The present study seeks to elucidate the nature of first language transfer effects on the perception of English lexical stress in different intonational contexts. Pitch (F0) is one of the acoustic cues to lexical stress in English, but crucially, pitch can only serve as a cue to stress when interpreted within the intonation system of English. In contrast, Mandarin is a tonal language where pitch carries greater functional weight for signaling lexical contrasts. Associating pitch to lexical contrasts, Mandarin listeners might therefore assume a one-to-one relationship between pitch and lexical stress in English. To examine whether Mandarin learners of English make this assumption, the current study tested their stress perception in four different intonations: H*L-L%, L*H-H%, H*H-H%, and L*L-L%, where F0 cues to stress were realized differently in each intonation. The results of the stress identification task showed that Mandarin listeners falsely associated higher F0 with stress and lower F0 with the absence of it, and their performance did not improve as they became increasingly proficient in English. The findings provide corroborating evidence for the Cue-Weighting Transfer Hypothesis by showing that the use of a suprasegmental cue can transfer from one phonological category (tones) to another (stress).

This study investigates whether high-variability (multi-talker) perceptual training is superior to low-variability (single-talker) perceptual training for enhancing Seoul-Korean listeners’ use of vowel quality and pitch cues to English lexical stress. Vowel quality and pitch are the two most important cues to English lexical stress in accented words (Tremblay et al., 2021). Seoul Korean does not have lexical stress, lexical pitch accents, or lexical tones. Seoul-Korean listeners completed a pre-/post-test sequence-recall experiment where they heard and recalled four English words uttered by different talkers. Experimental conditions: The words differed in lexical stress, which was cued by: (i) vowel quality, pitch, and duration; (ii) vowel quality and duration; (iii) pitch and duration; or (iv) duration. Control condition: The words differed in their initial consonant (Seoul-Korean contrast). Listeners completed one of two training types (8 × 30-mins). The words in the training followed a cue distribution that mimicked naturalistic spoken English (Im et al., 2018). Results: Both training types improved listeners’ recall for (i)–(iii), suggesting enhanced use of vowel quality and pitch. Learning gains were greater with high-variability training, but training type did not interact with cue. Thus, high variability provides a global rather than selective enhancement of listeners’ use of cues to lexical stress.