Tid: Tisdag 21 juni 2016, kl. 10:00-12:00
Plats: C307, Södra huset, Frescati

Much is known about infants’ emerging ability to detect mispronunciations of familiar words, or learn novel words, involving minimal-pair consonant contrasts. However, recognizing words despite systematic phonetic variation (e.g., across regional accents), which we refer to as phonological constancy, is also crucial for lexical acquisition. The first studies to examine phonological constancy (Best et al., 2009; Mulak et al., 2013) found that English-learning 14-15 month olds recognize words familiar to toddlers when spoken in their native accent but not in an unfamiliar English accent, whereas 18-19 month olds recognize words in both accents. We concluded from this that the older but not the younger children had acquired phonological constancy. Subsequently, we found a positive relationship between the emergence of phonological constancy (cross-accent word recognition) and the size of the children’s expressive vocabulary.

In a recent series of studies, we selected target words that limited the primary cross-accent phonetic discrepancies to either vowels and/or consonants, and to differences that constituted either “less-nativelike” tokens of the same native phoneme category (Category Goodness: CG), or crossed a boundary into a contrasting native phoneme (Category Shifting: CS). The results indicate that when cross-accent discrepancies in word pronunciation are limited to CG differences in vowels or consonants, even 15-month-olds can recognize familiar words in both accents. However, with CG vowel differences a reliable age difference in overall listening preference was found (native accent preferred at 15 months; novel accent at 19 months), whereas with CG consonant differences neither age showed any accent preference, implying lower sensitivity to consonant than vowel variations. By comparison, when target words instead showed CS accent differences in vowels alone or in both vowels and consonants, even 19 month olds failed to recognize them in the unfamiliar accent, suggesting that categorical shifts in perceived vowels disrupts word recognition (as in adults). Importantly, when only cross-accent CS consonant discrepancies were present, both age groups again recognized the familiar toddler words in both accents (see Figure 1), without no accent preferences. This suggests the same insensitivity to CS as to CG discrepancies in consonants.

But how do young children adapt to accent variation? Like adults, infants show perceptual benefits for talkers they have heard before, so we asked whether this benefit extends to recognition of new words by familiarized talkers; and whether such benefit might persist over 4-5 months. Two groups of 19 month olds were tested on word recognition in AusE and in an unfamiliar accent: Group A had no prior exposure to the unfamiliar accent or to any of the talkers; Group B had participated 4-5 months earlier in a study that had used different word sets by the same talkers/accents. Group B’s brief prior exposure (5-10 minutes) did indeed enhance recognition of novel words by the same talkers. Although both groups recognized the toddler words in both accents, performance was significantly better in the experienced group.

Implications of these findings for theories of consonant versus vowel contributions to word recognition, as well as the models on the role of talker- and accent-specific episodic memory in early lexical retrieval, will be discussed.

Hjärtligt välkomna!

Iris-Corinna Schwarz och Hatice Zora