Siraya: Retrieving the phonology, grammar and lexicon of a dormant Formosan language by Alexander Adelaar (review)

Of the 25 or so indigenous languages of Taiwan, those of the west coast, probably among the earliest to separate from Proto-Austronesian (PAn), have largely died out in the past couple of centuries under the impact of Chinese penetration. Siraya, once spoken in southwest Taiwan in the region of present-day Tainan and Kaohsiung, became extinct— Adelaar writes “dormant” so as not to preempt attempts at reviving it by descendants of Siraya speakers—in the late nineteenth or early twentieth century. Seventeenth-century Dutch missionary sources are particularly extensive and important for Siraya and its northern neighbor Favorlang due to the presence on parts of west Taiwan of the Dutch Verenigde Oostindische Compagnie (East India Company) between 1624 and 1661, and the opening of these regions to proselytizing by Dutch Calvinist missionaries invited by the Compagnie (historical background recounted by Adelaar on pp. 8–13). After 1895, Japanese linguists were able to collect fragmentary word lists from the last Siraya speakers (now available in Li and Toyoshima 2006), but the bulk of our documentation, aside from a set of land contracts studied by Li (2010), consists of seventeenth-century Dutch sources: (a) translations into a dialect of Siraya of the Gospel of St. Matthew and the Heidelberg Catechism, both by Daniel Gravius (1661); and (b), in a markedly different dialect, a Dutch–Siraya word-list known as the “Utrecht Manuscript” (UM), plus four short dialogs between schoolchildren, both by anonymous authors. Of these two, the Gospel and Catechism dialect is the more extensively documented: it is primarily that dialect that Adelaar’s book aims at restoring, although he makes reference to the Utrecht Manuscript dialect when the need arises.

Adelaar has been working on seventeenth-century Siraya for many years, his first publications dating back to the 1990s (1994, 1997). In this enterprise, his native knowledge of Dutch and understanding of the history of Dutch phonology and European orthographic practices are valuable assets: he notes (6) that the spelling in the Gospel shows the hand of several editors following different orthographic traditions, this being compounded by widespread spelling inconsistencies. The 400-page book under review is currently the single most complete resource on the Siraya language, although it does not include a historical phonology. It consists of four parts: an introduction, a grammatical analysis, a glossed edition of ten chapters (out of twenty-eight) of St. Matthew, and a Siraya–English lexicon.

The introduction, 15 pages altogether, provides relevant background information on the sources and literature on Siraya, dialect variation, Formosan languages and Austronesian prehistory, the historical setting of Dutch–Siraya language contact, the history and ethnography of the Siraya people, and so on. Adelaar shows that Siraya dialect variation must have been much broader than the Gospel and UM dialects, but he is agnostic on the question of whether the Taivuan and Makatau groups, distinguished by Tsuchida, Yamada, and Moriguchi (1991) on the basis of word-lists collected by early twentieth-century Japanese scholars, are dialects of Siraya or different languages altogether.

The notes on the authorship of the Siraya Gospel (6–7) rely on a 1931 Leiden University PhD thesis by W. A. Ginsel: Daniel Gravius, the official author, may have put together the final version of a text elaborated by his predecessors. What is not entirely clear is to what extent the translation of the Gospel of St. Matthew represents native Siraya or displays features resulting from imperfect learning by the Dutch ministers.

In matters of subgrouping (8–9), Adelaar follows Blust (2009), where Siraya is placed together with Basai-Trobiawan, Kavalan, and Amis within an “East Formosan group” on the basis of the merger, which these languages uniquely share, of two PAn phonemes: *j and *n. In Blust’s interpretation, *j was a voiced palatalized velar stop [gʲ] while *n was an alveolar nasal. The merger of the two...