Sequential Voicing in Old Japanese

Abstract

This study tackles sequential voicing in Old Japanese with a focus on three matters: (a) the interaction of the eight vowels and aspirated consonants; (b) the association of the written system and sequential voicing; and (c) the interaction between the combinationality of each constituent and sequential voicing. Four hundred and seventy-two compound nouns of Old Japanese were collected from the corpus ‘The Japanese Lexicon: A Rendaku Encyclopedia’ by NINJAL. The findings reveal that (i) /k/ has the largest token number of sequential voicing and /p/ has the second largest token number, followed by /s/ and /t/; (ii) regarding the eight vowels /a/, /e₁/, /e₂/, /i₁/, /i₂/, /o₁/, /o₂/ and /u/, /a/ is most likely to form a [N₁-N₂] whose initial consonant is /k/, /p/ and /t/. It is not likely for the vowel /a/ to invite a voiced ‘/s/-initial’ N₁’; /o₁/ and /o₂/ are both likely to combine with a voiced consonant /k/; /t/ and are less likely to yield a voiced /s/; /e₁/ is more likely to invite a voiced consonant than /e₂/, but /e₁/ does not invite a voiced /p/; /e₂/ does not yield a voiced /s/; and /i₁/ is likely to take a voiced consonant than /i₂/. /k/ and /t/ are the two consonants that are most likely to be voiced when forming a N-N with N₁ ends with /i₁/. /i₂/ does not invite a voiced /p/, /s/ or /t/; /u/ never results in a voiced /s/; (iii) there is a split in the characters that renders a voiced phoneme or an unvoiced phoneme; and (iv) the semantic relationship of N₁ and N₂ in [N₁-N₂] that bears sequential voicing is of six types, of which the most frequent relationship of N₁ and N₂ is [Modifier - N₂]. The [prefix-N₂] construction is not subject to sequential voicing.