Journal Of Memory And Language 61 (2009) 412–422

Journal of Memory and Language 61 (2009) 412–422

Contents lists available at ScienceDirect

Journal of Memory and Language
journal homepage: www.elsevier.com/locate/jml

Perceptual learning of co-articulation in speech
Cynthia M. Connine *, Laura M. Darnieder
Department of Psychology, Binghamton University, State University of New York, Binghamton, NY 13901, USA

a r t i c l e

i nf o

a b s t r a c t
Four experiments investigated the novel issue of learning to accommodate the co-articulated nature of speech. Experiment 1 established a co-articulatory mismatch effect for a set of vowel–consonant (VC) syllables (reaction times were faster for co-articulation matching than for mismatching stimuli). A rhyme judgment training task on words (Experiment 2) or VC stimuli(Experiment 3) with mismatching information was followed by a phoneme monitoring task on a set of VC stimuli; training and test stimuli contained physically identical (same condition) or new (different condition) mismatching co-articulatory information (along with a set containing matching co-articulatory information). A third group received no training. A co-articulatory mismatch effect was foundwithout training but not when the same mismatching tokens were used at training and test. Both word (Experiment 2) and syllable (Experiment 3) training stimuli eliminated the mismatch effect; overall reaction times were somewhat slower when the training stimuli were words. Perceptual learning generalized to new tokens only when the acoustic manifestation of the critical co-articulatory information inthe training stimuli was sufficiently large (Experiments 3 and 4). The results are discussed in terms of speech processing and perceptual learning in speech perception. Ó 2009 Elsevier Inc. All rights reserved.

Article history: Received 3 November 2008 Revision received 2 July 2009 Available online 4 August 2009 Keywords: Spoken word recognition Perceptual learning Speech co-articulation Speechperception

Listeners are confronted with a dizzying array of variation in the language they hear. The variation can take the form of alternative pronunciations of words from both familiar and novel dialects, speech categories that differ from a listener’s native language, within talker variability and contextually conditioned segmental variation. The range of variability found in speech wouldrequire a system capable of keen sensitivity to detailed information and at the same time able to make rapid adjustments to those details. The present research addresses the novel issue of learning to accommodate the co-articulated nature of speech. Co-articulation is a pervasive aspect of spoken language and refers to a production process in which a speaker adjusts their production of a segment toaccommodate the surrounding segments. A consequence of the co-articulated nature of speech is that the acoustic information relevant to a given segment is distributed over adjacent (and
* Corresponding author. Fax: +1 607 777 4890. E-mail address: connine@binghamton.edu (C.M. Connine). 0749-596X/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jml.2009.07.003sometimes non-adjacent) segments. Listeners must learn what information is relevant for a given segment and integrate the distributed sources of information to arrive at a percept. The contextual variation in co-articulated segments is due in part to general production constraints generated by individual vocal tract configurations (Fowler & Saltzman, 1993) but other work indicates that there islanguagespecific variation (Maddieson & Emmorey, 1985; Sole, 1992; Strange, Weber, Levy, Shapiro, & Nishi, 2002). Given that languages and speakers vary in the way in which co-articulatory variation is instantiated in production, a listener must learn the general parameters of co-articulatory variability from the language environment. A classic example of the impact of vowel–consonant co-articulatory...