This thesis reports on findings from a study of sociolinguistic variation in second language speakers of English in New Zealand. The study combines quantitative methods of acoustic analysis and experimental design with qualitative methods of semi-structured interviews and content analysis. The study focuses on second language speakers’ variation in ‘passing for a native speaker’, that is, being regarded as a first language speaker. Variation in passing is explored from the perspectives of variation in production and perception. 18 second language speakers of English (first language Korean and German) and 6 first language speakers of English were recorded in four different settings (family, friends, services, and university). In the production study, the second language speakers’ monophthongal vowels are analyzed in comparison with the first language vowels and New Zealand English ones. The speakers were found to style-shift in their production of the first and second formants of certain vowels in different settings: the German speakers were more English-target-like in the services setting and the Korean speakers were more English-target-like in the services setting and less English-target-like in the family setting compared to the university one, exhibiting a continuum of native-likeness in the three settings. Three perception experiments complement the production analysis. Two of these focus on the effect of setting in accentedness perception and passing for a native speaker, and one explores the effect of social information (namely, ethnicity) on accentedness perception. The speakers were found to receive a different accentedness rating depending on the recording setting and whether or not the listener was aware of their ethnicity. Specifically, some speakers were rated less accented in the services setting and some in the family setting compared to the university one. Also, Asian speakers were rated similarly for accentedness both when the listeners were provided with video input and when they were not, but Caucasian speakers were rated more accented when the video input was available. Additionally, the thesis addresses passing for a native speaker of different English varieties in an experimental context. It reveals interesting trends in the speakers’ variation of passing in different settings and passing for native speakers of different varieties. The family setting was conducive to passing, and some speakers passed for a native speaker of the same variety more often than for a native speaker of other varieties and some vice versa. Finally, the second language speakers’ beliefs about passing and listeners’ comments on their decision-making in identifying the origin of the speakers are investigated. The results showed that the speakers believed that first (and short) encounters with strangers were conducive to passing. A variety of linguistic and extralinguistic listener comments was revealed. Taken together, the results paint a complex picture of variation in second language speakers’ production, accentedness perception, and passing for a native speaker. The findings suggest that speakers vary in their production according to audience and in the construction of their identities. The perception experiments highlight the effect of listener expectation on their perception. These results have implications for how we understand sociolinguistic variation in second language speakers.