This study examined whether native-language (L1) phonotactic constraints affect second-language (L2) speech segmentation and, if so, whether L2 proficiency influences word segmentation strategy. L2 speakers at different English proficiency levels and native English speakers were instructed to spot consonant–vowel (CV) words (e.g., pea) and consonant–vowel–consonant (CVC) words (e.g., peak) in bisyllabic nonwords. All groups significantly missed CV targets in the nonwords with clear boundaries that leave illegal residue syllables. Native Japanese speakers did not use L1 phonotactics to segment L2 words but syllabified them while being affected by possible-word constraints.

How learners pronounce the vowels before the syllabic nasal is investigated in this study. Acoustic analysis of what is being pronounced by the learners was examined by an evaluation experiment. When the first and second formants of the vowels recorded in the production experiment were measured, it was found that the distributions of Chinese learners and native Japanese speakers were different. Especially for learners, the distributions tended to overlap between /u/ and /e/. The distribution range of /a/ and /i/ was almost the same as that of native speakers of Japanese, regardless of their Japanese learning history. Next, we conducted an evaluation experiment on the naturalness of pronunciation by 37 native speakers of Japanese. As a result, among the vowels that preceded the learner’s syllabic nasal sound, /u/ had the lowest naturalness regardless of the Japanese learning history. In the results of acoustic analysis and evaluation experiments, it was found that the longer the learning history, the closer the pronunciation was to the distribution of native Japanese speakers and the higher the degree of naturalness. The vowels that precede the syllabic nasal produced by Chinese learners may be interfered with by their first language, and in particular /u/, /e/, /o/ may have different sound qualities from native Japanese speakers. This was confirmed from the acoustic and perceptual aspects of the experiments.