霧台魯凱語元音串及重疊詞與音韻之互動研究

This study examines vowel cluster repair in Budai Rukai, using 1,298 tokens from 288 hours of elicitation. Three main strategies are attested: glide formation, vowel coalescence, and insertion of glottal stops or glides. Glide formation typically applies when the first vowel is high; coalescence is common with identical or certain heterogeneous vowels; glottal stop insertion occurs more with non-high vowels. Repair patterns are shaped by four interacting factors. Positional effect is primary: word-final clusters often remain disyllabic, with high vowels taking glide insertion (e.g., /ia/ → [i.ja]) and non-high vowels taking glottal stop insertion (e.g., /?a/ → [?.?a]); non-final clusters usually become monosyllabic, often with glide formation. Prosodic constraints also influence outcomes: avoidance of CGVG syllable structure is consistent and exceptionless, while avoidance of uneven trochaic feet applies only in certain environments, often with pa-, ma-, or wa- prefixes, and admits exceptions. Morphological source matters: clusters from pa-, ma-, or wa- may remain disyllabic in non-final positions, while those from ka-, la-, mu-, or ki- tend toward monosyllabicity; suffix-derived clusters behave like root-internal ones. Vowel quality is position-sensitive across all cluster types: heterogeneous clusters often remain disyllabic word-finally but monosyllabize elsewhere; identical clusters (/uu, ii, aa/) usually coalesce, with /aa/ retaining disyllabicity in specific contexts; /u?/ simplifies to [u]; /ua/ and /au/ follow general patterns, though /au/ word-finally always glides to [aw], with repair influenced by morphology. Overall, vowel cluster repair in Budai Rukai reflects the interaction of positional, prosodic, morphological, and segmental factors, with CGVG avoidance emerging as the most stable prosodic principle.