Our results indicate that native Mandarin listeners could understand the English minimal pairs produced by native Mandarin speakers better than native English and EnglishMandarin bilingual listeners did. This is consistent with the findings in Bent and Bradlow  and Hayes-Harb et al , which indicates that ISIB does exist for Mandarin. Nevertheless, such ISIB was not found across all consonant and vowel contrasts, which was not explored by previous studies before. Previous studies on ISIB adopted a broader approach and did not focus on specific consonant and vowel contrasts. For example, Bent and Bradlow  checked listeners’ transcriptions of simple sentences and counted the correct transcription of key words. Major et al  examined ISIB by testing listeners’ comprehension of short lectures. As pointed out by Hayes-Harb et al. , when listeners were presented with sentences, a mixture of factors might be at play, some of which may be irrelevant to ISIB. Hayes-Harb et al. , Bent et al.  and Smith et al.  used isolated words as stimuli, and studied word-final stops but none examined specific consonant contrasts and none included any vowel contrasts on ISIB. · Our study incorporated both types of segmental contrasts in the experiment and has shown that even within the same segmental type, intelligibility could also differ. Sharing the same phonetic and phonological system between the speakers and listeners does not guarantee intelligibility in an across-the-board manner. It is particularly revealing to find that even NM listeners could perform at chance level for some contrasts. Future studies on ISIB should investigate which contrasts are more conducive to ISIB and which are not, and whether ISIB is language-specific. The main research question of this study is whether ISIB can be extended to bilingual L2 learners who also share the phonological system which gave rise to ISIB. Our results show that BI listeners performed similarly with NE listeners who did not know Mandarin at all, which gave a negative answer to the research question. However, before we can firmly reject this hypothesis, there are some factors which we should re-consider. First, although the BI listeners had plenty of exposure to Mandarin, they did not have much exposure to Mandarin-accented English. Being native English speakers, they might have relied more on their native English knowledge to distinguish the minimal pairs in the experiment, rather than transferring their learned Mandarin knowledge to understand Mandarin-accented English. Second, the proficiency level of the bilingual learners may also play a role. The BI listeners may need to have an even higher Mandarin proficiency level in order to benefit from the ISIB. Third, the sample size of the listener groups (n = 9) may be too small to yield significant differences. This was further aggravated by the fact that there was more individual variation in the BI group than the NE group, as evident in the larger standard deviations in both accuracy (Table 1) and confidence level (Table 2) of the BI group. Whatever the reason, our results indicate that simply sharing a phonological system may not be enough to give rise to ISIB. More studies with bilingual L2 learners are needed to explore the complex mechanisms behind ISIB. Finally, it will be interesting to further investigate Mismatched Interlanguage Speech Intelligbility Benefit (MISIB) for Cantonese speakers/listeners. MISIB refers to the ISIB found when non-native listeners and non-native talkers do not share the same native language. A good example of MISIB can be found in Bent and Bradlow  where Mandarin listeners found high-proficiency Korean talkers more intelligible than native English talkers despite their difference in LI. As the features of the consonant and vowel contrasts of Mandarin-accented English used in this study are also found in Hong Kong English , it is quite possible that MISIB can be found for Cantonese listeners, but so far no study has investigated this possibility. Moreover, Mok, Setter and Low  have shown that juncture pairs in Hong Kong English are more intelligible than those in Singapore and British English. It will be useful to compare intelligibility of Mandarin-accented English and Hong Kong English with different listener groups. To conclude, our preliminary study has confirmed the existence of ISIB for Mandarin as shown in previous studies. We have also shown that individual segmental contrasts can differ in intelligibility. Our results call for further investigation of whether ISIB can be extended to situations where the talker and listeners do not match in terms of their native language, but share the same phonological system through L2 learning. Possible factors that can affect ISIB for bilingual L2 learners were discussed. It is hoped that our study has shed some new light on the interesting topic of intelligibility of non-native speech.