Native speakers of Japanese and Korean identified 6 American English monophthongs /i, ɪ, ɛ, ae, ɑ, ʌ/ and discriminated 6 vowel pairs /i/-/ɪ/, /ɛ/-/ɪ/, /ae/-/ɛ/, /ae/-/ɑ/, /ae/-/ʌ/, /ɑ/-/ʌ/ in in /hVt/, /pVt/, /pVn/ and /pVl/ frames. The two groups of listeners' identification and discrimination accuracy were submitted to 2 mixed-design ANOVAs, respectively, with 2 Listener Groups as a between-subject variable, and 6 frames and 6 vowels (or vowel pairs) as within-subject variables. As for identification, a main effect of vowel and frame are both significant (p<.001), and a three-way interaction of listener groups × vowels × frames is also significant (p<.001). Post-hoc pair-wise comparisons revealed that Korean listeners identified /ʌ/ significantly better than Japanese listeners. This is probably because Korean listeners can equate the English vowel to Korean /ʌ/. Japanese listeners' identification accuracy of /i/ is lower before /n/ and /l/, but that of Korean listeners was not affected by the context. As for discrimination, a main effect of the frames and vowel pairs was significant (p<.001). A three-way interaction of the 3 factors was not significant, but all the two-way interactions were significant (p<.001). Post-hoc pair-wise comparisons revealed that Japanese listeners' discrimination accuracy of /i/-/ɪ/ and /ae/-/ɛ/ is lower before /n/ and /l/ than before /t/, while that of Korean listeners was not affected by the context. Korean listeners discriminated /ae/-/ɛ/ least accurately of all the six vowel pairs.