Lexical and Fluency Changes in Young Children Who Stutter

1
Lexical and Fluency Changes in Young Children Who
Stutter

• Stacy A. Wagovich
• University of Missouri-Columbia
• Nancy E. Hall
• University of Maine

5th World Congress on Fluency Disorders July 28,
2006
2
Stuttering in the Context of Language Development

• The rationale
• Stuttering onset typically occurs in early
  childhood, between ages 20 and 50.
• Tremendous language growth occurs during these
  three years.
• At age 20, children are in Browns Late Stage
  I/Early Stage II by age 50 they are beyond
  Stage V.
• During this time, MLU develops from about 1.5
  morphemes to 5.0 morphemes.
• Browns 14 grammatical morphemes are mastered.
• Use of negation and questions are mastered.

3
Recent Studies of Language in CWS

• Group differences between young CWS and CWNS in
  language performance, with CWS showing relatively
  weaker (but not disordered) language skills
• Anderson Conture (2004)
• Anderson, Pellowski, Conture (2005)
• Bernstein Ratner Silverman (2000)
• Bajaj, Hodson, Schommer-Aikins (2004)
• Logan (2003)
• Pellowski Conture (2005)
• Silverman Bernstein Ratner (2002)
• Language skills of CWS exceeding those of CWNS
• Watkins, Yairi, Ambrose (1999)
• Bonelli, Dixon, Bernstein Ratner, Onslow (2000)

4
Measurement of Language Skills vs. Language
Development in Relation to Fluency

• Measurement of language over time enables
  description of developmental processes, rather
  than the status of skills at a particular time.
• Yairi colleagues
• Watkins Yairi (1997)
• Watkins, Yairi, Ambrose (1999)
• In relation to treatment
• Bonelli, Dixon, Bernstein Ratner, Onslow (2000)
• Onslow, Bernstein Ratner, Packman (2001)
• Lattermann, Shenker, Thordardottir (2005)

5
Interplay between Language and Fluency in CWNS

• Rispoli (2003)
• Cross-sectional design
• 52 children, ages 110 - 40
• Found that, as the children developed more
  advanced language, their rate of revisions
  increased, but their rate of stalls did not.
• Hall Burgess (2000)
• Longitudinal design
• Case study, female, age 29
• Found that, over time, speech disruptions changed
  from repetitions and pauses to an increased rate
  of revisions

6
CWS Fluency Changes over Time

• Throneburg Yairi (2001)
• Examined changes in fluency of CWS over a 3-year
  time period.
• Children who continued to stutter over 3-year
  period differed in the pattern of their
  stuttering from children who recovered from
  stuttering.

7
Lexical Skills in CWS

• Anderson Conture (2000) vocabulary testing
  and dissociations across language domains
• Pellowski Conture (2005) lexical priming
• Silverman Bernstein Ratner (2002)
  conversational vocabulary
• See Hall (2004) for a review

8
Purposes of the Study

• To examine how the conversational vocabulary of
  young CWS changes over time.
• To examine whether lexical diversity and rarity
  fluctuate across samples in correspondence with
  fluency fluctuations.

9
Research Questions

• Changes over Time - How does conversational
  vocabulary change over time, from samples
  obtained at the beginning of the study to samples
  obtained at the end?
• Changes with Stuttering Frequency - How does
  conversational vocabulary differ, comparing a
  childs most fluent samples to the childs most
  disfluent samples?

10
Participants

• 11 children participated in the study,
  contributing a monthly language sample for 10
  months
• Parents of each child reported that the child
  stuttered
• Two SLPs, each experienced in identifying the
  presence of stuttering, individually confirmed
  the presence of stuttered speech within the
  samples.

11
Participants (Contd.)

• Additional criterion for participation At least
  3 stuttered syllables (SS) on at least one
  sample over the course of the 10 sample study.
• Two children of the 11 did not meet this
  criterion and so are not included in the analyses.

12
Participant Information
13
Language Test Performance
14
Procedure

• Testing (1-2 testing sessions at the beginning of
  the study)
• PLS-4
• EOWPVT-2000 Edition
• ROWPVT-2000 Edition
• Vineland Adaptive Behavior Scales Long Form
• Careys Temperament Scales
• SSI-3
• P1 received the PPVT-3 instead of the ROWPVT

15
Language Sample Elicitation

• Ten 100-utterance language samples were obtained,
  each approximately one month apart.
• Samples were play-based and elicited by a
  graduate student examiner trained in language
  sample elicitation.
• All samples were elicited by the same examiner.

16
Representativeness of Language Sample Data

• External factors can impact the quality of a
  language sample (e.g., childs mood, toys, time
  of day, clinicians responses).
• Samples with an unfamiliar examiner can be less
  representative of the childs language skills.
• Samples of greater size and samples taken over
  multiple sessions can minimize the impact of
  these factors.

17
Language Sample Selection and Merging Analyses
Over Time

• First sample was considered a familiarization
  sample and was discarded from analyses.
• The subsequent 9 samples were merged into sets of
  3
• Samples 2, 3, and 4 ? Time 1
• Samples 5, 6, and 7 ? Time 2
• Samples 8, 9, and 10 ? Time 3
• Data from Time 1 and Time 3 were compared to
  analyze the effect of time on conversational
  vocabulary.

18
Language Sample Selection and Merging
Stuttering Frequency

• First sample was considered a familiarization
  sample and was discarded from analyses.
• The subsequent 9 samples were merged into sets of
  3. For each child
• Samples containing the least stuttering ? Set 1
• Samples in the middle ? Set 2
• Samples containing the most stuttering ? Set 3
• Data from Set 1 and Set 3 were compared to
  examine correspondence between stuttering
  frequency and conversational vocabulary.

19
Sample Coding and Analyses

• For this project, samples were coded according to
  CHAT conventions (MacWhinney, 2005)
• Fluency Analysis Stuttered disfluencies were
  coded manually
• Utterance-by-utterance agreement for 20 of the
  samples ranged from 82.1 to 98.1. Average
  agreement was 92.7

20
Language Sample Analyses

• Number of different words (NDW)
• Total number of words (TNW)
• Vocd (Malvern, Richards, Ngoni, Duran, 2004)
  CLAN program that computes D, a measure of
  lexical diversity that accounts for differences
  in samples lengths
• Lexical rarity number of different rare words
  in the samples (Beals Tabors, 1995 see
  Bernstein Ratner Silverman, 2000)
• crocodile, slither, kiwi

21
Results
22
Changes over Time TNW
d .87, large effect
23
Changes over Time NDW
d 1.2, large effect
24
Changes over Time Number of Different Rare Words
d .95, large effect
25
Changes over Time Vocd
d .72, medium to large effect
26
Growth over Time

• Language growth is not linear.
• For CWNS, language growth over the course of a
  year can be predicted (to some extent) based on
  previous language achievement.
• The same predictions for CWS may not hold,
  however.
• Watkins Johnson (2004)
• Bonelli, Dixon, Bernstein Ratner, Onslow (2000)

27
Proportional Change Index (PCI Wolery, 1983)
Vocd Gain Total Time in Study ____________________
______ EOWPVT Language Age (Mos.) Chronological
Age (Mos.)
28
Comparison of PCI and Simple Change
29
Stuttering over Time
30
Vocd for Recovering versus Other Pattern
Group
31
Stuttering Frequency Changes in Relation to
Language TNW
Indicates participant showing pattern of
recovery d .38, small to moderate effect
32
Samples Contributing to Stuttering Frequency Sets
33
Stuttering Frequency Changes in Relation to
Language NDW
Indicates participant showing pattern of
recovery d .49, moderate effect
34
Stuttering Frequency Changes in Relation to
Language Lexical Rarity
Indicates participant showing pattern of
recovery d .51, moderate effect
35
Stuttering Frequency Changes in Relation to
Language Vocd
Indicates participant showing pattern of
recovery d .46, moderate effect
36
Discussion
37
Conversational Vocabulary over Time

• Over time, most children showed an increase in
  the total number of words (7/9) and the number of
  different words (8/9) produced.
• Most children (8/9) produced more lexical items
  considered rare over time.
• Most of the children (6/9) showed an increase in
  vocd over time.

38

• Wolerys Proportional Change Index (1983)
  produced similar findings to the computation of
  simple change (Time 3 Time 1).
• The three participants who showed negative change
  in vocd each displayed a pattern that suggested
  recovery.
• However, the 4th participant seemingly headed
  toward recovery showed positive change in vocd.

39
Fluctuations in Conversational Vocabulary with
Changes in Stuttering Frequency

• Most children (7/9) had a higher total number of
  words and number of different words for samples
  that had more stuttering (Set 3) than samples
  that had less stuttering (Set 1).
• Most children (6/9) produced more rare word types
  in samples that had more stuttering.
• Slightly over half of the children (5/9) had
  higher Vocd scores for samples with more
  stuttering (moderate effect).

40
Limitations and Future Directions

• Limitations
• Small sample size
• Limited range of stuttering severity
• Wide age range
• Future Directions
• Studying a small developmental window with
  reference to language development milestones
  (similar to Rispolis method), and
• Including CWNS as language-matched controls

41
Acknowledgements

• We would like to thank the University of Missouri
  students who assisted in data collection and
  analysis - Meghan Malone, Dana Staley, Mallory
  Wood, Kim Wilt, Ellen Marschner, and Lauren
  Haney.
• Thanks also to Matthew Krause, CCC-SLP, for
  providing reliability coding.
• This research was supported by University of
  Missouri Research Council funding awarded to the
  first author.