This thesis seeks answers for two research questions: (1) how much information is needed for listeners to detect a foreign accent, especially in slightly accented speech? and (2) how does the listener's linguistic experience affect the detection of foreign accent? To investigate these questions, three listener groups with different L2 experiences (i.e., Native English monolinguals, Korean-English bilinguals, and Korean learners of English) were tested with relatively short English words (i.e., vowel /α/, monosyllabic, and disyllabic words) produced with and without slight accents, in a foreign accent detection task. The monosyllabic corpus included stimuli with different syllable structures (i.e., CV, CCV, CVd, and CCVd) as well as various segments, to test the influence of L1 phonotactics in foreign accent detection.

The results show that the native listeners did not need much information to detect foreign accent. They were able to detect a foreign accent even from hearing the vowel /α/, though they could do it more easily from hearing longer stimuli. The results also suggest that the L1 phonotactics is closely related to the listeners' sensitivity to a foreign accent, because the listeners exhibited differential sensitivity to the stimuli with “new” versus “similar” segments in terms of L1 phonotactics.

Regarding the listeners' linguistic experience, the results indicate that more L2 experience leads to higher sensitivity levels to a foreign accent. The analysis of the listeners' sensitivity and their linguistic experience further suggests that the age of first exposure to a massive L2 speaking environment plays a more important role than the length of residence in the L2 environment in detecting foreign accent. Nevertheless, even the non-native listeners with early exposure to the L2 environment do not have an identical perceptual system as that of monolingual native listeners.

In addition to these findings, this thesis proposes a perceptual model for foreign accent detection among listeners with different L2 experiences: listeners with limited L2 experience may use L1 categories rather than L2 categories as the norm in foreign accent detection, while listeners with enough L2 experience use L2 categories, leading to surprisingly better foreign accent detection by listeners with limited L2 experience.