Vowel Development in Children`s Writing: A Cross-Grade

Vowel Development in
Children’s Writing: A CrossGrade and Genre Comparison
Kenn Apel, PhD
Elizabeth Wilson-Fowler, MA
Danielle Brimo, MS
Florida State University
&
Cynthia Puranik, PhD
University of Pittsburgh
Introduction
• Spelling is a complex, linguistic skill that develops
over time (e.g., Apel, Masterson, & Niessen, 2004;
Bourassa & Treiman, 2001).
• Most developmental studies of spelling have focused
on consonants; vowel spellings are considered to be
difficult to produce early in development (e.g.,
Varnhagen, Boechler, & Steffler, 1999).
• Investigations of vowel development in spelling are
few (e.g.,Caravolas, Kessler, Hulme, & Snowling,
2005; Reece & Treiman, 2001; Treiman, 1993;
Varnhagen et al., 1999)
Overview of Presentation
• Knowledge about vowel development is important to
obtain an understanding of how children learn the
spelling system.
• In today’s presentation, we will:
– Briefly review what is known about young children’s vowel
spelling development, and
– Report findings from a cross-sectional investigation of the
vowel spellings of students in grades 1-4
Developmental Spelling
• Children’s writings gradually progress from
drawing and scribbles to letters as they gain
awareness of the English alphabetic system
(e.g., Gentry, 1982).
• When compared to the development of
consonant orthographic representations,
young elementary-age, English-speaking
students struggle more with the
representation of vowels in their spellings
(e.g., Treiman, 1993; Sawyer, 1999; Stage &
Wagner, 1992).
Developmental Spelling
• Initially, vowel representations are likely to
be omitted from children’s writings ( Ehri,
1986; Morris & Perney, 1984).
• Vowels that are located in the middle and end
of words are most likely to be omitted.
• For example, a young child might write “D” when
asked to write the word ‘dog’ or spell ‘cat’ as “KT,”
thus representing only one or more of the
consonants (Henderson, 1981).
Vowel Spelling Development
• Read (1971, 1975) studied the invented writings of
20 children ages 3 years to 6 ½ years and found
consistent patterns of errors which changed as the
children developed.
– Children represented vowels by either
substituting a different vowel letter or by
omitting the required vowel.
Vowel Spelling Development
• Read (1971, 1975)
– Vowel substitutions were most common for long and
short vowels.
• Short vowels were represented by the long vowel letter
that was closest in place of articulation and acoustically
to the targeted short vowel (e.g., letter-name spelling:
‘pet’ spelled as ‘PAT’.
• Long vowels also represented by the letter name of the
long vowel sound (e.g., ‘bak’ for ‘bake’)
• Vowels adjacent to liquids (e.g., little) and nasals (e.g.,
wagon) in two-syllable words were commonly omitted.
Vowel Spelling Development
• Jorm (1977) administered a 60 word spelling
test to 138 students, grades three through five
• The words consisted of four spelling structures:
CCVVC, CVCVC, CVCCVC, and CVCCVCC.
• Jorm’s findings indicated that children had
difficulty spelling unstressed vowel patterns and
the second vowel-letter in the CCVVC words (e.g.,
brain). This second finding was particularly true
for poorer spellers.
Vowel Spelling Development
• Stage and Wagner (1992) assessed the
nonsense word spelling ability of 187 students
in kindergarten through third grade.
– Students performed better with consonant
spellings than vowel spellings.
– The average proportion correct spellings for
vowels, which was less than 50%, was
substantially lower than for consonant spellings.
Vowel Spelling Development
• Treiman (1993) found similar results to both
Jorm (1977) and Stage and Wagner (1992) in
her seminal research with first grade students
– Students struggled with accurately representing vowel
patterns more than consonants because vowels have
“more different spellings” than consonants and “the
likelihoods of the alternatives are more similar” (p. 91).
– She concluded that it is the properties of English
orthography that explain why students have more
difficulty with vowels than consonants in their spelling.
Vowel Spelling Development
• Caravolas, Kessler, Hulme, & Snowling (2005)
assessed the spelling ability of British children in
their first two years of schooling.
– administered a 95 word spelling test of monosyllabic,
monomorphemic words that were common and highly
imageable objects and actions
– strongest, unique predictor of children’s accurate spelling
of vowels was the probability that the vowel phonemes
were represented by the target vowel grapheme in words
encountered by students in year one and two of school,
suggesting vowel spellings rely on MGRs
Common Error Analyses Across
Studies
• Previously, investigators have analyzed
children's vowel spellings using
– Correct vs. incorrect scoring
– Substitution/omission/addition scoring
• Including pattern analysis
– Legal/illegal scoring
• Legality can be based on phonological-plausible
spellings (e.g., ‘kat’ for ‘cat’)
Limitations of Past Research
• Much more information is required to understand vowel
spelling development.
• Most studies have taken a narrow approach to studying vowel
spelling development using dictated spelling tasks.
– Dictation tasks do not mirror typical writing situations or account for
the linguistic demands typically present when composing text (e.g.,
syntactic and semantic demands).
– Dictation tasks, though, prevent children from avoiding words they
can’t spell.
• Treiman’s (1993) assessed vowel spellings in children’s
spontaneous writing.
– Her task mirrored functional writing situations but were limited to a
single grade.
Limitations of Past Research
• Most investigators have sampled spellings of children from:
– One grade or a limited number of grades
– One genre
• Given that different writing discourse genres can tax a
writer’s abilities in different ways (e.g., Singer & Bashir,
1999), assessment of spelling ability involving different
genres may provide a broader picture of children’s
developing skills.
• Most investigators have only provided vowel accuracy results.
– Others (e.g., Green, McCutchen, Schweibert, Quinlan, EvaWood, & Juelis, 2003), have suggested that measures of
prevalence may be equally important.
Purpose of Study
• The current study took a broad approach to
the study of vowel spellings across grades 1-4.
Specifically, we asked the following questions:
– Does grade and/or genre affect:
• the prevalence of vowel spellings (short, long,
diphthong, r-controlled, and l-controlled) in children’s
written compositions?
• the accuracy of vowel spellings in children’s written
compositions?
• the source of error in children’s incorrect vowel
spellings?
Definitions
•
•
•
•
Diphthong: ‘ow’ ‘oy’
R-controlled: syllabic ‘r’ (father, stir)
L-controlled: syllabic ‘l’ (little, pedal)*
Narrative Discourse: description of a
happening/event
• Expository Discourse: writings meant to
describe, argue, persuade, provide an opinion,
etc.
* l-controlled vowels did not occur frequently enough for analyses
Method
• Participants
– 132 first through fourth grade students; 1st (29);
2nd (32); 3rd (39); 4th (32)
– Students chosen from schools that varied in
socioeconomic level
Writing Samples
• The children were given a piece of paper, pencil, and 15
minutes to answer to writing prompts:
– Narrative prompt: Imagine something special that happened in
your life. What was that special time? Who was involved? How
did it make you feel? (Did you learn anything from that
experience?)
– Expository prompt: Imagine an animal you would like for a
classroom pet. Why would you like to have that animal for a
classroom pet? What is so special about this animal? Why would
this animal be the best choice for a classroom pet? How would
the classroom benefit from having this pet?)
• No support from peers or adults in the classroom was
provided
Coding of Vowel Spellings
• Vowel categories:
– Short (e.g., hat)
– Long (e.g., rope)
– Diphthong (e.g., boy)
– r-controlled (e.g., butter)
• Vowels assessed in:
– Base words (e.g., hand, cake, bark)
– Within the base of affixed words (e.g., swimming,
recycle, animals)
Coding of Vowel Spellings
• Analysis of vowel categories were conducted if at least five
instances of the category occurred in the sample for base
words and within the base of affixed words
• Vowel responses were analyzed for:
– Prevalence: Number of instances of vowels within vowel
categories regardless of accuracy
– Accuracy: Percent correctly spelled out of total
opportunities
Coding of Vowel Spellings
• Vowel responses were analyzed for:
– Error Types:
• Phonologic: vowel phoneme not represented by a grapheme
(e.g., ct for cat)
• Orthographic: vowel phoneme represented with “illegal”
grapheme (e.g., tip for chip)
• MGR (mental graphemic representation): vowel phoneme
represented with a legal, but incorrect grapheme (e.g., keap
for keep)
• Homophone: vowel phoneme represented by a grapheme,
but result is a homophone of the word (e.g., one/won)
• Reversal: vowel phoneme represented, but order disrupted
by reversal between vowel and adjacent consonant (e.g.,
battel/battle)
Results
• Reliability (inter-rater agreement)
– Narrative: .90
– Expository: 90
• Construct Validity of Samples
– College students (blind to study) were trained to
assess expository and narrative samples; provided
definitions of each during validity task
– 15% of samples rated as narrative, expository, or
“not sure”
– 82% agreement between assigned genre and
intended genre
Results
• Text Production Information
– Length (# of words)
• Grade effect (F=21.42, p < .001, .87)
*1 < 2, 3, 4 *2 < 3, 4
• No genre effect
– Complexity (clauses per c-unit)
• Grade effect: (F=3.69, p < .014, .08)
*1 < 3, 4
• No genre effect
Results: Vowel Prevalence (M)
Grade (M
total
words)
1 (49)
(49)
2 (64)
(68)
3 (113)
(103)
4 (104)
(109)
Short
Vowel
Long Vowel Diphthong
RControlled
Narrative
25
20
0
3
Expository
26
20
2
2
Narrative
37
25
1
2
Expository
37
26
3
6
Narrative
63
45
3
12
Expository
61
44
6
10
Narrative
61
44
2
11
Expository
61
43
8
12
Does grade and/or genre affect
prevalence of vowel spellings?
Vowel Type
Grade
Finding
Short Vowels
1 < 2, 3, 4
2 < 3, 4
F=23.82 p<.001; .36
Long Vowels
1 < 2, 3, 4
2 < 3, 4
F=22.98 p<.001; .35
Diphthongs
NON-SIGNIFICANT
WITH
1 < 3, LENGTH
4
F= 17.81 p<.001; .30
SET AS A COVARIATE
2 < 3, 4
R-controlled
1 < 3, 4
2 < 3, 4
F=26.9 p<.001; .39
Length effect: 1 < 2, 3, 4;
2 < 3, 4
Does grade and/or genre affect
prevalence of vowel spellings?
• Genre Effect
– Diphthongs: Less prevalent in narrative than
expository (F=63.16 p<.001; .33)
– Disappears with length (# of words) entered as
covariate
Does grade and/or genre affect
accuracy of vowel spellings?
Vowel Type
Grade
Finding
Short Vowels
1 < 2, 3, 4
F=17.32 p<.001; .29
Long Vowels
1 < 2, 3, 4
F=6.77 p<.001; .14
Diphthongs
1 < 3, 4
2<3
F=4.86 p=.004; .16
R-controlled
(grade x genre effect: difference
between genres greatest for 4th
graders )
NS
Complexity effect: 1 < 3, 4
Does grade and/or genre affect
accuracy of vowel spellings?
• Genre Effect
– Diphthongs
• 1 and 2 more accurate in expository
• 3 and 4 more accurate in narrative
– (ANCOVA)Long Vowels
• More long vowels accurate in narrative
Vowel Error Types
• Phonologic: vowel phoneme not represented by a grapheme (e.g., ct for
cat)
• Orthographic: vowel phoneme represented with “illegal” grapheme (e.g.,
tip for chip)
• MGR (mental graphemic representation): vowel phoneme represented
with a legal, but incorrect grapheme (e.g., keap for keep)
• Homophone: vowel phoneme represented by a grapheme, but result is a
homophone of the word (e.g., one/won)
• Reversal: vowel phoneme represented, but order disrupted by reversal
between vowel and adjacent consonant (e.g., battel/battle)
• Examined percentage of children demonstrating at least one error in the
above categories
Vowel Error Types: Short Vowels
Narrative
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
17%
69%
55%
3%
7%
2
22%
50%
29%
3%
3%
3
18%
64%
31%
3%
10%
4
22%
53%
25%
6%
0%
1 > 2,4
1 > 2,3, 4
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
3%
62%
24%
0%
0%
2
12%
50%
3%
0%
0%
3
13%
49%
0%
0%
0%
4
12%
44%
6%
0%
6&
Expository
Vowel Error Types: Long Vowels
Narrative
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
14%
45%
31%
0%
0%
2
25%
28%
9%
16%
0%
3
15%
33%
38%
8%
0%
4
31%
56%
16%
12%
1%
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
7%
65%
17%
7%
7%
2
0%
47%
16%
12%
6%
3
5%
56%
31%
29%
5%
4
4%
50%
6%
22%
4%
Expository
Vowel Error Types: Diphthongs
Narrative
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
0%
0%
0%
0%
0%
2
0%
0%
0%
0%
0%
3
0%
5%
3%
3%
3%
4
0%
0%
3%
1%
0%
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
0%
14%
0%
0%
3%
2
0%
6%
3%
3%
3%
3
0%
8%
3%
3%
3%
4
0%
16%
0%
0%
3%
Expository
Vowel Error Types: r-controlled
Narrative
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
0%
7%
10%
0%
0%
2
0%
28%
12%
3%
3%
3
10%
20%
38%
8%
10%
4
12%
19%
12%
3%
10%
Phonological
Orthographic
(illegal)
MGR (legal)
Homonym
Reversal
1
3%
14%
0%
3%
7%
2
0%
22%
0%
0%
3%
3
0%
15%
5%
0%
3%
4
0%
34%
12%
9%
9%
Expository
Major Findings and Discussion
• Grade Effects on Vowel Spellings
– Prevalence
• For the most part, 1st graders wrote less vowels than
other three grades; 2nd graders wrote less than 3rd and
4th
• Findings mirrored and seemed to be due to length
effects
• Suggests prevalence may not be a useful measure for
developmental information
Major Findings and Discussion
• Grade Effects on Vowel Spellings
– Accuracy
• For the most part, 2nd through 4th graders were similar
in accuracy; 1st graders were less accurate
• Findings
– Do not mirror complexity effects
» “Trade-offs” to handle complexity do not appear to
affect vowel spellings
» 2nd graders may have reached a level of maturity
that does not change across primary grades
– Sampling issue: 4th graders “dipped” in overall literacy
skills
Major Findings and Discussion
• Grade Effects on Vowel Spellings
– Accuracy
• Findings support Read’s (1971, 1975) findings (for 3-6
year olds) that most errors occurred on short/long
vowels
• It appears difficulties with short/long vowels may be
due to varied orthographic representations of these
vowel categories vs. others (e.g., diphthongs) (Jorm,
1977; Treiman, 1993)
• Results support findings of Caravolas et al. (2005)
suggesting vowel development/accuracy relies heavily
on MGRs
Major Findings and Discussion
• Grade Effects on Vowel Spellings
– Error Types
• Minimum grade effects
– More grade effects for short and long vowels than other
categories
» May be due to prevalence and/or greater range in
possible orthographic representations (Jorm, 1977;
Treiman, 1993)
• Most errors were illegal or legal orthographic errors
– Phonemic awareness difficulties not impacting vowel spelling
greatly at these grade levels
Major Findings and Discussion
• Grade Effects on Vowel Spellings
– Error Types
• Similar distribution of error types across grades suggest
children can rely on, or use/misuse flexibly, the same
linguistic resources when spelling
– Consistent with the overlapping waves theory of spelling
development (e.g., Rittle-Johnson & Siegler, 1999)
» Spelling is a complex, non-linear developmental
process
» Spelling development is variable and children use
less and more “sophisticated” strategies across
development depending on the task
Major Findings and Discussion
• Genre Effects on Vowel Spellings
– Minimal difference across grade
• Prevalence: diphthongs
• Accuracy: long vowels, diphthongs
– More effects for error type
• Less errors in expository than narrative
– Short vowels – phonological and MGR (legal) errors
– Long vowel – phonological errors
• Less errors in narrative than expository
– Long vowels – orthographic (illegal) errors
– Diphthongs - orthographic (illegal) errors
Major Findings and Discussion
• Genre Effects on Vowel Spellings
– Minimal effect of genre may be due to
• Lack of differentiation in genre production
– However, construct validity was 82%
– Future investigations could examine optimal means for
eliciting the intended genre
• Vowel spelling is immune to reported genre effects
(e.g., Singer & Bashir, 1999)
– Future investigations could examine consistency in vowel
spellings across genres by word
Other Findings
• Children’s “choice” in words did not affect accuracy
– In the future, examine whether similar findings with other
sampling methods (e.g., dictation)
• Results reflect children with typical skills
– Future investigations should examine other populations
(e.g., children with LD/LLD)
• Results are aggregated across vowels within
categories
– Researchers could examine accuracy and error types by
specific vowels (e.g., long vowel a, short vowel e)
Summary
• Investigation is first to study vowel spellings
across grades and genres
• Vowel development information went beyond
accuracy and included prevalence and error
analyses
• Findings suggest that grade level affects vowel
spelling accuracy; genre mostly has an effect
on the type of vowel spelling error
Acknowledgment:
We thank Dr. Cynthia Puranik for
providing the writing samples for
this investigation