- The many faces of Austronesian voice systems: Some new empirical studies
In January 2002, the Australian National University in Canberra hosted both the Ninth International Conference on Austronesian Linguistics and the Fifth International Conference on Oceanic Linguistics. Rather than collecting all the papers from these conferences into conference proceedings, the editors have decided to produce more thematically unified volumes; this particular volume contains the papers on Austronesian "voice" (along with a handful of other related papers that were not presented at either conference).
The resulting collection of papers brings a variety of perspectives to bear on a number of quite distinct types of languages. The book contains a pair of papers on Tagalog, but no other language is analyzed in more than one paper; other languages discussed include Puyuma, Tongan, and many languages of Indonesia. As we might expect from a set of papers from a conference, the goals and approaches of the papers differ fairly sharply (the introduction does an excellent job of relating the papers to each other and to a number of larger issues). Many (though not all) of the papers ask a question that is essentially comparative; to what extent does Austronesian "voice" correspond to voice in a nominative-accusative language like English, or in an ergative language like Dyirbal? Do differences in voice entail differences in transitivity? Should we think of the nominal singled out by the voice morphology as being like the English subject? Or does it have some more discourse-theoretic role, such as topic or focus?
Questions like these have been the center of lively discussion at least since Schachter's (1976) observation that Tagalog seems to lack any clear analog of the "subject" familiar to us from other languages. For the questions not to be simple exercises in nomenclature, we have to develop definitions of the terms in question, based on syntactic or semantic tests that can be run in different languages. In the formal syntax literature, recent work on these languages has attempted to rise to this challenge, comparing the Austronesian subject to the clause-initial phrase in verb-second languages (Richards 2000, Pearson 2005), to phrases undergoing "object shift" in languages like Icelandic (Rackowski 2002), and to the absolutive nominal in ergative languages (Aldridge 2004).
Another approach to the problem has involved redefining the questions. Guilfoyle, Hung, and Travis (1992), for example, take advantage of arguments that the notion of subject is not straightforward to define even in well-studied languages. Following much work in the formal syntactic literature (see McCloskey  for an overview of relevant arguments) they assume that the subject in a language like English is associated with multiple positions in the syntactic tree, which are typically all occupied by the same phrase, but need not be; in current Minimalist theory, for example, these positions include the specifier of vP and the specifier of TP. On this type of view, we are free to say that the [End Page 501] Austronesian subject has some but not all properties in common with the subject in a language like English, by virtue of occupying some but not all of the positions the English subject typically occupies.
Their approach illustrates a way of modeling "similarity" of syntactic phenomena. First, we break a phenomenon down into its component parts; in this case, we associate the term "subject" with a set of syntactic positions, all typically occupied by the same phrase in English. Comparison with another language then allows us to sharpen our understanding of how the properties of the subject are distributed among the different positions, by allowing us to consider examples of phrases that occupy some but not all of the positions in question. The clearest analogy is with chemistry; we are discovering the syntactic atoms that combine to create complex syntactic objects like "subject" (and, as in chemistry, we should be prepared to discover that the properties of a complex object are not simply the sum...