「日本語レキシコン-連濁事典の編纂」研究発表会 概要

プロジェクト名
日本語レキシコン-連濁事典の編纂 (略称 : 連濁事典)
リーダー名
Timothy J. VANCE (国立国語研究所 理論・構造研究系 教授)
開催期日
平成26年6月14日 (土) 16:00~17:30
開催場所
国立国語研究所 3階 セミナー室 (東京都立川市緑町10-2)

発表者氏名・発表テーマ

"Shuri-Okinawan Rendaku"Leon A. SERAFIM (University of Hawaii, retired)

In many respects Shuri rendaku (R) is like the phenomenon with the same name in Japanese. In both, rendaku cannot be characterized as automatic - that is, in both, there are exceptions to the rule. In both, Lyman's Law (LL) characterizes some cases (i.e. voiced obstruent in the second element) where rendaku fails to appear. In both, there appear to be lexically governed conditions, either requiring rendaku or requiring its non-application.

In Okinawan, it has been claimed that, largely, rendaku does not affect the moras Sh/ši/ or Sh/fi/. However the Sh/fi/ generalization has too many exceptions, and the Sh/ši/ generalization may be true, but weak. Thus it may be better to characterize both sets of exceptions to rendaku as governed by the lexical items themselves, as second terms in their compounds.

A (seeming-)exception to Lyman's Law exists in a handful of compounds: a second element with a voiced obstruent (v'ed obst.) non-initial mora undergoes a change to a mora nasal (e.g. *Sh/bu/ > Sh/N/), thus depriving that element of the Lyman's-Law-required term. The result is that Lyman's Law does not operate and the compound does undergo rendaku:

*maye+kabuto 'bobbed+hair'(*maye lit. 'front'; x*...+gabuto [violation of LL])
*mai+kaNtu (2nd element lacks v'ed obst.; new candidate for R)
meH+gaNtu (R; no violation of LL)

While strictly speaking not rendaku itself, a phenomenon of unusually large numbers of initial voiced obstruents in (generally) natural-world-related nouns in native Okinawan words recalls the set of words with 2.5 accent in Japanese language history. This class may have gotten its start as a set of second elements of rendaku compounds (Sh/daki/ 'bamboo' ← Sh/X-"+taki/ 'X+bamboo'). This generalization that natural world items begin with voiced obstruents must have extended well beyond its origin as a mere calving-off of rendaku-affected items. At least that is what the single noun Sh/guǰira/ 'whale', suggests: even as a second element in a compound, it violates Lyman's Law, in which a second element Sh/ǰ/ should have prevented the voicing of Sh/k/; cf. J /kuǰira/ < OJ /kudira/ '(id.)'.