Written Expression of College Students With and Without Dyslexia and ADHD

1
Written Expression of College Students With and
Without Dyslexia and ADHD

• Chris Coleman, M.A.
• Noel Gregg, Ph.D.
• Donald Rubin, Ph.D.
• J. Mark Davis, Ph.D.
• Robert B. Stennett, M.S.
• The University of Georgia
• British Dyslexia Association 2001

2
Studies Presented

• Discourse Complexity in the Expository Essays of
  University Students With and Without Dyslexia and
  ADHD
• Assessment of Vocabulary and Verbosity in the
  Expository Essays of University Students With and
  Without Dyslexia and ADHD

3
Effects of Spelling Mistakes

• Young kids watch TV and see people who live in
  milion dollar homes, drive Ferries, and spend
  money like theres no tomorrow.
• Walk into any church on a Saturday and you can
  find two people exchanging rings and vowels.
• Im proud of the work I did as a candy stripper
  at St. Mareys Hospital.

4
Research Questions

• Beyond overt surface errors (e.g., misspellings),
  are there other differences between the essays of
  adults with dyslexia /- ADHD and the essays of
  adults without disabilities?
• Why do students with disabilities receive poorer
  marks on essays than their normal peers do?
• What accommodations are appropriate for
  university students with dyslexia /- ADHD during
  in-class writing assignments/exams?

5
Groups

• Group 1 (n77) diagnosed with dyslexia (LD)
• Group 2 (n44) diagnosed with attention-deficit/
  hyperactivity disorder (ADHD)
• Group 3 (n52) diagnosed with both dyslexia and
  ADHD (comorbid)
• Group 4 (n90) no history of diagnosis, special
  education, or learning problems (normal)

6
Identification of Subjects

• Groups 1-3 University students evaluated at the
  UGA Regents Center for Learning Disorders
  following a comprehensive (12-hour) assessment
  and in accordance with Georgia Regents criteria
• Group 4 UGA students enrolled in undergraduate
  speech communications courses

7
Measures

• Writing Task (30 min.)
• Vocabulary Counts
• Linguistic Variable Counts (54)
• Impressionistic Quality Ratings
• IQ Estimate
• WRAT-3 (Spelling)
• SAT Scores, GPA
• Demographic Information

8
Refinement of Subject Pool

• Initial n 325
• Controlled for
• Native language (English)
• Educational level (college)
• Learning history (no indication of learning or
  attention problems in Group 4)
• Writing instruction (passed English 101/1101)
• Estimated intelligence (average to high average
  range)
• Final n 261

9
Characteristics of Sample
10
Writing Samples

• 30 minutes, standardized instructions
• Sample prompt Discuss a major problem in high
  schools today and how it could be corrected.
• Processing of essays
• All subjects assigned arbitrary identification
  numbers
• Essays word-processed exactly as written
• Corrected version of each essay created in
  order to eliminate spelling mistakes, punctuation
  errors, and possibility of handwriting bias

11
Impressionistic Quality Ratings

• Multiple raters (acceptable ICC estimates,
  .73-.83)
• Georgia Regents rating formula applied to
  corrected essays by highly experienced raters
• Content/organization (1-4) (x2)
• Style (1-4)
• Conventions (1-4)
• Sentence Structure (1-4)
• ? ? ?
• Overall Quality (5-20 points per rater)

12
Impressionistic Rating Results
13
Intercorrelations of Rating Scales
14
Study 1 Discourse Complexity

• Purpose 1 To determine the frequency of specific
  syntactical/grammatical features in expository
  essays and the relationships of selected features
  to verbosity and quality
• Purpose 2 To compare feature patterns across
  Groups 1-4

15
Quantitative Linguistic Analysis

• Corrected essay files tagged by Bibers
  computerized analysis program (identification of
  the grammatical category and function of each
  lexical item)
• Initial results checked and edited (fixtagging)
• 54 linguistic features quantified
• Word counts, type-token ratios also generated
• All frequency counts adjusted to 1,000-word text
• Means and SDs obtained for each group

16
Structural Equation Modeling

• SEM Causal modeling in which The measured
  variables are used to make up latent variables
  and the main focus of the analysis is on the
  causal relations (the paths) between the latent
  variables... (Aron 1997)
• Four-factor model applied
• High degree of fit achieved with Group 4

17
Four-Factor Model

• Time (past tense, perfect aspect, stative be)
• Reference Elaboration (wh relative clauses
  subject, object, pied piping)
• Reduction (do as pro-verb, it, indefinite
  pronouns)
• Framing Elaboration (attributive adjectives,
  adverbs)

18
Downstream Factor/Variables

• Verbosity
• Fluency (word count)
• Type/Token Ratio (TTR)
• Quality (based on impressionistic ratings)

19
(No Transcript)
20
Study 1 Results

• Factor structure consistent across groups
• Factor loadings vary as a function of group
  membership
• Groups using the same linguistic structures, but
  in different distributions

21
Study 2 Vocabulary, Verbosity

• Purpose 1 To compare several quantitative
  methods of estimating vocabulary sophistication
• Purpose 2 To determine the importance of
  vocabulary and verbosity to quality
• Purpose 3 To compare regression patterns across
  Groups 1-4

22
Vocabulary Counts

• Initial type/token counts generated by
  concordance program, then refined by hand
• Additional counts
• Greater than 2 syllables (tokens, types)
• Greater than 3 syllables (tokens, types)
• Uncommon (freq rank gt100) words (tokens, types)
• Percentage of uncommon types
• Word length (from Biber CBA computer program)
• Length-adjusted counts also calculated
• (raw score) x (1000/tokens)
• Primary, reliability raters (ICC estimates
  .97-.99)

23
Best Model for Predicting Quality(verbosity as
step 1)
24
Best Model Applied to Each Group
25
Group Differences (Verbosity, Vocabulary)
26
Study 2 Results

• Verbosity strongest quality predictor
  (dyslexics fluency ADHD efficiency)
• Lexical sophistication/diversity (like verbosity)
  even more predictive of quality among dyslexic,
  ADHD writers than among general college sample
• Despite spelling/proofreading help, writers with
  disabilities still rated significantly lower on
  all scales
• Identification/accommodation hints
  (dyslexic and ADHD writers work quite
  differently)

27
Data Related to Example Essays
28
Answers to Research Questions

• Group differences in fluency, syntax, semantics,
  and quality evident despite
  spelling/proofreading assistance
• Primary reason for poor essay marks
  insufficient productivity
• Accommodation options extended time spelling
  and proofreading help thesaurus (consider
  individual strengths/needs)

29
Ongoing/Future Research

• Impressionistic ratings of uncorrected essays
• Relationships among different proxy counts
  (uncommon, syllables, Greco-Latinate)
• Group differences related to other types of word
  knowledge (spelling, receptive vocabulary)
• Discriminant analysis (identifying groups)

30
Contact Information

• Chris Coleman
• 335 Milledge Hall
• University of Georgia
• Athens, GA 30602
• ccoleman_at_arches.uga.edu