Adaptation to Atypical Contrastive Accent: The L2 Advantage

A growing number of visual world studies have reported that listeners process prosodic cues rapidly enough to predict upcoming speech, directing their eyes to a speaker-intended referent before the onset of the referring noun (Ito & Speer, 2008; Snedeker & Trueswell, 2003; Weber, Braun, & Crocker, 2006). These studies have converged on the idea that prosodic cues are used to guide processing decisions very early stages, and further enter into a larger debate regarding the prediction of words and structures in real time comprehension. The general conclusion is that language users draw on general statistical knowledge of how prosody functions in conjunction with other sources of information, such as contextual and world knowledge, to generate predictions about what information is likely to come next. The specific influence of prosody on predictive inferences – often referred as the anticipatory use of prosody, has been observed in multiple levels of representations, including structural prediction (e.g., Nakamura, Arai, & Mazuka, 2012; Nakamura, Harris, & Jun, 2019; Weber, Grice, & Crocker, 2006) and the interpretation of discourse and information structure (e.g., Ito & Speer, 2008; Dahan, Tanenhaus, & Chambers, 2002; Kurumada, Brown, Bibyk, Pontillo, & Tanenahus, 2014). More recently, a number of studies have manipulated whether or not a particular speaker’s prosody follows conventional use (e.g., boundaries or pitch accents that realize typical relationships between prosody and structure or meaning) and the extent that prosodic marking is informative to the listener. The results suggest that listeners monitor how specific speakers use prosodic information, and adjust the extent to which they rely on this information in anticipatory processing (Nakamura, Harris, & Jun, 2019; Roettger & Franke, 2019; Roettger & Rimland, 2020; Tanenhaus, Kurumada, & Brown, 2015). The observation that comprehension strategies change in response to linguistic input is sometimes taken as evidence for language adaptation, in which language users track statistical properties of the speech situation to dynamically update their