Sensitivity to orthographic structure influences letter position encoding

1
Sensitivity to orthographic structure influences
letter position encoding

• Nikki Pitchford, Tim Ledgeway,
• Jackie Masterson Maria Ktori

2
Introduction

• One of the first processes involved in
  orthographic processing is the encoding of letter
  identity and position
• This enables us to distinguish anagrams
• EG
• ABLE BALE
• NONDIALECTIC COINCIDENTAL

3
Introduction

• One of the first processes involved in
  orthographic processing is the encoding of letter
  identity and position
• This enables us to distinguish anagrams
• e.g.,
• ABLE BALE
• NONDIALECTIC COINCIDENTAL

4
Introduction

• Learning to encode letter identity and position
  accurately and rapidly is essential for becoming
  a skilled reader
• One mechanism by which orthographic learning may
  occur is extraction of statistical regularities
  embedded within orthography
• Orthographic regularities and irregularities may
  become encoded through increased exposure to
  print
• Accordingly, skilled readers should show
  sensitivity to distributional characteristics
  inherent to orthography

5
Introduction

• We have been investigating if letter identity and
  position encoding is modulated by statistical
  learning
• Used a visual letter search task
• Requires identification of a previously cued
  letter target embedded within a random 5-letter
  string
• For 50 of the trials the cued letter appears as
  a letter target within the random 5-letter string
• Position of the letter target varies across trials

6
Target Present (YES)
7
Target Present (YES)
8
Target Present (YES)
9
Target Absent (NO)
10
Target Absent (NO)
11
Target Absent (NO)
12
Aims Predictions

• To investigate if letter identity and position
  encoding is mediated by statistical learning
• If so, skilled readers of different orthographies
  should be sensitive to the distributional
  characteristics of their orthography
• If statistical learning is a general mechanism
  underpinning orthographic processing dyslexic
  readers may not be sensitive to orthographic
  regularities

13
Method

• To test these predictions we compared letter
  search performance of three groups of adult
  readers over a series of studies
• Skilled English (N28) Pitchford, Ledgeway
  Masterson (2008) Journal of Research in
  Reading, 31, 97-116
• Skilled Greek (N24) Ktori Pitchford (2008)
  Language Cognitive Processes, 23, 258-281
• Dyslexic English (N29) Pitchford, Ledgeway
  Masterson (in
• press) Quarterly Journal of Experimental
  Psychology
• Here, we explore differences in RT to identify
  correctly different letters across the 5 string
  positions in relation to positional letter
  frequency

14
Positional letter frequency

• Refers to the number of times a letter appears in
  a particular position in words of a specified
  length
• For English participants counts drawn from a bank
  of 20,000 words were used (Mayzner Tresselt,
  1965)
• EG Letter A Freq
• P1 ADULT 255
• P2 CAMEL 349
• P3 BLANK 383
• P4 PECAN 165
• P5 EXTRA 12
• For Greek participants counts were drawn from a
  bank of 35,000 words (Ktori, van Heuven,
  Pitchford, 2008)

15
Question 1

• Is letter identification speeded for most common
  positions?

16
Letter identification in the most vs least
frequent position

• Compared mean RT to identify correctly letter
  targets appearing in the most compared to least
  frequent position within 5-letter words across
  the set of 25 letters
• EG P1 P2 P3 P4 P5
• Letter A freq 225 349 383 165 12

17
RT differences across the most vs least frequent
letter position
Mean RT (ms)

Reader Group
18
RT differences across the most vs least frequent
letter position
Mean RT (ms)


Reader Group
19
RT differences across the most vs least frequent
letter position
ns
Mean RT (ms)


Reader Group
20
RT correlations across skilled and dyslexic
readers with RT

• Most frequent position
• Skilled and dyslexic readers RT correlated
  significantly
• r .75, (N 25), p lt .0001, 2-tailed
• Least frequent position
• Skilled and dyslexic readers RT did not
  correlate significantly
• r .21, (N 25), ns

21
Question 2

• Is letter identification sensitive to letter
  frequency across all positions?

22
Correlations with RT and letter frequency per
position
23
Correlations with RT and letter frequency per
position
24
Correlations with RT and letter frequency per
position
25
Question 3

• Is identification of frequently occurring letter
  sets (vowels) speeded compared to less frequent
  letter sets (consonants)?

26
RT differences across vowels vs consonants
Mean RT (ms)

Reader Group
27
RT differences across vowels vs consonants
Mean RT (ms)


Reader Group
28
RT differences across vowels vs consonants
ns
Mean RT (ms)


Reader Group
29
Conclusions

• Letter identity and position encoding is
  modulated by positional letter frequency in
  skilled readers
• Dyslexic readers show reduced sensitivity to
  positional letter frequency when identifying
  letters
• Sensitivity to orthographic regularities when
  encoding letters is likely to emerge with reading
  experience
• either from the bank of whole-word
  representations in the lexicon
• and/or at abstract letter units through a process
  of statistical learning
• Reduced sensitivity of dyslexic readers to
  orthographic regularities may arise from either
• an impoverished orthographic lexicon
• and/or a general deficit in statistical learning