Beyond Phonology

1
Beyond Phonology

• Rebecca Treiman
• Washington University in St. Louis, USA

2
Plan

• Young children have difficulty using phonology in
  spelling
• Common view Early spellings are random strings
  of letters
• Our view Nonphonological spellings reflect the
  graphic patterns to which children have been
  exposed
• Evidence from children exposed to Portuguese and
  English (with Tatiana Pollo, Brett Kessler)

3
Phonology is difficult

• It is difficult at first for children to
  understand that writing represents linguistic
  forms as opposed to features of objects

4
Phonology is difficult

• 2. It is often difficult for children to analyze
  spoken words into phonemes
• /bl/ as two phonemes spell blow as BO
• letter name spellings KR

5
Phonology is difficult

• 3. Childrens classifications of phonemes may
  not match those assumed by the writing system
• Is second element of /sp/ /b/ or /p/?
• Is first element of /tr/ /t/ or /t?/?
• SBIDR
• CHRUK

6
How do children respond to the difficulty of
phonology?

• Phase theories of spelling development (e.g.,
  Ehri)
• Prealphabetic spellers string letters together
  randomly
• HS quick
• Partial and full alphabetic spellers use letters
  to represent speech sounds

7

• Common view Nonphonological spellings are random
  strings of letters
• Our view Nonphonological spellings are not
  random but show that children in literate
  societies have implicitly learned about some of
  the graphic patterns of print from their exposure
  to it

8
Preliminary evidence

• U.S. 3 and 4 year olds show some knowledge about
  the written forms of their own first names, to
  which they are frequently exposed (Treiman et
  al., 2007)
• Horizontal string of symbols
• Latin letters as opposed to other characters
• Evidence that children exposed to different
  writing systems pick up graphic differences

9
Pollos study

• Participants
• 79 Portuguese speakers from Brazil (mean age 4
  10 yrs.mos.)
• 51 English speakers from U.S. (mean age 4 8)
• Spelling task
• 18 words CVC, CVCV, CCV
• 18 pseudowords CVC, CVCV, CCV

10
Pretest results
11
How do we differentiate between nonphonological
and phonological spellers?
gummi
light
dice
blow
pony
try
12

• List plausible phonemeletter correspondences
• /b/ ? B
• /a?/ ? I, Y, IGH, optional E at the end
• /t/ ? T
• List plausible spellings of test items
• e.g., some plausible spellings of bite /ba?t/
• BITE
• BIT
• BYTE
• BYT
• BIGHTE
• BIGHT

13

• Score each spelling child produced, determining
  how far it differs from the best matching
  plausible spelling (string-edit distance)
• BIT for bite receives a good score
• Compare the scores that a child gets when his or
  her spellings are paired with the actual targets
  that the child was trying to spell to the scores
  that the child gets when the pairs of spellings
  and target words are randomly rearranged
• Do this 10,000 times
• Compare actual scores to rearranged scores

14
Children for whom fewer than 5 of the rearranged
scores are better than the actual score
Phonological spellers
15
Children whose actual scores are neither
significantly better than chance (i.e., mean of
rearranged scores) nor more than 1 better than
chance
Nonphonological spellers
16
Groups of children

• Phonological spellers
• 31 Portuguese speaking (mean age 5 4)21
  English speaking (mean age 4 11)
• Nonphonological spellers
• 35 Portuguese speaking (mean age 4 8)
• 23 English speaking (mean age 47)

17
Letter patterns in texts to which children are
exposed

• Books
• books designed for young children in Brazil and
  the U.S.
• Childrens names
• Names of children in U.S. and Brazilian
  schools that are similar to those in the study

18
Letter frequency in childrens books
19
Letter frequency in childrens spellings

• Nonphonological spellers tend to use letters that
  are frequent in the language to which they are
  exposed
• r .64, p lt .01 for nonphonological spellers
• r .55, p lt.01 for nonphonological spellers who
  dont have the letter in their own name

20
Proportion of vowel letters in nonphonological
spellers productions
21
Bigram frequency

• Nonphonological spellers tend to use bigrams that
  are common in the language to which they are
  exposed
• Children exposed to Portuguese show more
  consonant-vowel alternations than children
  exposed to English

22
Alphabet sequence

• Children in the U.S. and Brazil sometimes see and
  hear the letters of the alphabet in order, as in
  the Alphabet Song and alphabet books
• Nonphonological spellers in both countries use
  letters in alphabetical order more often than
  expected by chance

23
Childrens own names

• Children often see their own written name, and
  they pay special attention to it
• Nonphonological spellers exposed to both
  Portuguese and English overuse letters from their
  own first names
• Use of a Sam more than Bob
• and similarly for other letters

24
Summary of findings about nonphonological
spellers

• Use letters that are common in their language
• Use bigrams that are common in their language
• Use letters from their own names
• Use alphabet sequences
• Produce somewhat different spellings depending on
  the written language to which they have been
  exposed

25
Statistical learning

• People track relative frequency of individual
  elements
• People track frequency of groups of elements that
  are processed together
• People are sensitive to gradations of frequency,
  not just high vs. low
• Such learning begins at an early age knowledge
  becomes more refined with increasing experience
  in a domain

26
Statistical learning and spelling

• Before they relate letters to sounds, children in
  literate societies begin to learn about
  frequencies of letters and pairs of letters
• Later, children begin to track frequency of
  various sound-to-spelling links
• Using phonology in spelling and reading is
  difficult, but children pick up useful
  information about their writing system before
  they begin to learn how phonology is symbolized
  in print

27
Phonological spellers

• Common view Beginning alphabetic spellers rely
  on phonological information they dont know or
  use graphotactic patterns
• Our view Graphic knowledge continues and deepens
  in phonological spellers
• Evidence
• Intrusions in spelling correlate with letter
  frequency
• Even young children prefer nonwords with common
  letters and bigrams (e.g., jeeb vs. jiib for
  English)

28
Conclusions

• Phonology is important for spelling and reading,
  but its hard
• Learning about graphic patterns is easier it
  begins early and is important throughout the
  course of spelling development
• A statistical learning perspective can help us
  understand the acquisition of phonological and
  graphic knowledge

29
Implications and future directions

• Researchers and educators should not ignore early
  nonphonological spellings
• Can these spellings help us to identify, at an
  early age, children who will have reading and
  spelling problems?

30
Thank you!
31
(No Transcript)
32
Childrens Own Names
33
Alphabet