Levels of lexical stress contrast in english and their realization by L1 and L2 speakers

Following our previous extended paradigm of Pairwise Variability Index (PVI) to analyze both F0 and duration, we analyze and compare primary, secondary and tertiary lexical stress in English as it is realized by L1 English, L1 Taiwan Mandarin and L1 Beijing Mandarin speakers. Chao's tone letter system was also adapted to describe differences in degree of stress contrast produced by L1 and L2 speakers. 20 multisyllabic English words of varying stress patterns produced by 32 speakers were selected for analysis. Results indicate that: 1) the most acoustically robust stress contrast in L1 English is found between primary stress and secondary/tertiary stress; the distinction between secondary and tertiary stress is not strong enough to be categorical; 2) a substantial difference in degree of contrast realization between L1 and L2 speakers exists with respect to F0, but not duration; 3) words whose stress patterns were consistently transcribed across dictionaries require a five-point scale to distinguish native from non-native speakers' realization of English lexical stress, whereas for inconsistently described words, a three-point scale is sufficient. The results suggest that for L1 speakers, robust contrast of lexical stress/no-stress differentiation is critical, a 2-way contrast represented by a simple three-point scale; whereas more fine-grained secondary/tertiary differentiation, which requires more distinct 3-way contrasts as reflected in a five-point scale is less crucial. L2 speakers' failure to achieve the same differentiation patterns, as exemplified by their production data, is predicted to have a strong effect on the naturalness and intelligibility of L2 speech.