Phonetic priming, neighborhood activation, and PARSYN

Abstract

Perceptual identification of spoken words in noise is less accurate when the target words are preceded by spoken phonetically related primes (Goldinger, Luce, & Pisoni, 1989).The present investigation replicated and extended this finding. Subjects shadowed target words presented in the clear that were preceded by phonetically related or unrelated primes. In addition, primes were either higher or lower in frequency than the target words. Shadowing latencies were significantly longer for target words preceded by phonetically related primes, but only when the prime-target interstimulus interval was short (50 vs. 500msec). These results demonstrate that phonetic priming does not depend on target degradation and that it affects processing time. We further demonstrated that PARSYN-a connectionist instantiation ofthe neighborhood activation model-accurately simulates the observed pattern of priming.Virtually all current models of spoken word recognition share the assumption that the perception of spoken words involves two fundamental processes: activation and competition (see P. A. Luce & Pisoni, 1998;Marslen-Wilson, 1989;McClelland & Elman, 1986;Norris, 1994). In such activation-competition models, the hallmark of the discrimination process is competition among multiple representations of words activated in memory. As a result, the role of competition has been a primary focus of research and theory on spoken word recognition in the last few years (e.g., Cluff & Luce, 1990;Goldinger, Luce, & Pisoni, 1989;Marslen-Wilson, 1989;McQueen, Norris, & Cutler, 1994;Norris, McQueen, & Cutler, 1995;Vitevitch & Luce, 1998, 1999.One example of an activation-competition model is the neighborhood activation model (NAM; P. A. Luce & Pisoni, 1998). According to NAM, stimulus input activates a set (or neighborhood) of acoustic-phonetic patterns in This research was supported in part by Research Grants ROI DC 026580 I-A2 and R29 DC 02629-03 from the National Institute on Deafness and Other Communication Disorders. National Institutes of Health . We thank Dennis Norris and an anonymous reviewer for their advice and comments. We also thank Jim Sawusch for many helpful discussions and Michael S. Cluff for his assistance in running subjects. Correspondence concerning this article should be addressed to P A. Luce, Department of Psychology, State University of New York, Buffalo, NY 14260 (e-mail: paul@deuro.fss.buffalo.edu).memory. Patterns are activated to the degree to which they match the stimulus input (see Marslen-Wilson, 1989, Morton, 1969, for similar proposals). These acoustic-phonetic patterns then activate a system of word decision units that are tuned to the patterns. Throughout the recognition process, the word decision units monitor three sources of information: (1) the activation levels of the acoustic-phonetic patterns to which the units are tuned, (2) higher level lexical information (specifically, lexical frequency), and (3) the overall level of activity in the entire system of decision units. It is assumed that each of the de...

Time Course of Frequency Effects in Spoken-Word Recognition: Evidence from Eye Movements

“…According to Goldinger et al, a model in which frequency operates directly on activation levels (either on resting activations or connection weights) would predict more inhibition from high-frequency primes than from lowfrequency primes because high-frequency primes should produce stronger competing activation. However, Luce, Goldinger, Auer, and Vitevitch (2000) reported simulations with a connectionist model (''PARSYN'') in which frequency is instantiated in connection strengths (i.e., the weights of lateral connections between phonemes are proportional to forward and backward transitional probabilities), which correctly predicted the trend for stronger priming by low-frequency items. As Luce et al note, PARSYN is similar to a class of models, including TRACE, which presumably would make similar predictions.…”

Lexical viability constraints on speech segmentation by infants

“…In addition, spurious word forms that span existing word boundaries (such as speech in grasp each) may correspond to lexical items and lead to their activation in the candidate set (Gow & Gordon, 1995;Shillcock, 1990;Tabossi, Burani, & Scott, 1995;Vroomen & de Gelder, 1997). Consequently, other models have assumed that lexical candidates can be activated at any point in the speech stream (Luce, Goldinger, Auer, & Vitevitch, 2000;McClelland & Elman, 1986;Norris, 1994). Indeed, many words are apparently not properly identified until after their acoustic offset (Allopena, Magnuson, & Tanenhaus, 1998;Goldinger, Luce, & Pisoni, 1989;Luce et al, 2000).…”

“…Consequently, other models have assumed that lexical candidates can be activated at any point in the speech stream (Luce, Goldinger, Auer, & Vitevitch, 2000;McClelland & Elman, 1986;Norris, 1994). Indeed, many words are apparently not properly identified until after their acoustic offset (Allopena, Magnuson, & Tanenhaus, 1998;Goldinger, Luce, & Pisoni, 1989;Luce et al, 2000). In any case, the competition among the activated candidates winnows the set to the word that is ultimately extracted and recognized.…”

Listeners recover /t/s that speakers reduce: Evidence from /t/-lenition in Dutch

“…This divide is mirrored in models of spoken-word recognition. In most models (e.g., Lahiri & Reetz, 2002;Luce, Goldinger, Auer, & Vitevitch, 2000;McClelland & Elman, 1986 /eix=l=k/ 'actually' and natuurlijk /natyrl=k/ 'naturally'. Ernestus (2000) showed that these forms may be pronounced as [eik] and [tyk] in casual speech.…”