Background Infants and toddlers have detailed representations for their known vocabulary items

1. Phonological Neutralization and the Representation of Lexical Tones in Mandarin-speaking Toddlers Jun Gao 1, Rushen Shi 2, Aijun Li 3 gaojungscass@126.com, shi.rushen@uqam.ca, liaj@cass.org.cn • Background • Infants and toddlers have detailed representations for their known vocabulary items • Consonants (e.g., Swingley & Aslin, 2000; Fennel & Werker, 2003; Halle & de Boysson-Bardie, 1996) • Vowels (e.g., Mani & Plunkett, 2007) • No study has examined the nature of representations of lexical tones during early vocabulary development • Lexical tone contrasts are perceived by tone-learning infants at the preverbal stage (e.g., Gao, Shi, & Li, 2010; Mattock & Burnham 2006; Tsao, 2008) • Lexical tones in Mandarin & Tone 2 – Tone 3 neutralization • Four tones (See Fig. 1): • Tone 1 (high level) • Tone 2 (high rising) • Tone 3 (low dipping) • Tone 4 (high falling) • Fig. 1 Time-normalized and pitch-normalized F0 contours of the four lexical tones in Mandarin, produced by a female native speaker. (data from Lee Sung Hoon, Graduate School, Chinese Academy of Social Sciences) • Tone2-Tone3 sandhi rule: Tone3-Tone2 in the environment of a following Tone 3 • i.e., Tone 3 Tone 3 → Tone 2 Tone 3. • e.g., xiao3 (“small”) + gou3 (“dog”) → Tone 2 Tone 3 • Research questions • Experiment 1 • Hypotheses •  If toddlers lack the full knowledge of the Tone 2-Tone 3 sandhi rule, the two phonemic tones may be represented as free variations (i.e., non-distinguishable) in the lexicon; •  If toddlers have acquired the Tone 2-Tone 3 sandhi rule (i.e., knowing that Tone 2 and Tone 3 are phonemic and are only neutralized in specific context), they should have distinctive representations for the two tones. • Methods • Participants:16 Mandarin - learning toddlers; 19-29 months old • Speech stimuli • Four monosyllabic key familiar words, two in Tone 2 and two in Tone 3 • Tone 2 (rising tone): yang2 (“sheep”), chuang2 (“bed”) • Tone 3 (low dipping tone): ma3 (“horse”), wan3 (“bowl”) • The Tone 2 and Tone 3 targets were in a non-neutralizable context in this study • Procedure • Intermodal preferential looking procedure (See Fig. 2) • Two side-by-side pictures of familiar words per trial (6.5 secs); 24 trials • Two test trial types: • Correct pronunciation (CP): • yang2, chuang2, ma3, wan3 • Mispronunciation (MP): • yang2 mispronounced as yang3 • chuang2 mispronounced as chuang3 • ma3 mispronounced as ma2 • wan3 mispronounced as wan2 • Results • Measure:proportion of looking to the target • target looking time divided by the sum of target looking time and distractor looking time • Analysis window:starting 375msec from the onset of the first production of the target, to 2 sec (See Fig. 3) • Fig. 3The structure of a trial • Fig. 4Proportion of looking time to target in CP vs. MP trials • Comparison to chance level (0.5): children recognized the target words in CP and MP • CP significantly above chance, p = 0.000, 2-tailed • MP significantly above chance, p = 0.001, 2-tailed • CP vs. MP comparison: mispronunciation of tones did not impede target recognition • CP & MP not different, p = 0.751, 2-tailed • Toddlers did not distinguish Tone 2 vs. Tone 3 in familiar words, possibly due to Tone 2 – Tone 3 sandhi operations in the input; infants lack full knowledge of this sandhi rule. • General Discussion • Experiment 1 & 2 taken together, toddlers showed •  No distinctive lexical representations for Tone 2 versus Tone 3 (neutralizable in one specific context); •  No distinctive lexical representations for Tone 2 versus Tone 4 (non-neutralizale contrast); •  No distinctive lexical representations for Tone 3 versus Tone 4 (non-neutralizale contrast) • Preverbal tone-learning infants can perceive lexical contrasts in Mandarin and Thai (Gao, Shi, & Li, 2010; Mattock & Burnham, 2006; Tsao, 2008). • Possible explanation for the non-discrimination of tonal contrasts: •  The present task not sensitive enough: familiar-familiar pairings • Subsequent experiment: A more sensitive task: familiar – unfamiliar pairings (White & Morgan, 2008); preliminary results show an MP effect for tonal contrasts that are never subject to neutralization (Tone 2 vs. Tone 4; Tone 3 vs. Tone 4). References • Fennell, C. T., & Werker, J. F. (2003). Early word learners' ability to access phonetic detail in well-known words. Language and Speech, 46(2-3), 245. • Gao, J., Shi, R., & Li, A. (2010). Categorization of lexical tones in Mandarin-learning infants. Paper presented at the the Fifth International Conference on Speech Prosody, 2010, Chicago. • Hallé, P. A., & de Boysson-Bardies, B. (1996). The format of representation of recognized words in infants' early receptive lexicon. Infant behavior and development, 19(4), 463-481. • Mani, N., & Plunkett, K. (2007). Phonological specificity of vowels and consonants in early lexical representations. Journal of Memory and Language, 57(2), 252-272. • Mattock, K., & Burnham, D. (2006). Chinese and English infants' tone perception: Evidence for perceptual reorganization. Infancy, 10(3), 241-265. • Swingley, D., & Aslin, R. N. (2000). Spoken word recognition and lexical representation in very young children. Cognition, 76(2), 147-166. • Tsao, F.-M. (2008). The Effect of Acoustical Similarity on Lexical-Tone Perception of One-Year-Old Mandarin-Learning Infants. Chinese Journal of Psychology, 50(2), 111-124. • White, K. S., & Morgan, J. L. (2008). Sub-segmental detail in early lexical representations. Journal of Memory and Language, 59(1), 114-132. • Acknowledgement • This work was supported by the CASS Key project fund and the Chinese Social Science Fund to the third author and the first author, and grants from NSERC, SSHRC and CFI to the second author. • Affiliations • 1.3. Institute of Linguistics, Chinese Academy of Social Sciences, China • 2. Université du Québec à Montréal, Canada Experiment 2 Do toddlers distinctively represent tonal contrasts that are not subject to neutralization? • Methods • Participants:18 Mandarin - learning toddlers; 19-29 months old • Speech stimuli • The monosyllabic key familiar words of Experiment 1: • Tone2 (rising tone): yang2 (“sheep”), chuang2 (“bed”) • Tone3(low dipping tone): ma3 (“horse”), wan3 (“bowl”) • Procedure • Intermodal preferential looking procedure • Two side-by-side pictures of familiar words per trial (6.5 secs); 24 trials • Two test trial types: • Correct pronunciation (CP): • yang2, chuang2, ma3, wan3 • Mispronunciation (MP): • yang2 mispronounced as yang4 • chuang2 mispronounced as chuang4 • ma3 mispronounced as ma4 • wan3 mispronounced as wan4 • (Note: Tone 2 & Tone 3 are never neutralized with Tone 4 in any context) • Results • Fig. 5Proportion of looking time to target in CP vs. MP trials • Comparison to chance level (0.5): children recognized the target words in CP and MP • CP significantly above chance, p = 0.005, 2-tailed • MP significantly above chance, p = 0.06, 2-tailed • CP vs. MP comparison: mispronunciation of tones did not impede target recognition • CP & MP not different, p = 0.687, 2-tailed • Toddlers did not distinguish Tone 2 vs. Tone 4, nor Tone 3 vs. Tone 4 in familiar words. • Fig. 2 One example of the pictures presented to the subjects • Do lexical-tone-learning toddlers • possess distinctive tonal representations • for familiar words? • Does tonal neutralization affect • toddler’s tonal representations of • familiar words?