Abstract

Abstract: Automatic scoring of language proficiency is becoming a point of growing importance because the number of English-as-a-second-language learners has been steadily increasing worldwide. Two common issues in this field are: a) the lack of publicly available data specifically designed and annotated for automatic assessment, especially as regards spoken data, and b) the lack of consistency and coherence in human assessment. Although second language proficiency cannot be assessed on the mere basis of the presence of errors, this aspect is highly consistent. In this talk, we will present an approach to automatically predict the scores of spoken language responses of ESL learners (TLT-school corpus and ICNALE corpus) leveraging written data (EFCAMDAT corpus) and exploiting the presence of grammatical errors modelled across 5 CEFR proficiency levels thus tackling both the aforementioned problems.

Bio: Stefano Bannò is a second-year PhD student in Cognitive Science at University of Trento in a joint programme with the SpeechTeK research unit of Fondazione Bruno Kessler (FBK) with a project on automatic assessment of spoken language proficiency. Before starting the doctoral course, he obtained a MA in Philology and Linguistics and a BA in Classics at University of Trento. During his MA, he spent a research period at the Lautarchiv of Humboldt University in Berlin. Besides his academic career, he has worked as a musician and a secondary school teacher. He is currently working at the Department of Engineering at Cambridge University as a visiting student.