A study of valency-changing devices in Proto Oceanic (review)

The volume under review is a revised version of the author's Australian National University PhD thesis. While it does not break any new ground (which it did not set out to do), the monograph is a solid systematization of results arrived at by previous studies and it provides new insights. Its central concerns are the system of verb classes in Proto-Oceanic (POC) and some of the valency-changing devices POC possessed. (One valency-changing device that is left out of consideration is the prefix *paRi-, which probably had several functions and would merit a study of its own. And valency-decreasing reduplication is considered only briefly.)

The volume is divided into eight chapters, followed by two appendices. One of the appendices lists the languages—Oceanic and non-Oceanic—referred to, and the other is a list of reconstructions together with supporting data. Most of the reconstructions are for POC; others are for lower-level subgroups. The introductory chapter provides background information on the Oceanic group of languages and deals with the conventions employed in the study.

Chapter 2 is a discussion of verb classes. This is highly relevant to the study, because the author argues, and usually demonstrates convincingly, that there is a correlation between the semantic types of verbs and their morphosyntactic characteristics. Evans adopts Chafe's (1970) classification of states of affairs into states, processes, actions, and process-actions. This distinction works reasonably well, but there are occasional problems. Intransitive verbs are divided into Actor-subject verbs and Undergoer-subject verbs. When an intransitive verb has one or more transitive, derivationally related counterparts, the relation may be of an applicative type or of a causative type. In the applicative type, the S argument of the intransitive verb corresponds to the A argument of the transitive verb; and in the causative type, the S argument of the intransitive verb corresponds to the O argument of the transitive verb.

On the basis of data from present-day languages Evans posits a distinction between Undergoer-subject and Actor-subject verbs for POC. Undergoer-subject verbs participated in causative relations, while Actor-subject verbs participated in applicative relations. Derivational relations between intransitive and transitive verbs could operate in either direction: a transitive verb could be based on an intransitive verb, and an intransitive verb could be based on a transitive verb. Among the valency-changing mechanisms considered by Evans, transitivization employed the causative prefix(es) *pa[ka]- and the suffixes *-i and *-akini; and detransitivization employed the prefixes *ma- and *ta-, and reduplication.

Evans provides examples of Undergoer-subject and Actor-subject verbs of various semantic classes in a number of Oceanic languages. In many instances, there is agreement among the languages with respect to verbs of the same semantic type. For example, Undergoer-subject verbs with the meaning 'be closed, shut' uniformly participate in causative derivations: 'close, shut s.t.; cause s.t. to be closed, shut'. And Actor-subject "corporeal" verbs regularly participate in applicative relations. The category of corporeal verbs includes verbs of excretion and secretion, and verbs of consumption; for example 'drink (INTR)' and 'drink s.t.'.

However, there are also verbs where there is variation among the languages, even when one and the same etymon is involved. As Evans points out, reflexes of POC *roηoR (her *loηoR) 'be heard'(?), 'hear'(?) are Undergoer-subject verbs ('be heard') in some languages and Actor-subject verbs ('hear') in others.

Verbs with the meaning 'die, be dead' that are reflexes of POC *mate are problematic in a different way. In most languages the verbs are transitivized by means of prefixes that reflect POC *pa[ka]-, and the derivations are of the causative type: 'kill; cause to die, cause to be dead'. However, in at least some Cristobal-Malaitan languages, reflexes of *mate participate in an applicative derivation, where a transitive suffix is added to the...