Automatic assessment of spoken interaction in second language (AASIS)

The construct of spoken interaction traditionally deployed in assessment entails acoustically discernable pragmatic competences, while visual cues, such as gazes, mimes and movements are rarely included in rating scales used by human raters of interactive tasks despite their crucial role in human encounters In AASIS we focus on adult students of L2 Finnish from diverse backgrounds. First, we address the verbal and non-verbal features of interaction competence in L2 Finnish dialogues that affect human raters’ assessments of L2 interaction skills, and second, we develop the ways to measure these features in ASR-based assessment. We will develop automatic speech recognition (ASR) for second language (L2) learners’ speech that extends to recognize also the sounds and visuals of non-verbal communication. AASIS improves the alignment between teaching, learning and assessment of speaking in educational settings and a better correspondence between everyday language use and testing.