180 likes | 346 Views
Are phonological representations feature based?. Do children think “tug” and “bus” sound similar?. Dr Julia Carroll Joanne Myers Department of Psychology University of Warwick, UK. Phonological Awareness. Pre-school children struggle with awareness of phonemes (Morias et al, 1974)
E N D
Are phonological representations feature based? Do children think “tug” and “bus” sound similar? Dr Julia Carroll Joanne Myers Department of Psychology University of Warwick, UK
Phonological Awareness • Pre-school children struggle with awareness of phonemes (Morias et al, 1974) • They have better performance on rhyme and syllable tasks • They have phonological sensitivity but not awareness (Carroll et al, 2003) What form could this sensitivity take?
‘Fuzzy’ or ‘underspecified’ representations • Elbro (1998): • Children and adults with dyslexia have phonological representations which are not fully specified • Walley (1993) • Typically developing children have ‘global’ representations • What form could these representations take? • Organised by phonetic features rather than phonemes
Feature based representations? • Many feature based errors in early speech • Storkel (2002) • Children are likely to classify words together if they share manner of articulation • Only true in sparse phonological neighbourhoods • Jusczyk et al (1999) • Nine month olds show sensitivity to: • Word onsets • Shared manner in word onsets • Varying results so far – small samples, single tasks
Current Research • 125 children tested: • 59 from Reception (m= 4;11 years) • 66 from Nursery (m = 4;0 years) • Five measures used, four focused on: • Explicit classification: • Forced Choice (based on Storkel, 2002) • Sound Families • Implicit memory • Word Production • Memory confusions
Hypotheses • Young children code words according to phonetic features – particularly manner of articulation • Therefore, they are more likely to say words are similar if they share manner of articulation. • They are likely to be confused by words sharing phonetic features in priming or memory tasks
Forced Choice classification tug Same Phoneme Manner Place Different Onset + nucleus tug tough bus son mum mud young mum Rime tug hug Based on Storkel (2002)
Forced choice n = 93 • Main effect of relationship to target (F(4,368) = 197.7, p<.01, η2 = .68 • Interaction between word type and relationship (F(4,368) = 11.5; p < 0.01. η2 = .11) • Main effect of neighbourhood density (F(1,92) = 10.6, p < .01, η2 = .10)
Forced Choice • Significant effect of relationship • Children are more likely to classify words as similar if they share manner of articulation • Interaction with manner matching being particularly common in the rime • Significant effect of neighbourhood; with more same ratings given for sparse words. • Words are less fully represented in sparse neighbourhoods?
Sound families – shared body “beach” “beam” “beak” “bead” “bean” Vowel = same Vowel = different Same Manner Place Different Manner Place Different Onset + nucleus beep team meat read cot food shirt
Sound Families – shared rime “bat” “chat” “sat” “cat” “that” Vowel = same Vowel = different Same Manner Place Different Manner Place Different Nucleus + rime pat map gas ham mop wheel ring
81.1% shared a vowel 62.2 % were rhyming words 16.8% shared vowel and manner in the coda Only 2% shared a vowel without shared manner in coda Can you tell me words that sounds like… ? You can even make up words! Only CVCs were analysed Word Production - Coda (n = 79) “Bed” “Nut” “Shop” “That”
81.1% shared a vowel 4.0 % shared onset and vowel 11.6% shared onset with a different vowel 36.9% shared some phonetic features and vowel 39.8% shared vowel but no phonetic features in onset Can you tell me words that sounds like… ? You can even make up words! Only CVCs were analysed Word Production - Onset (n = 79) “Bed” “Nut” “Shop” “That”
Memory confusions: Treiman & Breaux, 1982 “[bis]” • Children and adults taught three name – animal associations • Two words share a common phoneme • Two words share ‘global similarity’ • Which pair are most likely to be confused? • Adults – common phoneme pair • Children – globally similar pair “[bun]” “[diz]”
Memory Confusions “[mern]” “[moab]” “[vit]” Mern and moab share one phoneme Mern and vit share place of articulation (broadly) 1 manner and phoneme 1 place and phoneme 1 manner and place Time A Time B = 24 hours later
Memory Confusions At both time points, 63% of all errors were confusions At both times, distribution of the confusions differs from the random distribution: Time 1: χ2 (3) = 18.01, p < .01 Time 2: χ2 (3) = 81.22, p < .01
Summary • Explicit classification • Words sharing manner are more likely to be classified as similar • This is particularly strong word finally • Implicit tasks • Children seem to regard rhyming words as the primary ‘sounds similar’ response • Shared manner is important in the coda in word production • Children make higher levels of manner based and lower levels of anomalous confusions than expected.
Conclusions • Young children do represent phonetic features within words • More evidence for manner than place • More common word finally • This is true in implicit and explicit tasks • The effect is not just a learnt strategy but present in representations • This may be why pre-literate children have sound sensitivity, but not phoneme awareness