Idea Transcript
Segmental and supra-segmental accuracy of nonword repetition in bilingual children ASHA Convention Atlanta, GA 2012
Sue Ann Lee, Sherry Sancibrian, & Dalan Gore
Disclosure • No financial or non-financial interest to disclose.
Background • Nonword repetition (NWR) has been widely used for evaluating phonological short-term memory. • In clinical practice, NWR tasks have been recommended as a component of less-biased child language assessment protocols.
Previous studies • Gathercole & Baddeley (1990) ▫ Nonword repetition in 6 children with SLI and 6 age-matched and younger children with typical development. ▫ The children with SLI scored significantly lower than both groups of children with typical development.
Previous studies • Thal, Miller, Carlson, & Vega (2005) ▫ 44 typically developing and 20 children that had a history of language delay were studied. ▫ Children that had a history of language delay scored significantly lower than children with typical language development.
Previous studies • Munson, Edwards, & Beckman (2005) ▫ Compared NWR of typically developing children to children with phonological disorders ▫ Children with phonological disorders were less accurate overall, but both groups repeated high frequency sequences of phonemes more successfully than low frequency.
Previous studies • Calderon & Guiterrez-Clellen (2003) ▫ Compared NWR performance of 21 Spanishspeaking children with SLI with that of 21 typically developing Spanish-speaking children ▫ Spanish-speaking children with SLI performed significantly worse than the typically developing Spanish-speaking children
Factors related to NWR performance • Possible potential prosodic influences on poor NWR • Hanson et al. (1999) ▫ Swedish children with SLI were six times more likely to omit unstressed syllables in weak-strong syllable combination than strong-weak syllable combination
Factors related to NWR performance • Linguistic experience ▫ L1 appears to influence NWR performance in L2 (Gutierrez-Clellen & Simon-Cerejido, 2010; Summers et al., 2009.
• Windsor, Kohnert, Lobitz, & Pham (2010) ▫ Investigated NWR between bilingual SpanishEnglish and monolingual English-speaking children with and without language impairment.
Previous studies • Windsor, Kohnert, Lobitz, & Pham (2010) ▫ Spanish-English bilinguals showed lower performance on English NWR tasks than English monolinguals ▫ NWR performance relies on native language experience and may be sensitive to language impairments.
Previous studies • Carter, Dillon, & Pisoni (2002) ▫ Examined prosodic characteristics of NWR task in children with cochlear implants ▫ Accuracy of number of syllables and primary stress ▫ 64% of NWR had the correct number of syllables ▫ 61% of NWR had the correct placement of primary stress.
Previous studies • Dillon, Clearly, Pisoni, & Carter (2004) ▫ 39% of consonants were produced correctly. ▫ Coronals were produced with greater accuracy than labials or dorsals. ▫ No difference across consonant manner or voicing.
Needs • Further insight into how linguistic background may influence NWR performance is needed. • Further investigation into prosody and production accuracy of consonant and vowel categories across various linguistic background is needed.
Purpose • To analyze the potential influence of various linguistic backgrounds on children’s English NWR accuracy. • To analyze segmental and supra-segmental accuracy of English NWR.
Relevance • Information obtained may prove useful for refining NWR measures to make them more appropriate for children from various linguistic backgrounds and to identify specific factors that influence their performance.
Methods • Participants ▫ ▫ ▫ ▫
15 Korean-English bilingual children (KE) 15 Chinese-English bilingual children (CE) 12 Spanish-English bilingual children (SE) 15 Monolingual English-speaking children (ME)
• No speech, language, hearing or other health problems
Bilingual children • All bilingual children acquired their home language first and learned English at Englishspeaking daycare, preschool, kindergarten. Onset of English language exposure
# of KEB # of CEB
# of SEB
Before 3
1
2
3
After 3-Before 5
7
8
3
After 5
7
5
6
Vocabulary and Phonological awareness 140
120
Standard Scores
100
80 PPVT
60
ELLA
40
20
0
ME
KEB
CEB
SEB
Consonants
Vowels
Stress-timed vs. Syllable-timed • Stress-timed: English ▫ Trochaic vs. iambic
• Syllable-timed: Spanish, Chinese, Korean
Dominant syllable types • English: mono-syllable • Korean: di-syllable • Chinese: mono-syllable • Spanish: multi-syllable
Data Collection • 64 Nonword repetition stimuli ▫ 2-3 syllable nonwords (Edwards et al., 2004) ▫ 4-5 syllable nonwords (Gathercole et al., 1994)
Edwards et al. (2004) • /vugim/ • /bɪdæg/ • /donug/ • /mɪnæmp/ • /kɛdəwəmb/ • /tikəpo/
Gathercole et al. (1994)
Data collection • The adopted nonwords were produced by two native English speakers • The recorded stimuli were played via computer • Each child was asked to repeat what he/she heard
Transcription • All responses were transcribed by native English speakers using IPA. • A second transcriber transcribed 20% of NWR ▫ Mean reliability for consonants = 89% ▫ Mean reliability for vowels = 87%
Data analysis • All data were entered into the LIPP (Oller & DeAngelo, 2001)
Data analysis: Segmental analysis • Percentage of consonants correct (PCC) ▫ (length of syllable * groups)
• Consonant accuracy: total & each syllable length • Percentage of vowels correct (PVC) ▫ (length of syllable * groups)
• Vowel accuracy: total & each syllable length
Data analysis: Supra-segmental analysis • Percentage of accuracy for primary stress • (length of syllable * groups)
NWR: Percentage of Consonant Correct 100
90
PCC
80
70
ME KEB
60
50
CEB SEB 2 syllables
3 syllables
4 syllables
5 syllables
Total
Consonant Manner
NWR: Percentage of Vowel Correct 100
90
PVC
80
70
ME KEB
60
50
CEB SEB 2 syllables
3 syllables
4 syllables
5 syllables
Front vowels
More lax vowel errors than tense or corner vowels KEB & CEB produced more errors than ME & SEB
Central vowels
No significant differences among groups
Back vowels
More lax vowel errors than tense and corner vowels
Supra-segmental analysis
2 Syllable NWR
3 Syllable NWR
KEB: more voicing errors
4 Syllable NWR
SEB: more place and manner errors than CEB and ME
5 Syllable NWR
SEB & KEB more errors than ME & CEB
2 Syllable NWR
3 Syllable NWR
4 Syllable NWR
5 Syllable NWR
2 Syllable NWR
3 Syllable NWR
CEB & KEB more vowel errors than SEB
4 Syllable NWR
No group differences
5 Syllable NWR
CEB & KEB more errors than ME & SEB
2 Syllable NWR
3 Syllable NWR
4 Syllable NWR
5 Syllable NWR
2 Syllable NWR
3 Syllable NWR
4 Syllable NWR
5 Syllable NWR
Discussion 1. Analyze performance according to consonants. •
•
Consonant accuracy: SE and KE groups displayed lower accuracy than monolingual and CE groups for 4 and 5 syllable NWR. English and Chinese contain larger consonantal inventories than Korean and Spanish. This may affect production accuracy of fricatives.
Discussion 2. Analyze performance according to vowels. • •
Vowel accuracy: CE and KE groups displayed lower accuracy than monolingual and SE groups. Consonant accuracy may yield less-linguistically biased results than vowels, on which greater differences emerged as word length increased.
Implications • Results provide further support for the view that, even in bilinguals with relatively strong English skills, NWR tasks do not eliminate the role of linguistic experience. • NWR tasks are often used to tap PSTM, yet there are clearly language-specific effects not only in terms of production accuracy, but also seemingly, on the underlying processes that influence performance.
Consonants in NWR • Consonants in NWR have been handled differently: • Dollaghan & Campbell ▫ Excluded late 8 sounds (sibilants & liquids)
• Gathercole & Baddeley ▫ Included late 8 sounds
Vowels in NWR • Vowels in NWR have been handled differently: • Dollaghan & Campbell ▫ tense vowels bc of increased perceptibility (and thus violate typical English metrical pattern which includes weak syllables)
• Gathercole & Baddeley ▫ both tense and lax vowels, and included weak syllables
Prosody in NWR • Prosody in NWR have been handled differently: • Dollaghan & Campbell ▫ Used only tense vowels, contrast with the typical English stress patterns.
• Gathercole & Baddeley ▫ Contained weak syllables with a reduced vowel, natural prosodic patterns of English