Speech Sound Disorders: Red Flags for Literacy Problems

Speech Sound Disorders: Red Flags for Literacy Problems Jonathan Preston Haskins Laboratories preston@haskins.yale.edu Mary Louise Edwards Syracuse University mledward@syr.edu 1

Outline Defining speech sound disorders (SSD), phonological processing, literacy Known risk factors: Which children with SSD are likely to have literacy problems? Types of speech errors Relating speech errors to phonological processing in preschoolers with SSD Summary 2

Outline Defining speech sound disorders (SSD), phonological processing, literacy 3

Speech Sound Disorders (SSDs) Difficulty producing and/or using the sounds of the adult language, in comparison to peers. Includes both articulation and phonology, as well as possible cases of CAS. Does not include (in this study) SSDs with a clear etiology. 4

Prevalence of SSDs Approximately 32% of all communication disorders are SSDs (Slater, 1992). 2% 6 % of school age children have SSDs. Many children with SSDs also have language problems; estimates from 35 50% (Tyler, 2002) to 75% 85% (Shriberg & Kwiatkowski, 1988). 5

SSDs and Literacy Phonologically based reading and spelling skills are often lower for individuals with a histories of SSDs (Lewis & Freebairn, 1992). 50 75% of children with SSDs have been found to have long term academic problems (Lewis et al., 2006). 6

Some Early Predictors of Literacy Acquisition Phonological awareness Phonological (short term) memory Phonological recoding in lexical access 7

Phonological Processing Phonological Awareness Phonological Memory Rapid Naming Literacy 8

Phonological Awareness The awareness of speech sounds and the ability to reflect on and manipulate them (Tunmer, Pratt & Herriman, 1984). Involves abilities such as rhyming, segmenting, blending, and manipulating sounds (e.g., moving or reversing sounds). 9

Phonological Awareness and Literacy Phonological awareness is a critical skill for literacy development. Performance on phonological awareness tasks is a good predictor of future reading ability (Webster & Plante, 1992). 2/3 of children with SSDs between the ages of 3;6 and 16 have low PA (Broomfield & Dodd, 2005). 10

Phonological (Short Term) Memory Temporary storage/retention of speech sound information (Baddeley, 2003). Assessed by tasks such as repetition of syllables or nonwords. 11

Phonological Recoding in Lexical Access Often referred to as Rapid Naming (RN) Assessed by tasks that involve rapidly naming pictures of familiar objects, letters, etc. Relationship with reading skills, especially reading fluency 12

Outline Defining speech sound disorders (SSD), phonological processing, literacy Known risk factors: Which children with SSD are likely to have literacy problems? 13

Syllable/Rhyme- Level Awareness Phoneme- Level Awareness Reading/ Spelling 14 Standardized Effect Size -3.0-2.0-1.0 0.0 1.0

Which children with SSD are most likely to have literacy problems? Co occurring Language Impairment Robust evidence suggesting SSD + LI = increased risk for literacy problems (Catts, 1993; Lewis et al., 2004; Nathan et al., 2004) 15

Syllable/ Rhyme Level Awareness Phoneme Level Awareness Reading/ Spelling Syllable/ Rhyme- Level Awareness Phoneme- Level Awareness Reading/ Spelling 16 Standardized Effect Size -5.0-4.0-3.0-2.0-1.0 0.0 1.0-5.0-4.0-3.0-2.0-1.0 0.0 1.0

Which Language Skills? Much research indicating that, for children with SSD, language skills predict additional variance in phonological processing and literacy scores Semantics Receptive vocabulary explains ~25% of variance in PA in preschoolers (Rvachew & Grawburg, 2006; Preston & Edwards, in press) Age 5 6 semantics composite score significantly predicts age 7 9 literacy and PA (Peterson et al. 2009) 17

Which Language Skills? Morphosyntax Age 4 ½ MLU strong predictor of age 8 ½ reading (Bishop & Adams, 1990) Age 5 6 Expressive Syntax score explains significant variance in 7 9 yr PA and reading skills; speech skills only weakly predicted PA (Peterson et al., 2009) CELF Word Structure and Recalling Sentence scores related to Reading Recognition & Reading Comp in children grades K 2 (r>.57) (Bernhardt & Major 2005) 18

Which Children with SSD are most likely to have literacy problems? Nonverbal Cognition Nonverbal IQ found to relate to reading outcomes in children with SSD (Bird et al., 1995; Peterson et al., 2009) Index of general developmental delay? 19

Which Children with SSD are most likely to have literacy problems? Speech perception Preschool speech perception (as assessed by SAILS) predicted kindergarten PA (Rvachew, 2006) Perception problems poor phonological representations? 20

Which children with SSD are most likely to have literacy problems? Persisting Speech Problems Case Studies (e.g., Stackhouse, 1997) Lower PA scores for children ~6 yrs who still have a SSD vs. those whose SSD normalized (Nathan et al., 2004; Raitano et al., 2004) Children ages 10 14 with residual rhotic errors showed deficits in phonological awareness, phonological memory, and rapid naming (Preston & Edwards, 2007, 2009) 21

Which children with SSD are most likely to have literacy problems? Childhood Apraxia of Speech A neurological childhood (pediatric) speech sound disorder in which the precision and consistency of movements underlying speech are impaired in the absence of neuromuscular deficits (e.g., abnormal reflexes, abnormal tone) (ASHA, 2007) Clinical and theoretically important Motor based disorder resulting in problems with phonological bases of literacy 22

Which children with SSD are most likely to have literacy problems? Childhood Apraxia of Speech Children ages 6 11 with CAS below controls on reading and spelling (Stackhouse, 1982) Preschoolers with SSD followed to ages 8 9. CAS < S+LI < S only on spelling and reading comprehension (Lewis et al., 2004) 23

Which children with SSD are most likely to Severity have literacy problems? Inconsistent Findings: not strongly predictive of phonological awareness or literacy skills if speech is measured using standardized tests or PCC (Peterson et al., 2009; Preston, 2008; Rvachew, 2006, 2007; Smith et al., 2005) Is this a measurement issue? 24

SSD Severity and PA STUDY AGES N SPEECH MEASURE PA MEASURE Correlation r Preston & Edwards (in 4-5 yr 43 GFTA-2 standard score PA Principal Component (.15) press) PCC (.22) Bird & Bishop (1992) 5-6 yr 14 Percent distinctive features correct Onset identification.60 Onset segmentation and.47 matching Rhyme judgment (.35) Rhyme generation (.42) Major & Bernhardt 3-4 yr 19 PCC-R Metaphonology composite ρ=.57 (1998) Cons Matches ρ=.47 Percent Vowels Correct ρ=.64 Word shape match ρ=.62 CVC matches ρ=.52 CVCV matches (<.47) Nijland (2009) 21 PCC Percent Vowels Correct Rhyme judgment ρ=.48 ρ=.54 Smith et al (2005) 5-8 yr 187 GFTA cons errors Phonological Awareness -.31 zpcc-r Principal Component.10 Speech1 (log odds of SSD, composite of 7 variables) -.09 Rvachew (2006) Pre-K Kinder. Speech2 (probability of SSD, composite of 7 variables) -.11 47 GFTA-2 cons errors (Pre-K) Phonological Awareness Test -.45 Pre-K GFTA-2 cons errors (Kinder.) Phonological Awareness Test -.54 Kinder. (-.38) GFTA-2 cons errors (Pre-K) Phonological Awareness Test Kinder. -.47 First Grade Rvachew (2007) Pre-K Gr.1 68 GFTA-2 cons errors (Pre-K) Phonological Awareness Test Rvachew et al (2004) 3-4 yr 34 GFTA-2 cons errors Phonological Awareness Test (-.23) Webster & Plante (1992) 6-8 yr 22 Number of intelligible words (11 SSD) Pseudoword segmentation Word-syllable segmentation Word-phoneme segmentation.62 (.32).77 25

Correlation Coefficients: Severity and PA 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 26

SSD Severity and Literacy STUDY AGES N SPEECH MEASURE LITERACY CORR. MEASURE r Catts (1993) Kinder. Gr.1 56 GFTA consonant errors Word Attack (<.29) Word Identification (<.29) Stein et al (2004) Pre-K SSD & 177 PCC in conversation Word Attack.18 siblings Word Identification.12 188 GFTA percentile Word Attack.13 Word Identification.19 Rvachew (2007) Pre-K Gr.1 68 GFTA-2 errors (Pre-K) PCC (Pre-K) First Gr sight word reading (-.16) (.10) GFTA-2 errors (Pre-K) PCC (Pre-K) First Gr nonword reading -.39.30 Lewis et al 2002 ~9 yrs 52 GFTA percentile Test of Written Spelling 3 Total Predictible Unpredictible.39.43.32 Webster & Plante (1992) 6-8 yrs 22 (11 SSD) Number of intelligible words Printed word recognition (.35) 27

Correlation Coefficients: SSD Severity and Literacy 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 28

Theoretical Argument Is there a common mechanism that underlies speech sound production and (pre)literacy? One possibility: Accuracy of Phonological Representations (Bird et al., 1995; Rvachew et al., 2003, 2004; Rvachew & Grawburg, 2006; Senechal et al., 2004; Sutherland & Gillon, 2005) In production, poorly specified phonological representations may be indicated by atypical (unusual) speech sound errors 29

Types of Speech Sound Errors In the present study speech sound errors are classified as: Typical error patterns Atypical error patterns Distortions 31

Typical Error Patterns Common in typical phonological development, e.g., Final Consonant Deletion Cluster Reduction Velar Fronting Stopping of Fricatives Gliding of Liquids 32

Atypical Error Patterns Unusual or not commonly observed in typical phonological development, e.g., Initial Consonant Deletion Cluster Production Backing to Velars Glottal Replacement 33

Distortions Errors that are within the correct phoneme category (sub phonemic), e.g., Dentalization Lateralization Labialization Derhotacization (of /r/) 34

Research on Types of Speech Sound Errors & PA Mixed results re: phonological processing & error types Trend for 6 yr olds with SSD & low PA to produce more atypical /unusual errors (Leitao et al. 1997) no statistics, just observations Children with more atypical errors at 6 yrs scored lower on phonological processing at 12 13 yrs (Leitao et al., 2004) n= 7 per group, did not consider vocabulary differences Preschoolers with low PA had more typical omissions than preschoolers with normal PA (Rvachew et al., 2007) Kindergarteners with low PA had more atypical substitutions than peers with normal PA (Rvachew et al., 2007) 35

The Current Study Is the relative occurrence of different types of speech sound errors related to phonological processing in preschoolers with SSD? Considers both the frequency and types. Theoretical assumption: poorer performance on phonological processing tasks and less precise speech sound output both reflect weaker phonological representations 37

Participants Monolingual English speaking children with SSD of unknown origin Primarily clinical referrals Ages 4 5 Not yet in kindergarten 38

Part I: Screening, n=51 Speech Sound Production Goldman Fristoe Test of Articulation 2 <90 Oral Mechanism Screening, Hearing Screening Receptive Language (met at least 2 of 3 criteria) Peabody Picture Vocabulary Test 4 >80 Clinical Evaluation of Language Fundamentals: Preschool 2 Sentence Structure subtest >6 Concepts & Following Directions subtest >6 Nonverbal Cognition Differential Ability Scales Pattern Construction Subtest T score >37 39

Part II: Experimental Measures, n=43 Speech sound production 125 word picture naming task (480 consonants) Digitally recorded, phonetically transcribed Analysis of speech sound errors: Percent Consonants Correct (PCC) and Frequency of Different Types of Errors Distortions per consonant Typical sound changes per consonant Atypical sound changes per consonant 40

Experimental Measures: Part II Phonological Awareness Adapted from previous research (Bird et al., 1995; Rvachew & Grawburg, 2006) All early predictors of later literacy All require nonverbal responses Tasks assess: Rhyme Onset Matching Onset Segmentation & Matching Blending 41

Rhyme 16 items 42

02 43

Onset Segmentation and Matching 10 items 44

07 45

Onset Matching 10 items 46

03 47

Blending 12 items: 6 items spoken in two parts (onset rhyme): C VC 6 items spoken in three parts: C V C 48

Phonological Memory Syllable Repetition Task (Shriberg et al., 2006, 2009) 2 4 syllable sequences Included /ba, na, da, ma/ Example /naba, daba, nada, maba/ 50

Rapid Naming Monosyllabic: 5 pictures repeated 6 times in different sequences Disyllabic: 5 pictures repeated 6 times in different sequences 51

Example 52

Data Analysis Correlational design to examine concurrent relationships between speech sound production and phonological processing 53

RESULTS: Summary of PA tasks Principal Component Analysis summarized four Phonological Awareness Tasks 54

Phonological Awareness Composite (Principal Component) 55

Sound Error Types per Consonant 56

Results: PA Distortions Typical Sound Changes Atypical Sound Changes Correlation with PA ns ns r = -0.36 57

Results: PA Types of errors that reflect weak phonological representations will predict variance in PA beyond receptive vocabulary and age Variance in Phonological Awareness 58 Note: values are adjusted R 2

Results: PA Types of errors that reflect weak phonological representations will predict variance in PA beyond receptive vocabulary and age Atypical Sound Changes Typical Sound Changes Distortions Age & Vocab 33% Variance in Phonological Awareness 59 Note: values are adjusted R 2

Results: PA Types of errors that reflect weak phonological representations will predict variance in PA beyond receptive vocabulary and age Atypical Sound Changes Typical Sound Changes Distortions Age & Vocab 33% 5.9% Variance in Phonological Awareness 60 Note: values are adjusted R 2

Results: PA Types of errors that reflect weak phonological representations will better explain variance in PA than will PCC. PCC = 0% change Age & Vocab 33%* Variance in Phonological Awareness 61 Note: values are adjusted R 2

Results: Phonological Memory An analysis that considers three types of sound changes will predict variance in phonological memory beyond age and receptive vocabulary. Variance in Phonological Memory (Syllable Repetition Task) 62 Note: values are adjusted R 2

Results: Phonological Memory An analysis that considers three types of sound changes will predict variance in phonological memory beyond age and receptive vocabulary. Age & Vocab (nonsig) 5.6% Atypical Sound Changes Typical Sound Changes Distortions Variance in Phonological Memory (Syllable Repetition Task) 63 Note: values are adjusted R 2

Results: Phonological Memory An analysis that considers three types of sound changes will predict variance in phonological memory beyond age and receptive vocabulary. Age & Vocab (nonsig) 5.6% Atypical Sound Changes Typical Sound Changes Distortions 30.8% * Variance in Phonological Memory (Syllable Repetition Task) 64 Note: values are adjusted R 2

Results: Rapid Naming An analysis that considers three types of sound changes will predict variance in rapid naming beyond age and receptive vocabulary. Variance in Rapid Naming Tasks 65 Note: values are adjusted R 2

Results: Rapid Naming An analysis that considers three types of sound changes will predict variance in rapid naming beyond age and receptive vocabulary. Age & Vocab (nonsig.) 5.3% Atypical Sound Changes Typical Sound Changes Variance in Rapid Naming Tasks Distortions 66 Note: values are adjusted R 2

Results: Rapid Naming An analysis that considers three types of sound changes will predict variance in rapid naming beyond age and receptive vocabulary. Age & Vocab (nonsig.) 5.3% 9.9% * Atypical Sound Changes Typical Sound Changes Variance in Rapid Naming Tasks Distortions 67 Note: values are adjusted R 2

Summary More atypical sound changes associated with lower performance in 3 phonological processing domains (phonological awareness, phonological memory, rapid naming) Implications for early literacy problems May be diagnostically useful for SLPs May help to explain some inconsistencies reported in other studies relating SSD and phonological processing Distortions and typical sound changes were not associated with phonological processing Caveat: Still assess phonological processing in ALL preschoolers with SSD 69

Overall Summary Based on existing literature, children with SSD most at risk for (pre)literacy problems would include: Co existing language impairment Childhood Apraxia of Speech Persisting speech problems (~6 yrs and beyond) Low Nonverbal IQ Speech perception deficits Severity? Many Atypical errors 70

Where do we go next? SSD and literacy problems: Etiological similarities and differences Better measurement and classification of SSD Error patterns can be informative Neural phenotypes: what can the brain tell us about how these cognitive skills are related or how subgroups might be defined? Can neural characteristics help predict outcomes? 71

Acknowledgements American Speech Language Hearing Foundation Student Grant in Early Childhood Language Development NIH Training Grant to Haskins Laboratories (C. Fowler, PI) T32 HD 7548 8 72

Thank You! Questions? 73