A Language Endowment Spectrum Examining Evidence from a Developmental Perspective

1
A Language Endowment Spectrum? Examining Evidence
from a Developmental Perspective

• Susan Ellis Weismer
• Dept. of Communicative Disorders
• Waisman Center

2
Early Language Learning ProjectLongitudinal
investigation of late talkers

• Funding provided by NIH NIDCD grant
• R01 DC0371, Linguistic Processing in Specific
  Language Delay (Ellis Weismer, PI Evans and
  Chapman, Co-PIs)

3
Research Team

• Lab Managers
• Chris Hollar
• Heather Lohmeier
• Liz Schoonveld
• Kathy Schumacher
• Doctoral/Postdoctoral Students
• Maura Jones Moyle
• RaMonda Horton-Ikard
• John Heilmann
• Beth Roos
• Andy McDuffie
• Beth Coppoc Gunshor

• Student Hourly/PA
• Danna Accola
• Laura Jennings
• Jen Mulder Johnston
• Susan Bunton
• Laurie Eisenband
• Molly OShea
• Erin Green
• Kanika So
• Carrie Noskowiak
• Sheila Shippy
• Kelly Hickman
• Jim Kessenich
• Kari Faber
• Mindy Mentink
• Jessica Witt

4
Why Study Late Talkers?

• Interest in examining the full range with respect
  to rate and patterns of language development
• Interest in identifying toddlers who may be at
  risk for language impairment

5
(No Transcript)
6
Who Are Late Talkers?

• Typically identified by late onset of productive
  vocabulary
• Varying criteria cut points, ages
• Terminology
• ELD, SLI-E, LT
• Late talkers vs specific language impairment

7

• Separate lines of research focused on Late
  Talkers and children with SLI

8
Recent Theoretical Claims

• Childrens language abilities and disabilities
  - can be conceptualized in terms of a language
  endowment spectrum (Rescorla, 2000, 2002)
• Claim stems from the normal distribution account
  of SLI (Leonard, 1987, 1991)

9
Language Endowment Spectrum

• Typical talkers___ LB __ LT___ SLI
• According to this view, lower language skills are
  thought to reflect weaknesses in various skills
  that subserve language (e.g., auditory
  processing, working memory, motor planning)

10
Broad Research Questions

• What is the developmental course for late
  talkers? What are predictors of language
  outcomes?
• What is the extent of overlap between patterns of
  development and linguistic processing in typical
  talkers, late talkers, and children with SLI?
• Is this range of developmental language ability
  best characterized as a continuum or as distinct
  categories?

11
Contributions of Early Language Learning Project

• Included focus on linguistic processing measures
  as well as standard, knowledge-dependent measures
• Examined developmental patterns and learning
  mechanisms
• Identified late talkers using gender-based norms
  and nonverbal IQ only
• Larger sample

12
Participants and Methods
13
Initial Identification

• Late talkers identified at 20
• 10th percentile cut-off on the McArthur-Bates
  Communicative Developmental Inventory (CDI) used
  to identify late talkers vs. typical talkers

14
Participant Characteristics at 26
15
(No Transcript)
16
(No Transcript)
17
Assessment Measures
18
Assessment Measures
19
Early Word Learning
20
Why examine word learning?

• Vocabulary deficits are hallmark of late talkers
• Children with SLI have been shown to have
  deficits in novel word learning and fast mapping
  processes
• Early indicators of problems may be predictive of
  later language outcomes

21
Novel Word Learning Study Research Questions

• Do late talkers exhibit limitations in fast
  mapping processes compared to typically
  developing toddlers?
• Do late talkers display similar word learning
  patterns as typical talkers?

22
Novel Word Learning Study Participants
23
Novel Word Learning Task

• Two nonsense words - koob and tade- presented
  3 times each within puppet play activity
• Novel words (object labels) presented with 2
  familiar words
• Production and comprehension assessed after each
  exposure phase

24
(No Transcript)
25
Group Differences in Novel Word Learning
26
(No Transcript)
27
(No Transcript)
28
Analysis of Learning Patterns

• Amount of Exposure

29
(No Transcript)
30
(No Transcript)
31
Analysis of Learning Patterns

• Stimulus Features of Nonwords

32

• Phonotactic Probability refers to the likelihood
  of a particular sound sequence (diphone
  probabilities)
• Neighborhood Density refers to the number of
  phonologically similar neighbors, i.e. words
  differing by a single phoneme substitution,
  deletion, or addition

33
Stimulus Features
34
(No Transcript)
35
Individual Learning Patterns

• Analysis of individual learning patterns revealed
  that about 66 of NL group displayed a koob
  advantage
• 43 of the LTs (17) showed a koob advantage
• 20 of the LTs (8) showed a tade advantage
• 37 of the LTs (15) showed no preference

36
MLU for koob Advantage
37
Novel Word Learning Summary and Conclusions
38

• Summary LTs demonstrated lower total
  comprehension and production accuracy scores and
  displayed less steep learning slopes across 3
  exposure trials than NL controls
• Conclusion Late talkers display limitations in
  fast mapping processes compared to typically
  developing toddlers

39

• Summary Both groups comprehended koob better
  than tade and the NL group also showed a
  production advantage for koob
• Conclusions The majority of children in both
  groups appeared to demonstrate better learning of
  the low neighborhood density nonword
• The groups displayed qualitatively similar
  patterns with respect to their sensitivity to
  features of the nonwords

40
Relation between Lexical and Grammatical Skills
41
Theoretical Debate

• Dual-mechanism view lexical and grammatical
  development mediated by distinct mechanisms
  (Pinker 1991 Marcus et al., 2002)
• Single-mechanism view emphasizes
  interdependence between lexicon and grammar
  (Marchman Bates, 1994 Marchman et al., 2004)
• Continuity between lexicon and grammar in early
  acquisition

42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
Lexical-Grammatical Summary

• Findings support single-mechanism account for
  both typical and late talkers
• Both typical talkers and late talkers displayed
  semantic bootstrapping (influence of lexical
  skills on later grammatical abilities)
• Unlike typical talkers, late talkers did not
  display syntactic bootstrapping overlap with
  grammatical weakness in SLI

46
Developmental Patterns and Prediction of Language
Outcomes

• Early development
• Short-term Outcomes

47
Late Bloomers at 36
Total 40
48
Prediction of 36 MLU for Late Talkers
49
Prediction of 36 MLU for Late Talkers
Novel word accuracy CDI and PLS R2.65
50
Language Status Classification Results
Classification Variables Bayley (26), MLU
(26), PLS-Express. (26)
Novel Word Comp. (26)
51
Treatment Status
20 4 (10) 26 9 (22.5) 36 9 (22.5)
Number of Late Talkers
52
Treatment Classification Results
Classification Variables SES, PLS-Express
(26), Bayley (26), Arizona (26)
53
Developmental Patterns and Prediction of Language
Outcomes

• Later development
• Long-term Outcomes

54
(No Transcript)
55
(No Transcript)
56
Language Outcomes at 56

• Using performance on the TOLDP as the outcome
  measure, only 3/40 late talkers (7.5) scored at
  least -1 SD for production
• None of the late talkers had clinically delayed
  comprehension
• Consistent with prior research (Paul, 1996
  Rescorla Lee, 2000 Whitehurst Fischel, 1994)

57
56 Treatment Status
Number of Toddlers
58
Type of Intervention Received by Late Talkers at
56

• 8 Speech
• 6 Articulation
• 1 Oral motor/Dyspraxia
• 1 Fluency
• 7 Language/Speech

59
Performance Relative to Comparison Sample

• Although most late talkers moved into normal
  range (i.e. no CLINICAL DELAY), they continued to
  perform more poorly than the comparison sample
  matched on SES and nonverbal cognition on various
  language measures
• Consistent with previous findings (Rescorla,
  2002 Thal Katich, 1996)

60
TOLDP-3 Quotients at 56
61
TOLDP-3 Subtest Standard Scores at 56
62
Performance on CLPT at 56
63
Nonword Repetition at 56
64
Performance on Discourse Tasks at 56
65
Leiter-R Composite Scores at 56
66
Social Communication Questionnaire Scores at 56
67
Examination of Individual Scores

• 10 late talkers scored gt 2 SD below mean of the
  comparison group on spoken quotient of TOLDP and
  on sentence imitation subtest
• 17 late talkers scored gt 2 SD below mean of
  comparison group on nonword repetition

68
Examination of Individual Scores

• Some late talkers were NOT less proficient
  talkers at 56
• 10/40 (25) exhibited language skills equivalent
  to comparison group on all measures at 56
• Possible to get a late start but truly catch
  up but children who display this pattern are in
  the minority

69
Prediction of Productive Language at 56 (TOLDP)
26 Predictors PLS-comp, CDI, Bayley, PLS-prod
R2 .52
70
Role of Comprehension

• Comprehension at 26 was strongest single
  predictor of language production at 56
• Consistent with findings of Thal and colleagues
  (1991, 2004) for prediction at 3 years
• However, Paul (1991, 2000) did not find that
  comprehension was significant predictor of early
  school-age outcomes

71
Conclusions and Future Directions

• Many of the findings from this project support
  the notion of a language endowment spectrum
• Considerable overlap in patterns and mechanisms
  of language development for typical and late
  talkers, as well as certain similarities between
  late talkers and SLI

72
Future Directions

• Need to explain factors underlying differential
  trajectories of late talkers who truly catch up
  vs those with consistently lower language
  endowment
• Re-consider notion that continuum can be
  characterized in terms of global severity levels
  (Rescorla Roberts, 2002)
• Apparent mismatch between percentage of late
  talkers with clinical language impairment at 5
  years and percentage of kindergarten children
  with SLI

73
Future Directions

• Notion of language endowment spectrum relates to
  broader theoretical debates about dimensional vs
  categorical accounts of language disorders across
  populations (Dollaghan, 2005 Gernsbacher, Geye,
  Ellis Weismer, 2005 Zhang Tomblin, 2005)

74
Thank you