Comments on S. R. Fischer's 'Mangarevan doublets: Preliminary evidence for Proto-Southeastern Polynesian'

Ssteven Roger Fischer (2001) has recently published arguments for an early Eastern Polynesian protolanguage distinct from Proto-Eastern Polynesian and Proto-Central Eastern Polynesian. Many aspects of the work constitute valuable contributions to fleshing out the early language (pre)history of Eastern Polynesia. Linguists will have many colleagues in archaeology and anthropology who welcome the presentation of some of F's data and conclusions. However, the data do not always mean what F says they do. He deviates from several important conventions in linguistics in his presentation and, in general, the work lacks a well-developed sense of language, population, and their interaction in the sociolinguistic sense.

Geologists have a concept called "uniformitarianism," which involves the general idea that geological formations to be observed today are the result of weathering, sedimentary, glacial, uplifting, meteoric, and other processes that, in the main, can be observed today. While there is no similar term with a similar meaning generally employed in linguistics, some writers (Christy 1983, Labov 1994:21-25) propose that we should simply use the geological term for our general sociolinguistic concept, which might be stated roughly as "In the main, sociolinguistic processes of prehistory fell within the general range of those to be observed today."

F violates this general principle in the present work and in an earlier one. Fischer (1992) presents a number of valuable linguistic observations, but then claims that they mean things, in the sociolinguistic sense, that they do not. Paraphrasing criticisms of that work from Marck (1996a and 1996b), the problem with Fischer (1992) is as follows.

He presents a study of the Rapanui language (Easter Island) where his grammatical comparisons are a unique addition to our knowledge of Eastern Polynesian internal relationships and are consistent with the view that the Rapanui language developed in isolation after an early break from other Eastern Polynesian languages. However, he reaches the unwarranted conclusion that, based on those observations, we can conclude that "one courageous canoeful of East Polynesians were the first and only ones to arrive on Rapanui until their descendants' historical encounter with Europeans in 1722" (Fischer 1992:187).

Such a conclusion is not recoverable from the comparative method of linguistics. F presents a well developed argument that careful examination of the Rapanui language reveals no non-Polynesian substrate, and a similar strong argument that post-settlement (of the "one courageous canoeful") influence from other languages cannot be demonstrated for the period prior to European contact. But we cannot infer from this that there was a single canoeful of initial settlers or that there was no contact with other people from the beginning of settlement to the time Europeans appeared. Neither does it establish that more people over a longer period of time were not involved in Rapanui's initial or even continuing settlement through some generations—a century or more.

His data simply show that such potential continuing influxes and later contacts had no perceptible linguistic results. Such lack of impact is actually common in what we know of same-source immigration and of even very meaningful later contacts from elsewhere in terms of world languages, even with borrowing of material and subsistence items and their sociolinguistic results.

It must be confusing, to nonlinguists, to read some of F's work and wonder how a linguist can infer the things he does. We cannot, and the present work (Fischer 2001) contains more of such errors. In addition, there is substantial misuse of historical and comparative linguistic concepts and terminologies.

There are two general kinds of problems with Fischer (2001). The first involves the linguistics. The second involves a kind of circularity, with archaeological and linguistic inference feeding upon each other more than they rightly should. The problems occur in that general order, so I will give my comments by page or section of the article, with a final summary.

The main linguistic problem I see is a lack of clarity about when we can refer to components of a language as substratum, and how that directs or informs us when we then have to talk about inheritance...