Like other Western Austronesian languages, Malay is well-known for its symmetrical voice system, but in this paper, McKinnon et al. (2011) report on something quite different in a little known Malay variety, Kerinci Malay, spoken in a valley near Mount Kerinci, the highest volcano of Indonesia. Kerinci has a distinction between an Absolute form and an Objective form for every transitive verb, e.g.
The focus of their paper is the syntactic distribution of the Objective form (they use the traditional but confusing term “oblique form”, or “O-form”), which they claim licences a null object pronoun. Since null object pronouns are often taken to be licenced by agreement morphology in generative grammar, they talk about the Absolute/Objective contrast as an agreement system. From my perspective, this looks suspiciously as if they are squeezing Kerinci into a conceptual framework that is familar from (and probably justified for) English-like languages, so I wondered whether the Kerinci facts can be looked at in a different way.
The facts that they take their null-object-pronoun analysis to explain are essentially the following:
(1) When a verb is used with an overt (definite or indefinite) object or when a definite-anaphoric object is implicit, the Objective form is used (e.g. ‘chased the monkey’, ‘chased a monkey’, ‘chased you’, or without overt object: ‘chased him’) (§2.1-2).
(2) When the object is unspecified, the Absolute form is used (‘I am going to eat right now’) (§2.3).
(3) When the verb is in the Object Voice (a construction somewhat similar to a passive), the Absolute form is used (§3.1).
(4) Object question-word fronting (‘Who did Budi shoot?’) is possible even with Active-Voice verbs, even though generally in Malay this is impossible (§4.1).
(5) Relativization from potential “islands” such as relative clauses is possible (‘The man that I met the woman who shot (him) died’) (§4.2).
Facts (1)-(3) are not surprising of the Objective form is simply a kind of transitive marker (and according to Mark Donohue, p.c., it may derive from the Malay suffix -i, a transitive/applicative suffix). The latter two facts follow if the null object pronouns can act as resumptive pronouns, because resumptive pronouns are known to allow speakers to circumvent island effects.
Now it seems to me that this reasoning is based on two presuppositions that are warranted only for English-like languages:
(i) Referents that can be inferred from the context must be syntactically present as pronouns, even if they are not pronounced.
(ii) Extraction restrictions can be circumvented only by resumptive pronouns (whether overt or null).
The first presupposition has long been suspect to linguists, and McKinnon et al. actually mention a very reasonable alternative: implicit anaphora, the contextual reconstruction of omitted arguments: Some languages “permit arguments that are understood from context to be omitted from the syntactic representation of the sentence” (p. 744). So why do they not adopt this alternative for Kerinci? In fact, implicit anaphora is widely attested in Standard Indonesian and all or most Malayic varieties, and they do claim that it exists in Kerinci, in addition to the null object pronouns (p. 743). But implicit anaphora cannot rescue potential island violations, so they think that they are forced to assume null object pronouns.
Now it seems to me that a much simpler generalization is possible that assumes neither (i) nor (ii): Circumvention of island violations is due not to the existence of (overt or null) pronouns, but to the existence of an overt clue pointing to the presence of the relevant argument in the conceptual structure. Gaps present the hearer with a processing difficulty, and overt resumptive pronouns are a crucial aid to processing (Hawkins 1999). Unpronounced resumptive pronouns, by contrast, would not help. In the case of Kerinci, the fact that verbs in their Objective form are overtly marked as having an object plays this processing-aid role.
If we no longer have a motivation to assume null object pronouns, this also means that any motivation for considering the Absolute/Objective contrast as a form of agreement disappears. In fact, even on McKinnon et al.’s analysis, using the term “agreement” is highly questionable, and they acknowledge that the Objective form codes “the morphological registration of the presence of a nominal argument” (n. 13), not person/number agreement in the usual sense. But later they object to the view that mere registration of the presence of an element is not agreement, citing Coppock & Wechsler’s (2012) discussion of Hungarian (n. 18). But Coppock & Wechsler in fact argue for agreement of the verb in definiteness in Hungarian, and a good case can indeed be made that in addition to agreement in gender and number (and occasionally in person), languages may occasionally show agreement in definiteness. But agreement always has to be “agreement in something”, and the Objective form of the Kerinci verb does not agree in anything with its object, it just registers its presence. Calling it an agreement form seems to be motivated purely by the analogy with person/number/gender inflection, which is usually called “agreement” and which in generative analyses is often claimed to licence null pronouns. Thus, the term “agreement” is extended here in its sense beyond recognition, in the typical generative fashion (think of “case”, “government”, and more recently, “applicative”).
Is it necessary to walk up all the way to Mount Kerinci, the highest volcano of Indonesia, to find a fairly English-like language there? It seems to me that this language does have something special in its Absolute/Objective contrast, though languages with regular overt transitive markers (such as some Polynesian languages) may actually be quite similar. What I have not seen before is an explicit discussion of the island-violation-rescuing effect of transitive marking, so this is something to look out for in other languages.
Coppock, E. & S. Wechsler. 2012. The objective conjugation in Hungarian: Agreement without phi features. Natural Language and Linguistic Theory.
Hawkins, John A. 1999. Processing complexity and filler-gap dependencies across grammars. Language 75, 244–285.
McKinnon, Timothy & Peter Cole & Gabriella Hermon. 2011. Object agreement and ‘pro-drop’ in Kerinci Malay. Language 87(4), 715–750.