The past decade has brought increased legislative pressure to assure English language “fluency” on the part of nonnative‐English‐speaking teaching assistants (NNS TAs). However, no consensus has been reached in second language assessment about the appropriate instruments to evaluate the spoken language proficiency of NNS TAs. We see as a major issue the fact that, as a field, we have not articulated the theoretical grounds on which to evaluate the merit of assessment instruments as they are used in NNS TA testing. In this paper, we use Bachman's (1990, 1991) framework of language testing and, in particular, his standard of “authenticity” for language testing instruments to evaluate three instruments of spoken language assessment— the SPEAK (Spoken Proficiency English Assessment Kit) test, the OPI (Oral Proficiency Interview), and a performance test—in terms of their “authenticity” as language tests for NNS TAs. We first look at the three test instruments as contexts for language use by examining what Bachman terms theirtest method facets. We then examine discourse elicited by the test instruments and consider the discourse differences arising from these differences in test context. Finally, we consider the three test instruments in relation to features of language use in the target context and draw conclusions regarding the authenticity of the instruments i