Workshop: Voice and Grammatical Functions in Austronesian

The Problem of Voice and Grammatical Functions

in Western Austronesian Languages

Simon Musgrave

University of Melbourne

Notes

1

I would like to thank Peter Austin for stimulating discussion of the ideas presented here, and for detailed comments on previous drafts. (Back to text)

2

The following abbreviations are used in glossing examples:

SUBJ - subject, TRM - term, OBL - oblique, DIR - direct, INV - inverse, OS(B) - Beneficiary oblique subject , OS(L) - Location oblique subject, N- - nasal prefix realised as homorganic nasal before a consonant, and as ng- before a vowel, RP - recent perfective aspect, PERF - perfective aspect, RD - reduplication, PROP - proper name marker, PN - proper name, 1sg - 1st person singular (Back to text)

3

This description of example 1b is not strictly true, The infix -in-- is actually an aspect marker which cannot co-occur with the suffix -in which is the true marker for the thematic status of the subject in this clause. (Back to text)

4

Artawa and Simpson in their contribution to this workshop argue that example 3a should be analysed as having a zero prefix in the verb form. (Back to text)

5

See also Dukes paper in this workshop for argument that Tongan should be analysed in this fashion. (Back to text)

6

The exact figures vary - see previously cited sources and also Cooreman, Fox and Givon 1984 and Shibatani 1988. In general, the text counts offer little support to either possible analysis in terms of traditional voice alternations. INV forms predominate in most studies, which certainly eliminates the nominative/accusative analysis, but DIR forms are far more common than would be warranted if such clauses are analysed as antipassives. See also our remarks below on the referentiality constraints which influence the form chosen in all the languages under consideration. (Back to text)

7

This issue is clouded by the fact that ng is also used to mark the possessive relation within noun phrases. Many previous treatments have therefore claimed that the two ngs in examples such as 6 are homophonous but functionally distinct. We see no reason to adopt this view. The marker may have distinct functions in marking predicate-argument relations and adnominal relations, but this does not mean that both functions can appear in the same level of constituency. (Back to text)

8

Gerudive nominalisations have a similar pattern. (Back to text)

9

See also Arka and Simpson's paper in this workshop for discussion of the Balinese facts. Arka and Simpson use the expression term complement for non-subject Agents. (Back to text)

10

The difference in tense in the translations of these examples is not significant. Schachter (1984 n.6) reports that his consultant translated DIR clauses as present tense, and INV clauses as past tense in the absence of any temporal context. But either clause type can be translated with either tense, given an appropriate context. (Back to text)

11

See Cena 1995 for some exceptions, and Adams and Manaster-Ramer 1988 for discussion of exactly what sort of definiteness is involved. (Back to text)

12

Indonesian has similar restrictions, but only on non-subject arguments. Indonesian unaffixed verbs can only be used with a pronoun, a proper name or a clitic Agent (Guilfoyle et al 1992:398). (Back to text)

13

Given that Schachter reports that P-SUBJ verbs are about twice as common as A-SUBJ verbs, the arguments in favour of treating Toba Batak as ergatively aligned are as valid as those for other W.An languages. There is, however, a well-known generalisation to the effect that no syntactically ergative language has ditransitive verbs. We would therefore take this construction as further evidence against treating W.An languages as syntactically ergative. (Back to text)

Created: 8 June 1998
Last modified: 12 June 1998

Authorised by: P. Austin, Professor and HOD, Linguistics and Applied Linguistics

1	I would like to thank Peter Austin for stimulating discussion of the ideas presented here, and for detailed comments on previous drafts. (Back to text)
2	The following abbreviations are used in glossing examples: SUBJ - subject, TRM - term, OBL - oblique, DIR - direct, INV - inverse, OS(B) - Beneficiary oblique subject , OS(L) - Location oblique subject, N- - nasal prefix realised as homorganic nasal before a consonant, and as ng- before a vowel, RP - recent perfective aspect, PERF - perfective aspect, RD - reduplication, PROP - proper name marker, PN - proper name, 1sg - 1st person singular (Back to text)
3	This description of example 1b is not strictly true, The infix -in-- is actually an aspect marker which cannot co-occur with the suffix -in which is the true marker for the thematic status of the subject in this clause. (Back to text)
4	Artawa and Simpson in their contribution to this workshop argue that example 3a should be analysed as having a zero prefix in the verb form. (Back to text)
5	See also Dukes paper in this workshop for argument that Tongan should be analysed in this fashion. (Back to text)
6	The exact figures vary - see previously cited sources and also Cooreman, Fox and Givon 1984 and Shibatani 1988. In general, the text counts offer little support to either possible analysis in terms of traditional voice alternations. INV forms predominate in most studies, which certainly eliminates the nominative/accusative analysis, but DIR forms are far more common than would be warranted if such clauses are analysed as antipassives. See also our remarks below on the referentiality constraints which influence the form chosen in all the languages under consideration. (Back to text)
7	This issue is clouded by the fact that ng is also used to mark the possessive relation within noun phrases. Many previous treatments have therefore claimed that the two ngs in examples such as 6 are homophonous but functionally distinct. We see no reason to adopt this view. The marker may have distinct functions in marking predicate-argument relations and adnominal relations, but this does not mean that both functions can appear in the same level of constituency. (Back to text)
8	Gerudive nominalisations have a similar pattern. (Back to text)
9	See also Arka and Simpson's paper in this workshop for discussion of the Balinese facts. Arka and Simpson use the expression term complement for non-subject Agents. (Back to text)
10	The difference in tense in the translations of these examples is not significant. Schachter (1984 n.6) reports that his consultant translated DIR clauses as present tense, and INV clauses as past tense in the absence of any temporal context. But either clause type can be translated with either tense, given an appropriate context. (Back to text)
11	See Cena 1995 for some exceptions, and Adams and Manaster-Ramer 1988 for discussion of exactly what sort of definiteness is involved. (Back to text)
12	Indonesian has similar restrictions, but only on non-subject arguments. Indonesian unaffixed verbs can only be used with a pronoun, a proper name or a clitic Agent (Guilfoyle et al 1992:398). (Back to text)
13	Given that Schachter reports that P-SUBJ verbs are about twice as common as A-SUBJ verbs, the arguments in favour of treating Toba Batak as ergatively aligned are as valid as those for other W.An languages. There is, however, a well-known generalisation to the effect that no syntactically ergative language has ditransitive verbs. We would therefore take this construction as further evidence against treating W.An languages as syntactically ergative. (Back to text)