Identification of input text characteristics that interfere with comprehension of second language listening texts

Abstract

Historically, computer-based second language help options (translations, culture notes, translations and glossaries) have been ignored by L2 listeners, despite that their use seems to improve comprehension. CALL scholars agree that training is paramount to address the help option neglect phenomena. The way this training should take, however, is still unclear. Some researchers suggest training should be done through the implementation of Interactive Virtual Tutors (IVT), but this suggestion has been explored primarily with grammar and vocabulary tasks, but not with computer-based listening tasks. An IVT, in this study, is a parser that highlights potential obstacles to understanding by analyzing the aural input at the linguistic level and suggests the most adequate help option that would allow the students to recover from comprehension breakdowns. The aim of this study and the first step for the conceptualization of an IVT is to identify input text characteristics that affect the complexity of oral texts. Accordingly, we measure the linguistic complexity of 13 authentic talks at the lexical, phonological, syntactical and discourse level. A full description of the criteria for input selection and analyses of the talks are provided. The article concludes with a reflection on limitations to automatic text processing tools for oral texts and provides suggestions for further research.