Abstract:

This paper argues that test criteria for oral assessment tend to be biased towards features more easily captured by written performance than spoken. This argument is carried forward at different levels reflecting the typical areas articulated in evaluation criteria for tests of speaking: discourse management; fluency; lexis and grammar. The paper draws on insights emerging from corpus linguistics, speech processing and sociophonetic studies to suggest that key features of oral communication may be undervalued in traditional approaches to oral assessment. Conversely, linguistic features which are more typical of written mode (for example, wide lexical range) and which are difficult for the native speaker to produce in spontaneous speech are valued positively in second language tests and regarded as key indicators of proficient speaking ability.

Author: Rebecca Hughes