The development of Global Englishes (GE) has urged language policy makers and language practitioners to re-address how English, as a lingua franca (ELF), should be taught and assessed to meet the various needs and goals of language learners. However, traditional assessment models for English language teaching (ELT) still adopt a native-oriented approach in many settings. This paper analyzes a speaking test from an intercultural communication course taught by both Chinese and international teachers of English at a Chinese university. The setting is regarded as a quasi-ELF context because students communicate in English with teachers of different L1s in both the classroom and extra-curricular activities. Students having the same L1s were assessed by a group oral test in which they discuss content-based topics. In this study, the researchers first analyzed the rubric used in the oral test and observed how the students used English to negotiate meaning in their discussion. The researchers then retrieved the discussion scores assigned by the teachers and interviewed the teachers about how they assessed the students' work. Linking to the rubric, the researchers also aimed to investigate the teachers' understandings and perceptions of GE when assessing the students' discussion. The findings show that although many teachers have heard of the notion of GE, it is not yet well-represented in English language assessment. This may be due in part to the top down language policy and the standardized forms of language assessment. This paper therefore argues for the importance of going beyond a native-oriented approach to assess language in practice for language assessment. This requires that both language policy makers and practitioners understand the complexity of English use around the globe and raise awareness of ELF in the near future.