LEXICAL EFFECTS IN REAL-WORD AND NONWORD SPEECH PERCEPTION: EVIDENCE FROM L1 MANDARIN–L2 ENGLISH SPEAKERS WITH DIFFERENT LEXICAL KNOWLEDGE

This study investigates how phonetic category structure and lexical information interact in second language (L2) speech perception, with a focus on the Ganong effect in L1 Mandarin–L2 English bilinguals. While previous research has established the presence of lexical bias in both native and non-native listeners, relatively few studies have examined the magnitude of the lexical effect or compared how it varies across phonetic categories along a perceptual continuum. The hypotheses test whether the TRACE model or the SLM/SLM-r models better predict the factors influencing changes in the magnitude of the lexical effect as lexical knowledge varies. Thirty L1 Mandarin–L2 English speakers participated in a phoneme identification task involving three word pairs, including a nonword baseline and two real word–nonword conditions. Lexical knowledge and years of immersion (YoI) were assessed using a vocabulary test and a language background questionnaire. The results showed no statistically significant correlation between the magnitude of the lexical effect and lexical knowledge. However, the results revealed a clear asymmetry in lexical effects across phonetic categories. A significant lexical shift was observed in the /p/ condition (beace–peace), while the /b/ condition (beef–peef) showed weaker and more variable effects. These findings suggest that lexical influence in bilingual speech perception is constrained by the condition of underlying phonetic categories. Building on this, the study proposes an integrated framework in which bottom-up phonetic activation and top-down lexical feedback interact dynamically.