InterSpeech 2021

A Preliminary Study on Discourse Prosody Encoding in L1 and L2 English Spontaneous Narratives
(3 minutes introduction)

Yuqing Zhang (BLCU, China), Zhu Li (BLCU, China), Binghuai Lin (Tencent, China), Jinsong Zhang (BLCU, China)
Relatively little attention has been devoted to the discourse-level prosodic encoding and speech planning in second language (L2) speech. This study reports a preliminary study on learners’ discourse prosody encoding pattern and makes a comparison with that of native speakers. Using a corpus of spontaneously produced picture story narratives, we analyzed general characteristics of prosodic units (PUs) and explored relationships between pitch encoding (cross-boundary f0 heights and f0 slopes) of PUs and the semantic completeness of PUs in English spontaneous speech by native speakers, beginning learners, and advanced learners. The results indicated that beginning learners showed neither sensitivity to semantic units in discourse (DUs) in their f0 encoding nor distinct signs of pitch-related preplanning based on DUs, suggesting improper phrasing of the least proficient non-native speakers. Both native speakers and advanced learners were sensitive to the initiation and termination of DUs in their prosodic encoding; however, only native speakers showed clear signs of DU-based preplanning. We argue that the observed between-group differences in L1 and L2 speech might be attributed to differences in the scope of speech planning, i.e., compared with native speakers, who mostly produce complete semantic units, learners’ speech is produced step by step with pauses between phrases.