Lecture 16 Language 2 Flashcards
(15 cards)
Definition of phoneme
Smallest sound unit that distinguishes words in a language (e.g., /p/ vs. /b/ in ‘pin’ vs. ‘bin’)
Vowel classification
2 dimensions:
a) High vs. low (tongue position)
b) Front vs back (lip posture & where in the mouth)
Consonant features
3 features:
a) Voicing (vibration)
b) Place of obstruction (e.g., bilabial)
c) Manner (how airflow is blocked)
Categorical perception
Hearing sounds as discrete categories (e.g., /g/ vs. /k/) despite continuous acoustic variation; present by 3 months
Statistical learning in phonemes
Infants identify native phonemes by detecting frequency/distribution patterns in speech input
Critical period for phonemes
By ~12 months, infants lose sensitivity to non-native phonemes (e.g., Japanese infants with /l/ vs. /r/)
Transitional probabilities (TPs)
Likelihood of syllables co-occurring; infants use TPs to segment words (e.g., ‘doggie’ vs. random chunks)
Saffran et al. (1996) finding
8-month-olds segmented nonsense words after 2 minutes of exposure using only TPs
Word segmentation challenges
No audible pauses between words; solved via TPs, prosody, and phonotactic rules
Pelucchi et al. (2009) result
Infants used TPs to segment real Italian words, showing applicability to natural languages
Computational modeling evidence
TP-based algorithms successfully segment child-directed speech corpora (Goldwater et al., 2009)
Domain-general statistical learning
TP tracking applies beyond language (e.g., visual sequences, actions)
Phoneme-vocabulary link
Better phoneme discrimination at 6 months predicts larger vocabulary at 13-16 months (Tsao et al., 2004)
Child-directed speech (CDS) role
Simplified pitch/grammar helps segmentation, but not strictly necessary (!Kung San study)
Universal to language-specific
Newborns discriminate all phonemes; by 12 months, only native-language contrasts
