10-ICAL historical-comparative papers (review)

The first of five volumes of papers presented at the Tenth International Conference on Austronesian Linguistics, held in Puerto Princesa on the Philippine island of Palawan in January 2006, this volume contains six papers by five authors: Isidore Dyen, John Wolff, Ronald Himes, Robert Blust, and Tom Laskowske. Two of the papers are by Wolff.

The historical linguist in me is gladdened to see a volume of historical-comparative papers, but the reader should note that this is no introduction to Austronesian historical linguistics. The papers by Dyen and Wolff make numerous assumptions that many, probably most, of their colleagues do not share, and there are many pitfalls for the unschooled reader. I have endeavored below to point out some of these, but it is not easy to do so without presupposing some knowledge of Austronesian historical linguistics on the readers’ part. At the opposite end of the scale, the paper by Robert Blust is a masterpiece of readable historical-comparative exegesis.

The first paper, Isidore Dyen's “Some evidence favoring the Central Hypothesis” (1–12), argues in support of the author's 1965 hypothesis of a Melanesian homeland for Proto-Austronesian (PAn) and against the Formosan Hypothesis that the homeland was in Taiwan.

Dyen associates the Formosan Hypothesis with Robert Blust (see Blust 1977), and begins in section 1 with a fierce attack on Blust's (1999) critique of Dyen's methods. The reader rapidly concludes that their positions are irreconcilable. It is not so much that Dyen and Blust would never have reached agreement about the Austronesian homeland (Dyen passed away in December 2008), but that the two were separated by a methodological gulf that could never be crossed because each made assumptions that were incompatible with the other's.

Dyen does subgrouping on the basis of lexicostatistics and of homomeries, a method unique to him (Dyen 1990). A homomery “is a collection of exclusively shared cognate sets.” Blust, on the other hand, argues on the basis of shared innovations, that is, on the basis of the classical comparative method that emerged from the work of the Neogrammarians in the late nineteenth century. The two methods are incompatible because Blust's work is based entirely on shared innovations, whereas Dyen's homomeries “are not a list of innovations, but a list containing innovations” (italics mine). Dyen describes this as “a subtle difference,” but does not expand on it. It is clear, however, that he assumes that a large collection of homomeries must include a large collection of shared innovations, even though he cannot identify them.

I see no reason to make this assumption. The “Homomeric grid for Formosan languages” shows that the Puyuma–Paiwan homomery contains 117 exclusively shared cognates, yet a close examination of Puyuma and Paiwan morphosyntax (now possible thanks to Teng 2008 and Chang 2006) suggests that the two languages are not particularly closely related. Blust (1999:47–51) is almost certainly correct in proposing that there have been numerous loans between the two. Exclusively shared cognates may also have arisen simply because words that are retained in both languages (from PAn, I assume) happen to have been lost in all other Austronesian languages for which relevant data are available: they would not need to be lost independently very many times for this to be true. The homomeric method thus shares with lexicostatistics the inability to distinguish innovations from retentions, and this puts it at odds with the comparative method.

It will be obvious from the foregoing paragraphs that I agree methodologically with Blust. I also agree with his conclusion that Taiwan is the Austronesian homeland. However, I know of no practising Austronesianist who employs Dyen's methods or who has published in support of the Central Hypothesis.

Dyen attacks Blust (1999:67) on two related grounds: first, that the latter's summary of Dyen's homomeric argument is Blust's “concoction,” and second, that Blust...