How Different Types of Linguistic Information Impact Voice Perception: Evidence From the Language-Familiarity Effect

• Published in: Language and speech 2023-12, Vol.66 (4), p.1007-1029
• Main Authors: Yu, Keke, Zhou, Yacong, Zhang, Linjun, Li, Li, Li, Ping, Wang, Ruiming
• Format: Article
• Language: English
• Subjects: Acoustics; Experiments; Familiarity; Identification; Indonesian; Lexical semantics; Mandarin; Phonology; Semantics; Semiotics; Sentence structure; Sentences; Speaker identification; Speech perception; Voice perception
• ISSN: 0023-8309; 1756-6053
• DOI: 10.1177/00238309221143062

Previous studies have suggested the effect of linguistic information on voice perception (e.g., the language-familiarity effect [LFE]). However, it remains unclear which type of specific information in speech contributes to voice perception, including acoustic, phonological, lexical, and semantic information. It is also underexamined whether the roles of these different types of information are modulated by the experimental paradigm (speaker discrimination vs. speaker identification). In this study, we conducted two experiments to investigate these issues regarding LFEs. Experiment 1 examined the roles of acoustic and phonological information in speaker discrimination and identification with forward and time-reversed Mandarin and Indonesian sentences. Experiment 2 further identified the roles of phonological, lexical, and semantic information with forward, word-scrambled, and reconstructed (consisting of pseudo-Mandarin words) Mandarin and forward Indonesian sentences. For Mandarin-only participants, in Experiment 1, speaker discrimination was more accurate for forward than reversed sentences, but there was no LFE in either sentence. Speaker identification was also more accurate for forward than reversed sentences, whereas there was an LFE for forward sentences. In Experiment 2, speaker discrimination was better for word-scrambled than reconstructed Mandarin sentences. Speaker identification was more accurate for forward and word-scrambled Mandarin sentences but less accurate for Mandarin reconstructed and forward Indonesian sentences. In general, the pattern of the results for Indonesian learners was the same as that for Mandarin-only speakers. These results suggest that different kinds of information support speaker discrimination and identification in native and unfamiliar languages. The LFE in speaker identification depends on both phonological and lexical information.