and syntactic roles

   	A	S	O
   ga	✓	✓
   o			✓

In Nias (Donohue and Brown 1999), we find alignment determined by verbal agreement, which has an overt category for A, and no agreement for S or O (thus following an ergative pattern). This alignment is confirmed by the case marking on NPs, which applies on NPs in S or O function and which is marked by initial mutation (illustrated here with the addition of n- on vowel-initial words), and is thus absolutive.

Nias:

(4) Manavuli sui  [_S n-ama-da           Tohönavanaetu]ba  Maenamölö.
return       again   MUT-father-1PL.IN.GEN Tohönavanaetu    LOC Maenamölö
'Ama Tohönavanaetu came back again to Maenamölö.'¹

(5) I-bözi   [_O n-ama-gu]       [_A Ama Dali].
3SG.R-hit      MUT-father-1SG.POSS    Ama Dali
'Ama Dali hit my father.'

(6) Nias case and agreement system, with syntactic roles

	A	S	O
AGREEMENT	✓
MUTATION		✓	✓

Although this scheme seems neat and useful, not all languages fit into the categories defined in figure 1. One obvious problem lies in identifying transitive clauses: what is the algorithm for determining which of two basic transitive patterns should count, for instance, as is the case in Tagalog (Maclachlan 1995)? Additionally, a language could conceivably have separate coding strategies for each of A, S, and O-this is referred to as a "tripartite system," several varieties of which will be discussed in section 5. More profound problems are encountered when the single arguments of intransitive predicates are coded in different ways depending on the semantic form of the verb. In languages with this system, typically, more patient-like arguments (fall, be sick, yawn) tend to be coded as Os, and more agent-like arguments (run, sing, go) are coded in the same way as As; the unity of the S syntactic role is thus questioned, and this claim has been explicitly explored by Foley and Van Valin (1984), and Dowty (1991). Because the intransitive category is not coded in one consistent manner, these alignment systems can be called split-intransitive, after Merlan (1985).^2...