摘要
語言的區分條件可以包括音韻、詞彙、構詞及語法幾方面的現象。觀察泰雅語與賽德語的音韻差異有六條，但泰雅語的萬大方言卻有四個音韻條件與賽德語相合，因此只剩下兩條勉強可用來區分泰雅語與賽德語。這兩種語言的詞彙差異遠比音韻的差異顯著。在約800個詞項當中，萬大有214個與其他泰雅方言相同，卻只有11個與賽德語相同。此外，萬大有50個獨特的詞項，跟這兩種語言都不同。因此，詞彙的證據顯示萬大為泰雅方言，殆無疑問。
土田滋先生曾提出三個條件來區分泰雅語的賽考利克方言群與澤敖利方言群，包括部分的音韻、構詞、詞彙的差異。本文支持他的觀點並且舉出更多的例證，包括22個詞項以示賽考利克、澤敖利、賽德語三者的不同。至於語法的不同，就有待進一步的研究了。
The reconstruction of Archaic Chinese *-b, *-d, and *-g is based on three main types of evidence: (1) rhymes in the Shih Ching, (2) hsieh sheng (phonetic compounds), (3) double readings in Ancient Chinese. However, there are problems in reconstructing *-b: first, it is not attested in the rhyming of the Shih Ching; second, there are only a few cases indicating close contact between *-b and *-p. The former problem can be explained by the fact that hsieh sheng represents an earlier stage than the Shih Ching and the change *-b＞*-d took place between these two periods. One purpose of this paper is to account for the latter problem.
Archaic Chinese *-b is rare, as is Proto-Austronesian *-b and Proto-Indo-European *b. Only a dozen reconstructed forms end with PAN *-b (Dempwolff 1934-38), and there are only a few cases of PIE *b in any position. The rare occurrence of *-b is thus a universal feature. Voiced stops are phonetically more complex than their corresponding voiceless stops, and are more susceptible to change than the latter, especially when they occur in the word- or syllable-final position. The reconstruction of *-b is partly based on the overall pattern of relationships in the sound systems. Parallel developments are shown in the different language families in this paper.
The reconstruction of Archaic Chinese *-b is supported with three types of evidence: (1) hsieh sheng, (2) double readings and (3) common etymology. Some supporting evidence is provided by a comparison of similar phenomena in the other languages.