Proceedings of the Annual Meeting of the Cognitive Science Society

Spoken language unfolds over time. Listeners cope with this by activating multiple lexical candidates which compete forrecognition (McClelland & Elman, 1986). Competition dynamics change with degraded speech (Brouwer & Bradlow,2016; McMurray, Farris-Trimble, & Rigler, 2017; McQueen & Huettig, 2012) but it is unclear whether this reflects thedegraded input, or functional adaptation. In two visual world paradigm experiments, listeners heard different levels ofdegraded (noise-vocoded) speech. Experiment 1 manipulated degradation level in blocks or interleaved across trials.Interleaving led to processing delays beyond that of degradation alone. We also found switch-costs when degradationlevel differed between trials. This suggests differences in lexical dynamics are not solely due to degradation level. Inexperiment 2, a visual cue indicated the degradation level before each trial. This reduced the delay and switch costs,suggesting adaptation before the input. These experiments support a role for central processing in dealing with degradedspeech.