Effect of semantic contexts on Cantonese tone perception

Abstract

PURPOSE: This study examined the effect of congruent semantic contexts and incongruent semantic contexts on the perception of naturally-produced ambiguous Cantonese lexical tones. METHODS: Eighteen monosyllabic words differed only in tone and produced in isolation by two speakers were used to replace the final syllable of four-syllable utterances in three different semantic conditions. In the first condition, the first three syllables formed carrier phrases that were semantically neutral to the final syllable (e.g., 跟住聽市“Next, listen to …”) In the second condition, the first three syllables formed an idiomatic expression with the last syllable (e.g., 國際城市 “international city”). In the third condition, the last syllable of an idiomatic expression was replaced by another syllable that differed from the target syllable by the tone, forming an unexpected non-sense utterance (e.g., 國際城思 “international + nonsense word”). Twenty native Cantonese speakers listened to and repeated the 144 productions. Two phonetically-trained experimenters independently coded the tones they produced and checked for reliability. RESULTS: Tones produced in isolation were identified with ceiling accuracy. However, when these productions were attached to utterance final position, tone identification accuracy, particularly accuracy of the level tones, was significantly lower. Investigation of the accuracy of level tones in the semantically neutral, semantically congruent and semantically incongruent conditions revealed that the accuracy in semantically congruent contexts was higher than in semantically neutral or incongruent contexts. CONCLUSION: Perception of tone height is affected by prosodic contexts. When the tone identity is ambiguous, semantic contexts that are congruent and highly associated with the target syllable are facilitative in tone recognition. Semantically incongruent contexts do not have negative effect for the identification of ambiguous tones. The results suggest that the use of semantic contexts can be a strategy adopted in aural rehabilitation to facilitate identification of lexical tones.