Individual and shared assessment of ESL students’ paired oral test performance: Examining rater judgments and lexico-grammatical features

Reflecting the importance of interaction in second language (L2) learning and teaching, standardized proficiency tests have adopted interactive speaking tasks for assessing oral proficiency. For example, in the IELTS speaking test, candidates answer questions and discuss topics with an examiner, while the Cambridge speaking tests ask candidates to interact with an examiner as well as with another test-taker. However, questions have been raised about the extent to which examiners play a neutral role in the interaction as well as the scoring consistency across examiners (He and Young 1998; Lazaraton 2002; Morton, Wigglesworth and Williams 1997; Nakatsuhara 2008). Because an examiner leads and controls the interaction, the resulting discourse structure can be asymmetric (i.e., low equality), which reflects imbalance in the power relationship between the examiner and the candidate (Lazaraton 1996; Ross and Berwick 1992; van Lier 1989; Young and Milanovic 1992). In addition, comparisons between authentic conversations and interview tests have revealed that examiner-candidate interaction shows greater similarity to institutional talk than naturalistic conversation (van Lier 1989). Therefore, concerns have been raised about the authenticity of oral interviews with examiners commonly used for oral proficiency assessment.