The literature on speech perception in noise shows that noise impairs both low-level processes (phonetic encoding) and higher-level processes (lexical-semantic integration). However, the degree to which both processes are affected is unlikely to be monotonic across noise levels and tasks.
Goal: To identify the conditions under which reliance on phonetic or lexical processing is particularly reduced and improve the ecological validity of existing models of speech perception in noise.
Relevance: A detailed understanding of how noise interacts with the low- vs. high-level components of the language system is central to models of speech recognition because it highlights listeners' reliance on cue trade-off mechanisms in realistic listening conditions. It can also contribute to targeted interventions on environmental conditions to improve intelligibility.
Main host institution: University of York
Second host institution: Radboud University Nijmegen
Hospital partner: Royal National Throat, Nose and Ear Hospital