Bayesian Analysis of Non - native Cluster Production

Under a variety of natural and experimental conditions, speakers and listeners distinguish among different types of non-native phonetic/phonological structure. For example, when asked to judge the acceptability of nonce words, speakers exhibit gradient preferences for certain non-native sound sequences over others (e.g., Greenberg and Jenkins 1964, Scholes 1966, Coleman and Pierrehumbert 1997, Vitevitch et al. 1997, Frisch et al. 2000, Treiman et al. 2000, Bailey and Hahn 2001, Frisch and Zawaydeh 2001, Hammond 2004, Shademan 2007, Albright 2009). Similarly, while it has long been known that structures absent from the native language of the listener are particularly susceptible to misperception (or ‘perceptual assimilation’; e.g., Massaro and Cohen 1983, Hallé 1998, Pitt 1998, Dupoux et al. 1999), recent research has established that not all non-native structures are misperceived or perceptually assimilated at the same rate (e.g., Moreton 2002, Berent et al. 2007, Haunz 2007, Kabak and Idsardi 2007, Yarmolinskaya 2010).