University of western Sydney,

Native language tuning effects on perceptyion of consonants, vowels and spoken words ». 22/04, 29/04, 13/05, 20/05.

Labex EFL: Catherine Best’s four seminar presentations

Who? Catherine Best is a Labex EFL International Chair in 2014 and will be working at the Laboratoire de Phonétique et Phonologie from April 14 to May 21. She is a Professor and Chair in Psycholinguistic Research at MARCS Institute and the School of Humanities and Communication Arts at the University of Western Sydney  (Australia), is the Editor-in-Chief of the journal Phonetica, and has also been selected as the ISCA Distinguished Lecturer (International Speech Communicaiton Association) for 2014-2016. Professor Best is most well-known for her Perceptual Assimilation Model (PAM) of nonnative speech perception and empirical findings within that framework, which has more recently been extended to cross-language perception by second language learners and bilingual listeners. Her current work has also extended the principles of cross-language speech perception to address how adult and infant listeners deal with regional accent differences within their own native language when recognizing spoken words.

Where? Salle Las Vergnas (Centre Censier, 3ème étage, 17 rue Santeuil, Censier-Daubenton)

When? Tuesday 22 and 29 April, 13 and 20 May, 4 to 6 p.m., except 29 April (5 to 7 p.m.)

What? Talk #1: April 22, 16-18h:

title: « Devil or angel in the details? Phonetic variation and the complementary principles of phonological distinctiveness and phonological constancy”


The phonetic patterns of ambient speech provide the raw materials for infants to discover the principles of their native language. By 10-12 months they show attunement to phonetic variations that are relevant in their language, and declining sensitivity to distinctions that are irrelevant to it, laying the cornerstone for mature listeners’ rapid and automatic recognition of native words. But what makes a phonetic distinction ‘relevant’ versus ‘irrelevant?’ The answer lies in how listeners relate the phonetic details of a word to its phonological structure, while taking into account the extensive phonetic variations in a given word across talkers, speech styles, and regional accents. Those phonetic variations are not “noise,” instead providing crucial information about two complementary principles that together define the phonological structure of words. One principle is phonological distinctiveness, which refers to language-specific minimal contrasts that meaningfully distinguish otherwise identical spoken forms. The complementary principle is phonological constancy, which permits listeners to recognize a word across talker and accent differences. A spoken word’s structure is co-defined by the phonetic variations that alter its phonological form and those that leave it intact. Discovering the balance between those two sides of native speech variability requires both episodic and abstract learning, which moves the child beyond attunement and into the realm of word recognition, and provides the foundation for adults’ rapid, automatic recognition of native language words.


Talk #2: April 29 (17-19h):

title: « Cross-language speech perception: Naive, second-language and bilingual listeners”

Talk #2: May 13 (16-18h):

title: « Spoken word recognition across regional accent variation: I. Native and second language adults”



Unfamiliar regional accents disrupt spoken word recognition by L2 adults and young L1 learners, can also trip up L1 adults, and often confuse ASR and smart systems. Little research, however, has addressed the aspects of non-native accents that hinder word recognition, or the processes involved. We used a Visual World task to assess how English regional accent differences influence the time course of spoken word recognition by L1 and L2 adults. Based on the principles of the Perceptual Assimilation Model (PAM: e.g., Best, 1995, and PAM-L2: Best & Tyler, 2007), we selected cross-accent differences in consonant and vowel pronunciation to fall into two broad categories: Category Shifting (CS) and Category Goodness (CG). CS differences refer to pronunciations in a relatively unfamiliar regional accent that L1 listeners are likely to perceptually assimilate to their native accent as a different phoneme than the speaker intended, e.g., Australian (AusE) listeners tend to hear TH in Cockney-accented THIEVES as [f]. CG accent differences instead involve non-native accented phoneme pronunciations that are likely to be assimilated instead as merely deviant pronunciations of the same phoneme in the listener’s native accent, e.g., AusE listeners hear the affricated T in Cockney-accented TINY as a /t/ with a deviant or marked pronunciation. Two listener groups of listeners, for whom AusE was the L1 or L2 (Chinese L1-Mandarin speakers), heard words spoken in AusE and two unfamiliar accents, Cockney-(CknE) and Jamaican-Mesolect-accented English (JaME), which display both CS and CG differences from AusE, primarily in their consonant (CknE) or vowel pronunciations (JaME). Listeners heard each word and identified it by clicking among printed choices of the target word, word onset competitor, word offset competitor, and phonologically unrelated distracter, or ³not there². Proportions of fixations to onset and offset competitors during the decision period indicate that cross-accent perceptual assimilation to the AusE accent played a key role in recognition of JaME and CknE pronunciations for both L1 and L2 listeners, especially at word onsets but also at offsets. Vowel and consonant variations affected lexical competition similarly in both groups, suggesting the L2 listeners had formed AusE-accented lexical representations, with one exception: for CknE words the L2 listeners failed to show the same assimilation (CS>CG) difference for word offset competitors as the L1 listeners. Thus, although the Mandarin listeners had formed AusE-like phonological categories for their L2-English, they paid less attention to consonant coda information than did L1-AusE listeners, suggesting a persistent influence of their L1-Mandarin, in which only nasal consonants can occur in coda positions. Implications of these findings for current perspectives on spoken word recognition and cross-language speech perception will be discussed

May 20 (16-18h): Spoken word recognition across regional accent variation: II. Development in young children

title: « Spoken word recognition across regional accent variation: II. Development in young children”

abstract: Numerous prior findings indicate that 19-month-olds detect minimal-pair phonological distinctions in newly-learned and already-known words more quickly and reliably, and under a greater range of task demands, than 14-15 month-olds. Those studies all employed single-phoneme manipulations (equivalent to « mispronunciations ») of words the child knew or had just been taught, which essentially tests the children’s working knowledge of phonological distinctiveness, i.e., recognition that minimal-pair phonemic changes transform a given word into another word or a nonword (Best, Tyler, Gooding, Orlando & Quann, 2009; Kitamura, Panneton & Best, 2013; Mulak & Best, 2013). However, such manipulations confound a phonetic change with a phonological change, and cannot alone clarify the nature of the developmental change between 14-15 and 19 months. The complementary ability to recognize phonological constancy in words despite differences in pronunciation, e.g., phonetic variations across regional accents, can help resolve the issue. Our research on toddlers’ recognition of familiar words pronounced in a regional accent of the native language that the child has not previously experienced, reveals a similar developmental trajectory for phonological constancy as for phonological distinctiveness. We reported the first finding that 19- but not 15-month-olds can recognize words in an unfamiliar accent (Best et al., 2009). Our follow-up studies have found that the younger age has difficulty recognizing even native-accented words when stimulus variability is increased (more speakers, words, and tokens), that vocabulary size rather than age per se is the correlate of stable versus unstable phonological constancy, that these developmental patterns hold up across eyetracking measures of word identification as well as listening preferences for familiar words, and more recently that perceptual assimilation (Perceptual Assimilation Model: Best, 1995; Best & Tyler, 2007) of the non-native-accented consonants and vowels to the listener’s native accent underlies the emergence of phonological constancy . The implications of the complementary relationship between phonological distinctiveness and phonological constancy for lexical and phonological abilities will be discussed.