ABSTRACT
One of the goals of second language (L2) speech research is to identify aspects of talker and listener behaviour that both facilitate and detract from effective communication. To achieve this, researchers elicit audio-perceptual (AP) evaluations of global properties of L2 speech – including accentedness, comprehensibility and intelligibility – from audiences of listeners. Substantial empirical evidence confirms the reliability of such data and points to the validity of the dimensions at issue. However, the AP approach clearly has limitations, and caution must be observed in interpreting judgement data. While ongoing research has helped resolve a number of methodological concerns, some questions about comprehensibility measurement have yet to be addressed, and confusions about the distinction between this construct and the other two still persist. To grasp the differences, it is essential to understand speech comprehensibility as a type of processing fluency, a concept widely employed in perceptual psychology that refers to a perceiver’s cognitive management of a stimulus. The ease with which a listener understands an utterance (comprehensibility) must be carefully distinguished from the amount of understanding that has actually taken place (intelligibility). Fortunately, well-established measurement techniques are available for assessing both accentedness and intelligibility. However, the methodology of comprehensibility assessment has received less attention. It has not been demonstrated, for instance, whether the usual approach of eliciting listener ratings on an equal-interval scale is suitable. To address this gap, the present chapter presents a brief study in which a nine-point linear rating was found to capture the dimension at least as well as other types of scaling.
