xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 005 20070604134445.0
007 cr mnu---uuuuu
008 070604s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001451
Fawcett, Kelly M.
Spelling development in young school age children
h [electronic resource] /
by Kelly M. Fawcett.
[Tampa, Fla] :
b University of South Florida,
ABSTRACT: Previous research investigations in the area of spelling development have adopted two approaches, the broad approach and the narrow approach. The broad approach suggests that spelling develops in sequential stages whereas the narrow approach focuses on individual linguistic patterns. However, research findings have revealed that children's spellings do not exhibit errors pertaining to specifically one stage or reflecting one linguistic element, yet a research void exists in resolving how these two approaches might intermix. This study examined the spelling errors of typically developing children in first through fourth grades (N = 400) to determine the quantitative and qualitative differences in misspellings among grade levels. Each grade level had an equal representation of children (N = 100) and male and female participants. The spelling errors were extracted from two writing samples completed by the children, a narrative and expository sample. In an attempt to combine the broad and narrow approaches, a coding system was designed to evaluate the linguistic category (phonological, orthographic, morphological) and specific features (letter name spelling, vowel error, digraph, etc.) of the spelling errors. The findings revealed a significant interaction between grade level and error type for phonologically-based spelling errors (1st graders made more errors than 2nd and 4th graders) and a greater number of morphological errors was noted in 4th vs. 2nd grade. No significant effects were noted for writing genre or gender. Analysis of performance patterns for specific linguistic category errors within and across grade levels revealed that all four grade levels committed the most phonological errors in the PSE (phonological silent /e/) and PSON (phonological sonorant clusters) categories. The OLN (orthographic letter name) and ODI (orthographic digraph) errors also occurred frequently in all four grades with first graders demonstrating significantly^more occurrences of the OLN than ODI error. Morphological findings revealed that first graders made significantly more MINF (morphological inflection) than MHOM (morphological homonym) errors and all four grades had significantly more MINF than MCON (morphological contraction) errors. A qualitative analysis regarding the most frequently misspelled words and most frequently encountered codes was also performed. The clinical and educational implications of these findings are discussed.
Thesis (M.A.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 96 pages.
Adviser: Ruth Huntley Bahr, Ph.D.
Qualitative scoring systems.
x Speech Language Pathology
t USF Electronic Theses and Dissertations.
Spelling Development in Young School Age Children by Kelly M. Fawcett A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science Department of Communicati on Sciences and Disorders College of Arts and Sciences University of South Florida Co-Major Professor: Ruth Huntley Bahr, Ph.D. Co-Major Professor: Elaine R. Silliman, Ph.D. Stefan A. Frisch, Ph.D. Date of Approval: March 31, 2006 Keywords: phonology, orthography, morphology, PO MAS, qualitative scoring systems Copyright 2006, Kelly M. Fawcett
i Table of Contents List of Tables................................................................................................................. ....iii List of Figures................................................................................................................ .....v Abstract....................................................................................................................... .......vi Chapter 1 : Literature Review.............................................................................................1 Theories of Spelling Development...............................................................................2 The Narrow Approach............................................................................................6 Spelling Errors as a Window into Linguistic Knowledge............................................7 The Emergence of Linguistic Knowledge in Spellings..........................................7 Phonologically-based misspellings...................................................................8 Orthographic misspellings................................................................................9 Morphological misspellings............................................................................10 Spelling Variations in Differing Groups of Children...........................................13 Typically developing children vs. children with language learning disabilities.......................................................................................................13 Assessments of Spelling.......................................................................................17 Spelling inventories........................................................................................17 Pretest-study-posttest......................................................................................18 Norm-referenced assessments.........................................................................19 The Effects of Composing on Spellings...............................................................20 Scoring Systems....................................................................................................21 Constrained vs. unconstrained approaches.....................................................22 Visual accuracy approach...............................................................................22 Orthographic legality approach.......................................................................23 Statement of the Problem............................................................................................25 Chapter 2 : Method...........................................................................................................29 Participants..................................................................................................................2 9 1st to 3rd grade......................................................................................................29 4th grade................................................................................................................30 Materials.....................................................................................................................3 1 Procedures...................................................................................................................31 POMAS.................................................................................................................32 Data Reduction......................................................................................................35 Qualitative Analyses.............................................................................................36 Agreement.............................................................................................................37
ii Statistical Analysis................................................................................................37 Chapter 3 : Results............................................................................................................ 39 Inter-Examiner Agreement.........................................................................................40 Overview of Subject Performance..............................................................................41 Statistical Analyses of Data for the Research Questions............................................44 Question 1: Grade Level Effects on Number and Types of Errors.......................44 Question 2: Did The Writing Sample or Gender Affect Number And Type of Spelling Error?..................................................................................................45 Question 3: Error Patterns within Grade Level.....................................................46 Phonological errors.........................................................................................46 Across grade patterns................................................................................47 Within grade patterns................................................................................48 Orthographic errors.........................................................................................48 Across grade patterns................................................................................48 Within grade patterns................................................................................49 Morphological Errors......................................................................................50 Across grade patterns................................................................................50 Within grade patterns................................................................................51 Questions 4: Qualitative Analysis of E rror Patterns Within Grade Level............53 Chapter 4 : Discussion......................................................................................................58 Question 1: Grade Level Effects on Number and Types of Errors.............................59 Question 2: Genre and Gender Effects on the Number and Type of Spelling Errors......................................................................................................................... ..62 Question 3: Error Patterns within Grade Level...........................................................63 Question 4: Qualitative analysis of Features within Phonological, Orthographic, and Morphological categories.............................................................66 Study Strengths and Limitations.................................................................................69 Strengths...............................................................................................................69 Limitations............................................................................................................70 Clinical and Educational Implications........................................................................71 Clinical Implications.............................................................................................71 Educational Implication........................................................................................72 References..................................................................................................................... ....75 Appendices..................................................................................................................... ...82 Appendix A Â– Coding System....................................................................................83 Appendix B Â– t-tests for Grade Level Eff ects on Number and Types of Errors.........87 Appendix C Â– Post Hoc Comparisons for Error Patterns across Grade Level............89 Appendix D Â– Comparisons of Specific Error Types Across Grade Level................91
iii List of Tables Table 1.1. Broad Approach Stage Theories.................................................................3 Table 2.1. Ethnic Representation of Pa rticipants in Grades 1 to 4............................30 Table 2.2. Grade Performance by Writing Sample....................................................32 Table 2.3. POMAS Coding Syst em Â– Phonological Errors.......................................33 Table 2.4. POMAS Coding System Â– Orthographic Errors.......................................34 Table 2.5. POMAS Coding System Â– Morphological Errors....................................34 Table 2.6. POMAS Coding System Â– Mixed Phonological-Orthographic Errors.........................................................................................................35 Table 2.7. POMAS Coding System Â– Mixed Morphological-Orthographic Errors.........................................................................................................35 Table 2.8. POMAS Coding System Â– Mixed Morphological-Phonological Errors.........................................................................................................35 Table 3.1. Total Number of Errors for Each Grade Level (N= 3,264)......................42 Table 3.2. Total Number of Errors in the Two Writing Samples According to Grade Level...............................................................................................43 Table 3.3. Total Number of Errors Acco rding to Grade Level Based on Sex...........43 Table 3.4. Most Commonly Misspelled Words in The Spelling Samples.................53 Table 3.5. Most Commonly Used Error Codes for All Grade Levels.......................56 Table A.1 Feature Errors............................................................................................83 Table B.1 Phonology Comparisons Across Grade Levels.........................................87 Table B.2 Orthographic Compar isons Across Grade Levels.....................................87 Table B.3 Morphological Compar isons Across Grade Levels..................................88
iv Table C.1 Bonferroni Post Hoc Testi ng Results for Phonological Error Type Comparisons per Grade Level..................................................................89 Table C.2 Bonferroni Post Hoc Testi ng Results for Orthographic Error Type Comparisons per Grade Level..................................................................90 Table D.1 Comparison of Error Type PFLP Between Grade Levels........................91 Table D.2 Comparison of Error Type PFPV Between Grade Levels........................91 Table D.3 Comparison of Error Type PSE Between Grade Levels...........................92 Table D.4 Comparison of Error Type PSON Between Grade Levels.......................92 Table D.5 Comparison of Error Type ODI Between Grade Levels..........................93 Table D.6 Comparison of Error Type OLN Between Grade Levels.........................93 Table D.7 Comparison of Error Type OLR Between Grade Levels..........................94 Table D.8 Comparison of Error Type OLS Between Grade Levels..........................94 Table D.9 Comparison of Error Type OVr Between Grade Levels..........................95 Table D.10 Comparison of Error Type MCON Between Grade Levels......................95 Table D.11 Comparison of Error Type MHOM Between Grade Levels.....................96 Table D.12 Comparison of Error Type MINF Between Grade Levels........................96
v List of Figures Figure 3.1. Comparison of Category Error Ratios by Grade Level............................45 Figure 3.2. Decreasing P honological Error Feature Use by Grade Level...................47 Figure 3.3. Decreasing Orthographic Feature Use by Grade Level............................49 Figure 3.4. Morphological Error Fe ature Use by Grade Level...................................51
vi Spelling Development in Young School Age Children Kelly M. Fawcett Abstract ABSTRACT Previous research investiga tions in the area of spelli ng development have adopted two approaches, the broad approach and th e narrow approach. The broad approach suggests that spelling develops in sequential stages whereas the narrow approach focuses on individual linguistic patterns. However, rese arch findings have reve aled that childrenÂ’s spellings do not exhibit errors pertaining to specifically one stage or reflecting one linguistic element, yet a research void exis ts in resolving how these two approaches might intermix. This study examined the spelling errors of typically developing children in first through fourth grades (N = 400) to determine the quantitative and qualitative differences in misspellings among grade levels. Each gr ade level had an equal representation of children (N = 100) and male and female partic ipants. The spelling errors were extracted from two writing samples completed by the chil dren, a narrative and expository sample. In an attempt to combine the broad a nd narrow approaches, a coding system was designed to evaluate the lingui stic category (phonological, orthographic, morphological) and specific features (letter name spelling, vowel error, di graph, etc.) of the spelling errors. The findings revealed a significant interac tion between grade level and error type for phonologically-based spelling errors (1st graders made more errors than 2nd and 4th
vii graders) and a greater number of morphological errors was noted in 4th vs. 2nd grade. No significant effects were noted for writing ge nre or gender. Analysis of performance patterns for specific linguistic category errors within and across grade levels revealed that all four grade levels committed the most phonological errors in the PSE (phonological Â– silent /e/) and PSON (phonological Â– s onorant clusters) categories. The OLN (orthographic Â– letter name) and ODI (orthogr aphic Â– digraph) errors also occurred frequently in all four grades with firs t graders demonstrati ng significantly more occurrences of the OLN than ODI error. Mo rphological findings revealed that first graders made significantly more MINF (morphological Â– inflection) than MHOM (morphological Â– homonym) errors and all four grades had significantly more MINF than MCON (morphological Â– contractio n) errors. A qualitative an alysis regarding the most frequently misspelled words and most freque ntly encountered codes was also performed. The clinical and educational implica tions of these findings are discussed.
1 Chapter 1: Literature Review Spelling instruction is incr easingly important in educat ion today (Graham, Harris, & Chorzempa, 2002). In previous years, spel ling instruction in th e classroom did not emphasize connections to reading and wr iting (Apel & Masterson, 2001; Goswami, 1992). This oversight has led to a decreased awareness that E nglish spelling is a patterned system. In general, spelling has been taught through rote teaching and memorization of a weekly spelling list, with little stress on th e importance of teaching patterns (Apel, Masterson, & Hart, 2004a; Goswami, 1992). Ho wever, a recent shift towards improving spelling assessment and instruction highlights the importance of spelling as the study of word patterns (Bear, Invernizzi Templeton, & Johnston, 2004). Current research focuses on improvi ng spelling instruction through the understanding of how spelling develops (B ear et al., 2004; Bern inger et al., 1998; Masterson & Crede, 1999). Comparisons have been made across groups of children to quantify errors (Bruck & Waters, 1988). Howeve r, minimal research exists regarding the qualitative assessment of spelling, which is pe rtinent for providing information regarding the types of linguistic errors children make. In turn, mo re specificity on individual linguistic patterns would enhance indivi dualized instructi on and intervention. This literature review discusses the re search pertaining to spelling development and assessment. The discussion begins with an overview of the theories of development that describe the errors th at are common throughout spelling development. The second
2 section discusses patterns of typical de velopment in the emergence of linguistic knowledge that supports spelling followed by a comparison of spelling errors in children who are typically developing versus those w ith a reading disability and those with a language learning disability. The third secti on presents various spelling assessments and quantitative and qualitative scoring systems used to assess spelling skills. The increased need for qualitative as opposed to quantitative assessments is then discussed. A brief comparison of written genres is made in the fourth section to increase awareness of the effects genres have on spelling. Finally, the st atement of the problem presents the studyÂ’s purpose and research questions. Theories of Spelling Development The broad approach and the narrow appr oach are two frameworks for describing how spelling develops. These approaches represent different perspectives that focus either on general stages of spelling deve lopment (the broad approach) versus the linguistic development of indivi dual spelling features (the na rrow approach). However, it is important to mention that both frameworks aim to achieve the same goal, providing a description of spelling development. The Broad Approach. The broad approach captures developmen tal patterns that signal changes in performance. This approach, qual itative in nature, subscribes to the concept of stages in spelling development (Bear et al., 2004; Reece & Treiman, 2001). Stage theory places patterns of development in various time fram es. While many researchers have developed their own stage theories, th ree of the most well known are those proposed by Gentry (1982), Henderson (1985) and Ehri (1986).
3 GentryÂ’s theory consists of five stages (precommunicative, semiphonetic, phonetic, transitional, and correct spelling). In contrast, Ehri (1986) proposed three stages (semiphonetic, phonetic, and morphemic). Hende rson (1985), like Gentry, also had five stages of development, but the stages diffe red in that Henderson be lieved in a life-long approach to spelling development, whereas Gentry proposed that complex spelling development could be completed during early academic instruction (Gentry, 2004; Treiman & Cassar, 1997). A complete descri ption of the primary stage theories is provided in Table 1.1, followed by a co mparison of the three theories. Table 1.1. Broad Approach Stage Theories. Gentry (1982) Henderson (1985) Ehri (1986) Stage 1 Precommunicative Strings together random letters; no concrete knowledge of the sounds the letters represent. Preliterate Meaningless marks on paper with a crayon or pencil; no understanding that writing represents speech. Semiphonetic Uses letters with no knowledge of the sounds that match. Stage 2 Semiphonetic Attempts to spell using the letters that match the sounds in the word; vowels and consonants in words Letter-name spelling Understanding that each sound represents a letter and letter names are Phonetic Demonstrates partial awareness of sounds and letters that match.
4 Table 1.1 (Continued) Gentry (1982) Henderson (1985) Ehri (1986) are usually represented as one letter (R=ARE). used to spell words. Stage 3 Phonetic All sounds represented but no orthographic rules applied. Within-word pattern Spelling has been learned from exposure to print during reading. Knowledge of sight words assists in spelling unfamiliar words. Morphemic Orthographic and morphological awareness skills are applied. Stage 4 Transitional No longer relies on sound to spell words; applies orthographic and morphological information to spellings. Syllable juncture Spelling rules are applied, such as doubling of consonants, to mark short vowels in words. n/a
5 Table 1.1 (Continued) Gentry (1982) Henderson (1985) Ehri (1986) Stage 5 Correct Spelling Concrete understanding of fundamentals of spelling and spellings are more likely to be correct. Derivational Principles Understanding of root word and the meaning it carries; develops throughout life. n/a Table 1.1 shows three different variations of stage theory. Ehri (1986) and Gentry (1982) share a similar view that Stage 1 cons ists of strings of le tters carrying no real meaning. Henderson (1985), on the other ha nd, includes random doodli ng with a writing utensil in Stage 1. Representations for Stage 2 are similar across all three researchers, revealing early knowledge of letters and the sounds repr esenting each letter. Stage 3 shows a greater amount of variation in that Ehri considers childre n in this stage to demonstrate advanced morphological skills while the Gentry and Henderson stages include only phonetic and or thographic skills. EhriÂ’s (1986) Stage 3 is more developmentally advanced than those of He nderson and Gentry. Ehri argued that the development of orthographic and morphological skills was the final st age where children learned word regularities during morphologi cal development, which then led to conventional spelling. Conventional spelling skills are thought to continue throughout life, and therefore, are not cl assified into a specific st age (Treiman & Cassar, 1997).
6 In contrast to Ehri, Henderson and Ge ntryÂ’s Stage 3 consisted of beginning spelling skills. Stages 4 and 5 were similar in that the child was learning and applying more advanced spelling rules. However, Henderson (1985) suggested that individuals would not completely master these skills becau se vocabulary continued to build and word roots, origins, and meanings continued to de velop. Gentry (1982), in contrast, believed that spelling skills become automatic because the child no longer relies on sound to spell, but is able to apply orthogr aphic and morphological inform ation to spell (Treiman & Cassar, 1997). The Narrow Approach While stage theory may seem to be an appropriate spelling framework, recent research questions the presumption that sp elling develops in specific stages (Reece & Treiman, 2001). Instead, aspects of phonol ogic, orthographic, and morphologic knowledge, in addition to mental graphemi c representations (Apel et al., 2004a), simultaneously interact during all levels of spelling development. Relative to this idea, Reece and Treiman (2001) presented evidence that first grade children were using phonologic and orthographic knowledge to sp ell. Thus, Reece and Treiman (2001) argued against stage theory in that multiple aspects of linguisti c knowledge interacted simultaneously within and across children to yield increasingly conventional spellings. Phonologic aspects, therefore, do not act in dependently of the other components. Consistent with the notions of the na rrow approach, Sulzby (1996) proposed the idea of repertoire theory in which spelling developed base d on an interaction of many different linguistic aspects. In other word s, all of the phonologic, orthographic, and morphologic skills a child has at any given time may interact to assist in spelling a word.
7 Thus, it is suggested that older childre n must access these linguistic components simultaneously to meet the demands of sp elling complex words (Apel, Masterson, & Niessen, 2004b). Misspellings occur because the linguistic complexity of the word exceeds the childÂ’s ability to utilize one or more linguistic components. In contrast to the broad approach, which classifies spelling development according to stages, the narrow approach an alyzes individual linguistic features and attempts to determine how these featur es affect childrenÂ’s misspellings (Reece & Treiman, 2001; Silliman, Bahr, & Peters, 2006). This approach typically limits spelling analysis to one feature at a time. While the broad and narrow a pproaches provide two different perspectives regarding spelling develo pment, both provide ways to classify error patterns in children and afford opportunities for valuable information to be gathered for instructional and intervention purposes. Spelling Errors as a Window into Linguistic Knowledge Spelling development frameworks, such as those found in the broad and narrow approaches, provide a way to examine childre nÂ’s emerging abilities to spell. While every child will not meet milestones at the same poi nt in time, similarities will be found across children. The Emergence of Linguisti c Knowledge in Spellings As described by Dodd and Carr (2003) children initially demonstrate phonological spelling errors in letter-to-sound a ssociations. These erro rs will appear as random strings of letters that carry no meaning (Dodd & Carr, 2003). However, upon entering kindergarten, most children have knowledge of letter names from routine activities, such as si nging the alphabet. They will then use this knowledge to assist in
8 spelling unfamiliar words, thus reducing the occurrence of random letter strings (Bourassa & Treiman, 2001). Phonologically-based misspellings As children learn to spell using increased alphabetic knowledge, error pa tterns occur that include letter name spellings and phonological violations. According to Bour assa and Treiman (2001), letter-name spellings substitute for vowel spellings and sequences of phonemes and occur most often in kindergarten and first grade children. The most frequent le tter name misspellings occur with the liquid phonemes / r / and / l / (Bourassa & Treiman, 2001) For example, early spellers may demonstrate errors, such as spelling eat as et elephant as lefit or far as fr In this case, the child has not developed an understanding of phoneme sequences, and therefore, spells the sequence with the si ngle letter name. Letter name spellings reduce over time as the result of increased exposure to print and formal instruction (Bourassa & Treiman, 2001). According to Bourassa and Treiman (2001) letter name spellings occur due to inexperience with the phonological structur e of the language and less print exposure. However, English pronunciation makes it difficult, at times, to decipher the phonemes in a word, thus leading to other types of spel ling errors, such as misspellings containing flaps (Treiman, Cassar, & Zukowski, 1994). A flap is a phoneme represented in the International Phonetic Alphabet (IPA) to phone tically represent the combination of /t/ and /d/ (Small, 1999). In the case of flaps, children of ten spell words such as city and dirty as cidy and dirdy due to the voiced nature of th e flapÂ’s pronunciation (Treiman et al., 1994). This error more likely occurs wh en children are capable of segmenting the word phonemically, with the outcome that the /t/ sounds like /d/.
9 In addition to the previously described erro rs with letter name spellings and flaps, early spelling errors also reveal difficulties w ith consonant clusters in the initial, medial, and final positions of words (Treiman, 1991). This difficulty arises from childrenÂ’s inexperience with dividing the cluster in to separate phonemes (Treiman, 1991). In other words, consonant cluster errors occur because the clusters are being analyzed as a one phoneme unit rather than as a single unit with two phonemes. For example, the word play, spelled phonetically as /ple/, has one unit containing two phonemes, /p/ and /l/, and a second unit containing /e/. However, the initial phoneme /p/ of the first unit followed by the phoneme /e/ of the second unit makes it challe nging for young children to understand that the /l/ needs to be represented separately from the /p/ In other words, writing /p/ for the first unit does not represent the /pl/ in play, as children of ten portray the spelling. Although the phonological process of cluster reduction in the initial position of words is most common, errors in nasal cluste rs, such as /nd/, are also prominent in beginning spellers (Treiman, 1991). This accounts for why young children will spell and as ad. These problems with nasal clusters also oc cur because nasal phonemes are difficult to hear when the child is decoding the word (Treiman, 1991). Orthographic misspellings. While phonological errors are prominent in beginning spellers, orthographic errors are also eviden t early in development (Cassar & Treiman, 1997). The presence of orthographic violati ons supports the idea that spelling does not develop in specific stages but build s on multiple linguistic factors. Orthographic errors include problems w ith consonant doubling and marking long vowel patterns through the use of silent Â–e. Beginning spellers make errors in consonant doubling when they understand a word c ontains a double consonant but do not
10 understand where the double consonant occurs. In written English, double consonants can occur in the middle of a word after a short vowel or at the end of a word. Double consonants do not occur in the beginning. For example, the word press has a doubled consonant in the final position. However, ch ildren might mark th e doubled consonant in the wrong position, such as ppres for press This type of error i ndicates an awareness of the need for a doubled consonant, but also illu strates the lack of integration of phonologic with orthographic knowledge to result in a correct spelling (Ca ssar & Treiman, 1997). Orthographic errors involvi ng the omission of silent Â–e also occur in young spellers, who lack the knowledge that when adding a silent Â–e to the end of a word, the preceding vowel is pronounced as a l ong vowel. For example, misspelling trade as trad indicates absence of the orthograp hic understanding of the silent Â–e Morphological misspellings. Morphological development consists of both inflectional and derived forms. Inflectional morphology involves maintaining the original root of a word but changing agreement, number, or possession with a grammatical marker, such as past tense Â– ed present progressive Â– ing or plural Â– s. Nunes, Bryant, & Bindman (1997), studied children in grades 2, 3, and 4 (N=363) to determine how they developed inflecti onal morphology skills. This study included a spelling task that incorporated regular past verbs, irregular past verbs, and nonverbs. Based on childrenÂ’s performance, the aut hors proposed that young children utilized inflectional markers, but failed to understand their meaning. In other words, the children represented the inflectional marker by spel ling the word phonetically, resulting in the word called being spelled as calld
11 Nunes et al. (1997) then proposed that, as children began to u nderstand and utilize inflectional markers, such as Â– ed they overgeneralized and applied Â– ed to words ending in /d/, including irregular past tense verbs, such as found, or nonverbs, such as cold Finally, children are observed to understand the meaning th at the inflectional marker represents and spell the wo rds correctly. The development of inflected morphology, as presented by Nunes et al. (1997), coincides wi th stage theory of development in that children spell the word phonetically without understandi ng its meaning prior to utilizing the morphological form. Stage theory suggests that knowledge of inflections is later developing. According to Bourassa, Treiman, and Kessler (in press), children actually utilize inflectional markers early in spelling development. In fact, Boura ssa et al. (in press) suggests that children utilize inflectional markers to help solve problems occurr ing as a result of phonological limitations. For example, if a child understands that wait ends in /t/, this information will assist him/her in correctly spelling waiting (an inflected form) sin ce the flapped /t/ makes the word more difficult to spell phonetically. To demonstrate how young children utili ze morphological knowledge, Bourassa et al. (in press) compared children who were dyslexic and typically developing to determine if both groups utilized inflectiona l morphology in the same ways. The typically developing children (N=25) ranged from grades 1 to 3 while the chrono logical ages of the children with dyslexia (N=25) ranged in age from 9;2 to 14;7 years. This group also scored below a grade 4 level on a standardized spelling meas ure. Results indicated that both groups performed similarly. Both more accurately spelled complex words that included an inflectional marker, such as rained than simple words, such as brand in
12 which they omitted at least one letter of the final nasal cluster. However, neither group utilized their knowledge of the wordÂ’s root. For example, both groups misspelled the root word lace as lase but spelled the inflected form correctly as laced Thus, as children continue to develop morphological understa nding, they could use their knowledge of inflections, such as the spelling of laced to correct the spelling of the simpler word lace In contrast to inflectional morphol ogy, derivational morphology alters the meaning of a word, which can include ch anging it from a verb to a noun among other changes (Carlisle, 2003). The general c onsensus is that derived morphological representations require a longe r period of time for their c onventional spellings to be mastered (Carlisle, 1987, 1988; Green et al ., 2003; Nagy, Berninger, Abbott, Vaughn, & Vermeulen, 2003), however, it has been found that derivational morphology does develop concurrently with inflected morphology. In other words, children do not wait until inflected forms are mastered before utilizi ng derived forms (Carlisle, 2003). For example, consider the suffix Â– able. Children as young as the preschool years have been observed to use this suffix, as in the word flyable. Although the added suffix is an overgeneralization, this is the first step in un derstanding and utilizing derive d meanings (Carlisle, 2003). A clearer and more consistent use of derived forms in writ ing appears to occur sometime between first and fourth grades (Carlisle, 1996). To reflect on how derivational morphology develops in spelling, a study by Green et al. (2003) should be considered. The purpose of this study, which included 3rd and 4th graders (N= 247), was to observe the use of inflec tional markers in their writing. Results revealed that inflected morphology was more accurately used than derived forms. Based on these results, Green et al. (2003) suggest ed several explanati ons for derivational
13 development. One of the most frequent obs erved derived forms was the addition of Â– ly The early development of these derived forms most likely occurs because children use these forms early in speech development. Transparent derived forms, such as dance dancer, also emerge initially because the base word is present in the derived word (Green et al., 2003). Errors would thus be more evident with opaque forms, such as magic magician Spelling errors at the phonologic, orthogr aphic, and morphologic levels, such as the ones just described, are a natural part of sp elling development. As children progress academically, it would be expected that erro rs would shift from primarily phonological to primarily orthographic and mor phological errors, especially morphological errors related to derivational meanings (Bear et al., 2004; Ehri, 1986; Gentry, 1982; Henderson, 1985). However, it cannot be expected that a ll children will develop in the same way, demonstrating identical errors at the sa me points in their spelling development. Therefore, variations in e xperience with academic language through reading and writing, as well as the quality of spel ling instruction, may create va riations in childrenÂ’s error types. Spelling Variations in Di ffering Groups of Children Typically developing children vs. childr en with language le arning disabilities. While spelling errors of typically developing children will vary, it is also important to consider how spellings of typi cally developing children differ from those of children with language learning disabilities ( LLD). One speculation is that error patterns between these two groups of children will differ with respect to the utilization of phonologic, orthographic, and morphologic knowledge.
14 Phonology, orthography, and morphology appe ar to develop in different phases with derivational morphology considered to be the most complex and, therefore, a later developing skill (Nunes et al., 1997). Since the broad approach (Bear et al., 2004) regards spelling as developing in phases, it would be appropriate and typical to see spelling errors occurring in advancing phases. However, chil dren with language le arning disabilities or other spelling delays may not demonstrate th ese patterns. Because these children struggle with the various kinds of linguistic k nowledge necessary fo r typical spelling development, it is likely that they may di splay many of the same errors evident in younger spellers (Treiman & Bourassa, 2000a). Most of the studies on spelling in ch ildren with atypical development have focused on struggling readers, especially ch ildren with reading di sability (RD) or dyslexia (Hauerwas & Walker, 2003). Some of these studies yielded results to support the hypothesis that struggling read ers performed similarly to young spellers. For example, Hauerwas and Walker (2003), studied 11-13 ye ar old children (N= 26) with spelling and reading problems (indicated by a stan dard score of less than 85 on the Wide Range Achievement Test 3 (WRAT-3; (Jastak & Wilkinson, 1984). They were compared to normally developing children of the same age (N= 31; as indicated by a standard score of 90 or above on the WRAT-3), and normally achieving second and third graders (N= 31), also selected based on a standard score of 90 or above on the WRAT-3. This study was designed to determine whether phonological defic its contributed to de ficits in inflected morphology (e.g., spelling inflected verbs, such as skip for skipped ) or whether limitations in orthographic and morphologic awareness were the primary contributing factors. All participants were given a phonological awaren ess task, which required the
15 deletion of syllables or phonemes, a morphologi cal awareness task in which a target cloze format was completed by adding the appr opriate inflected morphemes, and an orthographic awareness task where non-word s presented in pairs were identified. In addition, three spelling tasks were complete d. These tasks included spelling inflected verbs in a sentence context, spelling inflected verbs in a list format, and spelling base words from the inflected forms (e.g., jump for jumped ). Results indicated that the preadolescent s with reading and spelling difficulties (described as specific language impairments) showed particular difficulty with inflected morphology, which is mastered earlier than derivational morphology. For example, in comparison to the younger participants, th e preadolescents with reading problems misspelled inflected verbs in sentences by frequently omitting the past tense Â–ed form, as in jumped and waved, which were spelled as jump and wave Although the participants demonstrated errors across the spelling task s, it was in the inflected morphology where the most errors were evident (Hauerwas & Walker, 2003). Hauerwas and Walker (2003) analyzed spel ling errors quantitatively, but they did not analyze errors individuall y. In contrast, Silliman et al. (2006) included a qualitative system for the analysis of spelling e rrors. The Phonological, Orthographic, and Morphological Analysis of Spelling (POMAS ) was developed to evaluate whether spelling errors differed when comparing thr ee different groups of children, ages 6 to 11 years: a group with language learning disa bilities (LLD) (N = 8) a chronological age matched (CA) group (N = 8), and a spelli ng age matched (SA) group (N = 8). The purpose of the study was to assess quantitative and qualitative performance differences. Quantitative scoring systems included constrained and unconstrained systems and
16 orthographic legality. The qualitative scori ng system (POMAS) incorporated specific error codes based on linguistic category and feature type. Based on the application of different scor ing systems, results of the quantitative analyses indicated a significant difference fo r three of the four cat egories assessed with the CA group found to have significantly di fferent performance from the LLD and SA groups. Similar performance was found for th e SA and the LLD children, which was described as a result of delayed develo pment of interactions between phonologic and orthographic knowledge for the LLD group (Silliman et al., 2006). Of interest for the current study were th e qualitative differences in spelling performance. The qualitativ e analysis focused on lingui stic category (phonologic, orthographic, and morphologic) and feat ure differences among the three groups of children that the quantitative an alysis failed to distinguis h. For instance, the LLD group struggled with Â“rÂ” colored vowels, making errors like cos for curls Similar to the Hauerwas & Walker (2003) and Bourassa et al. (in press) resu lts, this group also demonstrated difficulties with the past tense -ed. For example, errors included spelling move for moved or crawl for crawled (Silliman et al., 2006). Moreover, children with a LLD exhibited more errors across the th ree categories while typically developing children resolved these errors at an appropr iate phase in the developmental spectrum. In regard to variations of spelling erro rs across groups of ch ildren, researchers have compared spelling errors at many le vels including typically developing, dyslexic and non-native speaking (Apel & Masterson, 200 1; Leybaert & Lechat 2001; Masterson & Crede, 1999; Silliman et al., 2006). Variat ion in spelling acquisition is evident across the groups due to home and instructional e xperiences and ability (Leybaert & Lechat,
17 2001). These differences were established qu antitatively in the Hauerwas and Walker (2003) study in which groups of children di ffered based on the total number of errors. The Silliman et al. (2006) study, on the othe r hand, revealed differe nces between groups of children based on the quality of the errors. Based on the findings previously discussed, it is of importance to implement both quantitative and qualitative assessments into the assessments of spelling that currently exis t in elementary classrooms to assist in identifying those children who may be exhibiting greater l iteracy problems. Assessments of Spelling Spelling assessments can occur in various ways to evaluate a childÂ’s knowledge of spelling skills. Gentry (2004) points out that while much of spelling instruction in the classroom is completed in the form of a weekly spelling test, the resulting information does not adequately assess the knowledge that students possess. As a result, remedial instruction cannot be determined. As disc ussed by Apel et al. (2004a; 2004b), the standard weekly spelling test assesses material in one context and signifies only if the child spelled the word correctly or incorrectly. However, spelling assessments can occur in the form of inventories, writing sample s, pretest-posttest studi es, or standardized assessments. These strategies are described in the following section. Spelling inventories. Bear et al. (2004) make the cas e that spelling inventories, or words specifically chosen by the teacher to represent various spelling patterns and features relating to different phases of spelli ng development, are an effective qualitative technique for spelling analysis. Bear and colleagues suggest th at this type of approach lends itself to analyzing speci fic types of errors that chil dren produce in their spellings versus a quantitative approach, which asse sses only the total number of errors.
18 Numerous types of spelling inventories ex ist or can be created by the teacher or speech-language pathologist and can be grade or level specific regarding the childÂ’s current spelling abilities considering phonologi c, orthographic, and morphologic skills. Words included in the list should represent di fferent spelling patterns at increasing levels of difficulty. The inventories are collected from each child in the form of a spelling test and charted according to the stage/phase or repertoire theory of spelling development (emergent, alphabetic, within-word pattern, sy llables/affixes, and derivational relations), as well as spelling features (Apel et al., 2004b; Bear et al., 2004; Henderson, 1985). Apel et al. (2004a) and Silliman et al. (2006) suggest that, when the broad and narrow approaches are integrated, the resulting info rmation can highlight the phase of childrenÂ’s spelling development and the linguistic compone nts mastered or sti ll in the process of emergence. However, for this type of assessmen t to be effective, it is important that the inventory words are not taught as examples during the intervention process (Bear et al., 2004). Pretest-study-posttest. In GentryÂ’s (2004) opinion, week ly spelling tests can be of significant importance for assessing spelling skills. Gentry (2004) argues that weekly spelling tests present a more rapid approach to understanding a childÂ’s difficulties. While many researchers believe that spelling is best assessed in a writing context (Berninger et al., 1992; Masterson & Crede, 1999) Gentry states that it is too cumbersome for the teacher to sift through the writing samples of entire classes to evaluate errors. In the pretest-study-posttest approach to spelling assessment, a form of response to intervention, the teacher obtains knowledge of each student Â’s errors through a spelling test. The teacher then addresses the errors by teaching the spelling patterns that lead to accurate
19 spellings. The spelling test is repeated at th e end of each unit to determine if the child demonstrates skill mastery (Gentry, 2004). The pretest-study-posttest differs from the spelling inventory in that the preteststudy-posttest assessment directly utilizes the weekly spelling test and the childÂ’s misspellings are obtained; therefore, the spelling inventory may be a more extensive record of spelling errors. In the inventor y approach in which the broad and narrow perspectives are combined, the errors are not only analyzed and reviewed in future spelling lessons, but also the errors can be classified according to linguistic category and their respective features. Norm-referenced assessments. Researchers often use standardized or normreferenced assessments when selecting partic ipants in spelling studies (Apel et al., 2004b). This type of assessment allows for a comparison of an individual childÂ’s spelling performance to a groupÂ’s performance. Thus, the level of spelling proficiency can be determined from the standard scores and pe rcentile rank derived from the childÂ’s test performance (Apel et al., 2004b). While it seems advantageous to compare a childÂ’s spelling performance to other children of the same age, the norm-referen ced assessment actually has minimal relevance for spelling intervention. These assessments do not recognize individua l error patterns in need of remediation, and thei r structure does not allow for assessment of all linguistic aspects of spelling (Apel et al., 2004a). Masterson & Crede (1999) make the case that writing in context is a more effective way to assess spelling errors than weekly spelling tests because the misspelled words can be compared to other words used in the sentence and, therefore, help in determining if othe r words affected the misspelling. The following
20 section describes how various written genres are used to assess spelling and considers how different genres require different knowledge bases and, therefore, may increase spelling errors as a consequence. The Effects of Composing on Spellings Spelling words in context is another a pproach to assessi ng spelling abilities (Gentry, 2004; Masterson & Crede, 1999). Be rninger et al. (1992) suggested that assessing spelling through written compositi ons, whether narrative or expository, was most effective because it revealed not only spelling abilities, but also the fluency of childrenÂ’s writing (how many wo rds were produced), as well as how children structured sentences. Another advantage of compositions is that children use words that are already in their vocabulary, and, therefore, are familiar to them (Paul, 2001) Narrative writing is generally considered ea sier for children to generate because the focus is placed on relationships between people and events. On the other hand, expository compositions are considered more challenging because they focus on factual information and ideas (Paul, 2001) and the topic may require knowledge that is less familiar to the writer (Scott & Windsor, 2000). Also, the structure of the expository genre differs considerably from na rrative organization in that expository structures do not necessarily follow a temporal order of events As Singer and Bashir (2004) note, world knowledge and the type of written genre affect the quality of a childÂ’s writing. Therefore, misspelling may occur less frequently in narr ative writing samples because children are accessing available knowledge about social rela tionships between characters and events. In comparison, expository samples require children to utilize le ss contextualized
21 information since they must focus on relations hips among ideas. It may be the case, that depending on the type of expository ge nre, misspellings will increase. When comparing spelling errors from two different genres, it seems appropriate to consider the differences in th e types and number of errors evident in each genre. Scott and Windsor (2000) compared expository a nd narrative writing samples in 60 children; 20 children with a LLD (mean age = 11;5 year s), 20 chronologically age (CA) matched children (mean age = 11;6 years), and 20 langua ge age (LA) matched children (mean age = 8;11 years). The study focused on compar ing the productivity, fluency, lexical diversity, and grammatical complexity of the three groups in na rrative and expository writing samples. Results revealed that both the children with LLD and the CA children had more difficulties with expository writing. However, the children with LLD demonstrated a greater number of grammati cal errors, including punctuation and spelling errors, in the expository writing sample. The ex pository compositions were also shorter in length and less fluent (Scott & Windsor, 2000). Spelling assessments, such as those just described, are used clinically and educationally. However, regardless of format, the results are meaningless if the analysis is not consistent with a particular scoring system. Scoring Systems When analyzing spelling errors, the type of scoring system must be predetermined. Various types of scoring system s exist and the type of system chosen will determine how misspellings are classified. Tr aditionally, scoring systems have focused on phonological errors, visual accuracy, and or thographic legality (Bruck & Waters, 1988). In contrast, the POMAS (Silliman et al ., 2006) permits the qual itative analysis of
22 linguistic spelling errors thr ough a scoring system composed of linguistic categories and features. These systems are discussed in detail next. Constrained vs. unconstrained approaches. Phonological errors can be analyzed using a constrained or unconstrained system (Bruck & Waters, 1988). In a constrained scoring system, a misspelled word is cons idered phonetically accurate if each phoneme occurred in the same place as in the target word, and, therefore, the word could be pronounced as the target word. Examples of phonetically accurate misspelled words under this system are reche for reach and kepe for keep (Bruck & Waters, 1988). Although the previous examples are misspelled, the Â–e at the end of each word marks the long vowel, and the word would be pronounced like the target (Bru ck & Waters, 1988). The unconstrained system, on the othe r hand, accepts a misspelling as phonetically accurate if each phoneme in the word is represented by a grapheme according to English pronunciation, such as rech for reach or necesite for necessity (Bruck & Waters, 1988). Alt hough the words are spelled inco rrectly, there is a match between phonemes and graphemes making the words phonetically plausible. In both cases, the long vowels are mark ed with the letter name -e Visual accuracy approach. Because spelling can occur through direct memory retrieval of the orthographic form of a word (Bruck & Wa ters, 1988), visual accuracy measures can also be used to analyze the or thography of spelling in a quantitative sense. Apel et al. (2004a) refer to this same appr oach as the bigram approach. With this measure, the amount of overlap between indi vidual letters in th e misspelling and the letters in the target word are assesse d (Apel et al., 2004a; Bruck & Waters, 1988). Bigrams focus on individual letters and their correct order in the misspelled word as
23 compared to the target word. The percenta ge of bigrams produces a visual accuracy score. An example from Bruck and Waters ( 1988) illustrates the bigram measure. The word nature has five bigrams: (na+at+tu+ur+re) and six letters. If a child spelled the word as nachure, the childÂ’s spelling would match the target word with 3 bigrams and 5 letters for a total score of 8 of 11 (five bigram s and six letters), ther efore, the percent of bigrams would be .73 for visual accuracy. The visual accuracy score is problematic for analyzing types of spelling errors for two reasons. First, the score is a representation of orthographi c similarities (i.e., letter order) of the correct and incorrect word rather than a repr esentation of the letters the child used to spell the word. For example, a child may represent all of the phonemes in a word but reverse the order of lette rs. Although the spelling is phonologically represented, the letter reversals within the word will yiel d a visual accuracy percentage representing minimal accuracy of the correct spelling. Furthe rmore, a misspelling of the word in this system would not tell where the error occurred or the type of error. Orthographic legality approach. The orthographic legality approach to misspellings analyzes whether or not the seque nce of graphemes used to spell a word is legal in English spelling (Treiman & Bourassa, 2000b). For example, mfbvg is not orthographically legal because the sequence of consonants does not exist in English. However, frip for trip does not violate English orthogra phy, although it is misspelled, it is a legal sequence of letters. Orthographic legality was designed to assess structural and positional orthographic knowledge. For example, when ad ministering a group of words, such as a spelling inventory, the spelling of each word is analyzed to determine if the words
24 contain orthographically legal se quences of graphemes. The ch ild receives credit even if the word is misspelled as long as the sequen ce of graphemes is legal. Therefore, the spelling error frip for trip although misspelled, would rece ive full credit using the orthographic legality approach because the order of phonemes is legal in English. To determine the percentage of orthographically legal spellings, the total number of words containing legal sequences is divided by the total number of words administered in the session (Silliman et al., 2006). While these three scoring approaches re veal how many spelling errors children make, the need for understanding the types of e rrors is equally important. A lack of focus on the quality of childrenÂ’s errors does not yield insight into u nderstanding childrenÂ’s patterns and the linguistic sources of their errors. The POMAS scoring system (Silliman et al., 2006) details a more qua litative approach to scoring. The Phonological, Orthographic, and Morphol ogical Assessment of Spelling (POMAS). The POMAS (Silliman et al., 2006) allo ws for a qualitative assessment of errors by linguistic category. In contrast to the other scoring systems previously described, the POMAS examines error patterns rather than concentrating solely on the total number of errors. First, misspellings are categorized according to the linguistic categories of phonology, orthography, and morp hology, which is an advantage of using this qualitative system. In addition, spelling er rors can be further classified by feature according to types of errors, such as tenses (inflectional morphology) deletions, clusters, and digraphs (Silliman et al., 2006). Each missp elled word can be classified into a broad category (phonology, orthography, morphology) and then further examined for specific features in a linguistic category.
25 As revealed previously, the process of teaching spelling seems challenging and the connection between spelling and other as pects of literacy de velopment are often overlooked. Different types of spelling assessm ents exist. These in clude pretest-studyposttests and spelling inventories, yet the process of rote memorization of weekly spelling words seems to persist (Apel et al ., 2004a). Weekly spelling tests, much like most of the spelling analyses, quantitativel y examine the errors children commit. The importance of shifting towards qualitative analys es assist in revealing exact error types, yet these types of analyses remain rare. Th e POMAS was the first qualitative analysis of its kind, particularly with re gard to children with a LLD. Statement of the Problem The research reviewed suggests that sp elling develops in phases. Although the phases of development vary among research ers, the basis of spelling development consists of phonologic, ort hographic, and morphologic know ledge and strategies for implementing this knowledge. To analyze the development of spelling patterns, various scoring approaches have been utilized. Thes e approaches tend to focus quantitatively either on the number of errors, visual accu racy scores, or orthographic and phonological information. The total number of errors is then compar ed across groups. What this information fails to reveal is that, although tw o different groups of participants may have the same number of errors in any given categ ory, the quality of th e errors may differ. Similarly, two groups who demonstrate different numbers of errors in any given category may demonstrate the same errors. In general, these quantitative scoring systems are weak in their ability to show patterns and types of errors within and across the phases of spelling development.
26 A need exists to develop a qualitative a pproach to analyzing spelling errors for determining whether or when a shift in error type (phonologic, orthographic, or morphologic) and features occurs in spel ling performance among children in different grade levels. In addition to providing a more detailed approach to classifying spelling errors, qualitative analyses of spelling may be a more effective approach to understanding the evolution of linguistic f eatures in misspellings as ch ildren progress academically. This study attempts to provide the informa tion overlooked by other sp elling studies and to fill the gaps needed for improved spelli ng intervention through the qualitative analysis of spelling errors in children in the lower elementary grades. The current study is a secondary outcome of a project initia ted by Berninger and colleagues (Berninger, Cartwright, Yate s, Swanson, & Abbott, 1994; Berninger, Whitaker, Feng, Swanson, & Abbott, 1996; Bern inger et al., 1992). These three studies included children in grades 1 to 9 (Ber ninger et al., 1994; Be rninger et al., 1996; Berninger et al., 1992). The first two studies (Berninger et al., 1994; Berninger et al., 1992) focused on how developmental skills, such as fine motor skills, working memory, orthography, phonology, and word finding, affect ed writing and read ing acquisition in grades 1 to 3 and 4 to 6. Results indicated that developmental skill s did influence writing acquisition in the younger grades when writ ing is introduced (Berninger et al., 1992). Lower level skills (automatic production of alph abet letters, rapid co ding of orthographic information, speed of sequential finger movement, visual-motor integration, and orthographic-phonological mappings) were found to affect the quality of writing. However, these skills must be developed before the quality of the writing can improve.
27 The second study (Berninger et al., 1994) a ssessed the relationship of cognitive skills to reading and writing. Cognitive skills assessed included: a) rhyme, b) semantic association, c) phrase, narrati ve and expository text, and d) listening recall. Results revealed that cognitive skills and reading and writing were related but not specific to one another. In the third study, Berninger et al. (19 96) assessed cognitive skills (planning, translating, and reviewing) of students in grades 7 to 9 during a writing task. The ability to be an effective writer consisted of these cognitive skills combined with the ability to produce fluent thoughts. Results indicated that strengths in one area of composition were not predictors of strengths in other areas. Stated another wa y, demonstrating strengths in planning did not indicate strength s in the ability to revise. While the purpose of the three studies va ried, they all had composition tasks in common. Across grade levels, students comple ted a narrative and expository writing sample with the same prompts. All were allotted five minutes to write. Then, misspelled words from both tasks were collected for the future analysis of spelling. The misspelled words for grades 1 to 4 are the focus of the current study. Because the current study centers on spelling development in the lower elementary grades, samples from grades 5 to 9 were excluded (Ber ninger et al., 1994; Berninger et al., 1992). However, since all of the data relati ve to the current study were collected in the same manner, inconsistencies should not exist and analysis of the data across grade levels can be conducted.
28 This study examined the qualitative error di fferences in grades 1 to 4 to determine whether grade level and type of writing sa mple had an influence on the type of misspellings. The four questions addressed whether: 1. The number and type of spelling er rors (phonological, orthographic, morphological) differed as a function of grade level. 2. The genre, expository or narrative, and gender affected the total number and/or specific error type of misspellings. 3. Patterns of performance re garding specific errors di ffered across and within grades for the phonological, orthographi c, and morphological categories. 4. Additional information was revealed thr ough a qualitative analysis of features within the phonologi cal, orthographic, and mo rphological categories.
29 Chapter 2: Method ChildrenÂ’s spellings are based on thei r knowledge and expe rience with the phonological, orthographic, and morphological components of language (Carlisle, 2003; Cassar & Treiman, 1997; Dodd & Carr, 2003; Hauerwas & Walker, 2003), which is influenced by reading and writing as they a dvance in grade levels. Therefore, the number and type of misspellings present in a writing sample may differ across grade levels. This study examined the different types of misspel lings in childrenÂ’s writing across grade levels 1 to 4. The data used in this stu dy were extracted from the writing samples of children gathered in the three different studies by Berninger et al. (B erninger et al., 1994; Berninger et al., 1996; Berni nger et al., 1992). The purpose of this study was to examine closely the spelling errors of young children to determine what, if any, error patterns consistently existed in their spelling development. The data collected by Berninger et al. (1994; 1992) was coded using the POMAS codi ng system (Silliman et al., 2006) and the errors were classified into broad (phonological, orthograp hic, morphologica l) and narrow (linguistic feature) categories. Participants 1st to 3rd grade. A total of 300 children in grades 1, 2, and 3 were selected from eight different elementary schools in three school systems in the Seattle, Washington area. Of the three school systems, one wa s suburban, one was suburban/rural, and one was urban. The 300 children who participated in the studies were selected from 570
30 volunteers. An equal number of girls and boys were selected to pa rticipate. MothersÂ’ educational level ranged from high school to college and beyond. Th e grade 1 childrenÂ’s age ranged in years and months from 6;6-8; 2, grade 2 ranged from 7;5-9;1, and grade 3 ranged from 7;6-9;11 (Berninger et al., 1992). Mean ages and standard deviations were not provided. 4th grade. While the Berninger et al. (1994) study included grades 4 to 6, this study focused only on the fourth grade data. A total of 100 4th grade children were chosen from five urban and suburban schools. An equal representation of boys and girls were included. In this sample, mothersÂ’ educati on level ranged from less than high school to college and beyond. Age ranges for the childre n were not provided (Berninger et al., 1994). The following table provides the per centages of ethnic re presentation for the children included for this project. Table 2.1. Ethnic Representation of Particip ants in Grades 1 to 4 (Berninger et al., 1994; Berninger et al., 1992) 1st-3rd grade 4th-6th grade Asian American 6% 14% African American 6% 10% Hispanic 3% 4% Caucasian 84% 70% Native American <1% 1% Other N/A 1%
31 Materials The study consisted of separate narrativ e and expository writing samples obtained from the 400 children. All were given paper an d a writing utensil (B erninger et al., 1992, 1994, 1996) and the examiner used a watch or timer to limit them to 5 minutes of writing per sample. The narrative writing sample was identical for all grad e levels, and began with the prompt Â“One day _____ had the best or worst day at sc hool.Â” The expository writing sample prompt was also identical for a ll grade levels, and began with the sentence Â“I like ____ because ____.Â” Procedures The children in grades 1 to 3 were test ed during individual sessions in a quiet space provided by each school. The assessment took place during the school year, sometime between February and May, 1990. At the beginning of the experiment, each child was assigned a number. Children with an even number completed the narrative essay first, while children assigned odd numbers completed the expository essay first. The examiner gave the child the prompt and five minutes to write. At the end of five minutes, each child was asked to read their writing sample to the examiner. The examiner then transcribed the childÂ’s re tell of the compositions on a separate sheet of paper and compared the transcribed composition with th e original composition (Berninger et al., 1992). Misspelled words occurr ing within the sample we re extracted from the composition and listed on a separate sheet of pa per with the correctly spelled word listed next to it. The grade 4 children were also tested in individual sessions in the sixth or seventh month of the school year. Similar to the task for grades 1 to 3, th e grade 4 participants
32 were required to write a narra tive and an expository essay wi th prompts identical to the compositions for grades 1 to 3. Table 2.2 provi des the mean number of words and clauses per writing sample for each grade level. Table 2.2. Grade Performance by Writing Sample Mean Words Narrative Mean Clauses Narrative Mean Words Expository Mean Clauses Expository 1st Â– 3rd 35.20 5.62 33.16 5.72 4th 57.47 7.94 55.61 8.37 After reviewing the compositions produced by e ach child, the examiner listed the childÂ’s misspelled words on a separate sheet of paper with the correct word listed next to it, based on the examinerÂ’s best judgment of th e misspelled word (Berninger et al., 1994). In summary, although 400 children differed in age, each grade level was given the same narrative and expository writing prom pts. The misspelled words from each childÂ’s samples were extracted, and these misspellings were analyzed in this study as a secondary analysis. The number of errors a nd type of each error was determined using a qualitative analysis system. POMAS. The POMAS scoring system (Silliman et al., 2006) classifies errors into the three broad areas of developmen t: phonology, orthography, and morphology. The POMAS also divides the three br oad categories into smaller subsets of linguistic features that describe qualitative diffe rences within each error cat egory. For example, an error would be coded as phonologica l, if the child spelled dresses as desses, because the child omitted the second letter in the consonant cluste r, and the complete sound structure of the
33 word was not represented. This type of error therefore would be clas sified under the code Phonological-Cluster Reduction (PCR). Orthographic errors were analyzed accord ing to misspellings that represented a digraph and the marking of sy llable juncture. Again using dresses as the sample word, if the child spelled the word as dreses, this error would be coded as an orthographic error due to the omission of the double Â“ sÂ” (Orthographic Digraph). Morphological errors were analyzed according to inflections, including the presence or absence of past, present, or present progressive tense, and derivations with and without phonological changes, as well as pr efixes/suffixes. For example, if a child spelled the word dresses as dress the error would be c oded as an inflectional morphological error since the plural marker Â– es (Morphological Inflection) was omitted. The error codes from the or iginal POMAS were included in the coding system for this study. However, modifications were made and many new codes were established due to the variety of words obtained from each participant contrasted with the original POMAS, in which each child spelled the same set of words. Tables 2.3 to 2.8 include a small sample of the linguistic features and the error category to which each feature was assigned. A full listing of the error codes can be found in Appendix A. Table 2.3. POMAS Coding System Â– Phonologi cal Errors (Silliman et al., 2006) Feature Word Example Short vowel cast Caste Long vowel cake Cak Diphthong found Fond Sonorant clusters and Ad
34 Table 2.3 (Continued) Feature Word Example Cluster reduction struck Stuck Vocalic /r/ curls Cos Table 2.4. POMAS Coding Syst em Â– Orthographic Errors (Silliman et al., 2006) Feature Word Example Digraph ship Sip Long vowel pattern keep Kipe Unusual vowel pattern found fowned Syllable juncture (consonan t doubling) stirring Stiring Syllable juncture (Â“yÂ” to Â“iÂ”) cries Cryes Table 2.5. POMAS Coding System Â– Morphol ogical Errors (Silliman et al., 2006) Feature Word Example Inflection-present tense cries Cry Derivational shift with phonologic change magician magishen Inflection-plural (-s pronounced /z/) curls Curl Inflection-plural (-es) dresses Dress Derivational Â– suffix later Late Agentive Â“erÂ” suffix prisoner Prison Derivation (no phonologic change) government govrment Derivation (phonologic change) magician megishen
35 Table 2.6. POMAS Coding System Â– Mi xed Phonological-Orthographic Errors (Silliman et al., 2006) Feature Word Example Letter reversals thier their Vowels missing double dbl Table 2.7. POMAS Coding System Â– Mi xed Morphological-Orthographic Errors (Silliman et al., 2006) Feature Word Example Misspelled root word resulti ng in phonologically accurate spelling Magician magishan Table 2.8. POMAS Coding System Â– Mi xed Morphological-Phonological Errors (Silliman et al., 2006) Feature Word Example Visually similar error car are Data Reduction. The misspelled words were entered in the Systematic Analysis of Language Transcripts Software (SALT) (Miller & Chapma n, 1991) for the purpose of coding the spelling errors. Each sample was id entified by grade level, gender, and type of writing sample (narrative vs. expository). Then, the errors were coded by category
36 (phonological, orthographic, and morphologica l) and feature using the POMAS scoring system. A specialized computer program was create d to assist with quantifying the results of the qualitative analysis. The program analy zed each line of coded information from the SALT files and tallied the number of times each error code was used. The total occurrences of each error per writing sample child, grade level, and across all grade levels were identified. Finally, the data obtained from the computer program was transferred into an Excel file for statistical analysis. The error f eatures belonging to the major categories (phonological, orthographic, morphological, and mixed) were grouped together and sums were obtained to determin e the total number of errors for each major category in the individual samples. For statistical analyses, these totals were normalized by the total number of errors produced by a pa rticular child. In a ddition, words in error were grouped together to assist in identifying commonly misspelled words. Qualitative Analyses. A qualitative analyses was completed to determine types of errors and error patterns that occurred in th e data. The first analysis completed compared the frequency of each error according to individual feature. The errors were compared within and across grade levels a nd according to the type of wr iting sample. An analysis of words frequently in error was also complete d. Although all of the ch ildren were required to write based on the same prompts, the words produced were unique to each child; therefore, the analysis of words frequently in error determined if children consistently used, and made, spelling errors on the same words. In conjunction with the analys is of words frequently in error, a second analysis was performed to determine variations in th e spellings of the same word. To complete
37 this analysis, each misspelled word was groupe d according to the intended word even if each misspelling varied. This type of analys is allowed determination of patterns in misspellings to see if there were words c onsistently misspelled by children and whether different children misspelled words exactly the same way. Agreement. A second trained examiner reanaly zed and coded the spelling data from 5% of the participants to determine interrater consistency. The participants included in the recoding were randomly selected from the grade 1 to 4 data with an equal number of boys and girls per grade level select ed. The second examiner recoded both the narrative and expository writing samples of each pa rticipant to assist in determining error patterns unique to the individua l. Six participants were selected from each grade level with an average of 7.583 words per participant. The second examiner was directly trained to use the POMAS coding system through the us e of spelling samples not related to the study. Errors observed during the training were reviewed and further examples of the same error type were provided. The agreement findings will be reviewed in the following chapter. Statistical Analysis. A four-way MANOVA was used to compare and analyze the quantitative data. The independent variable s were writing sample (narrative vs. expository), error type (phonol ogical, orthographic, morphol ogical, or combinations of these errors), gender (male/female) and grade level (1-4). The dependent variable was the frequency of occurrence in each error category. Post hoc tests were run when appropriate. Effect sizes were calculated. Two-way ANOVAs were also completed to compare and analyze differences in the types of errors the children made in the writing samples. The independent variable
38 was grade level (1-4) and the dependent variable was the specific error types. T-tests were completed for each comparison of major category (phonology, orthography, morphology) and grade level. Effect sizes were calculated as appropriate.
39 Chapter 3: Results This study was designed to describe the erro r patterns of children in grades 1-4 to determine whether the childÂ’s level of educa tion, type of writing sample, and sex had an influence on the type of errors seen in thei r spelling performance. The data was collected from two separate writing samples administer ed to the children dur ing a previous study (Berninger et al., 1994; Berninge r et al., 1996; Berninger et al., 1992). The spelling errors were transcribed for each child and coded fo r grade level and type of writing sample (narrative and expository). The data were an alyzed using an embellished version of the POMAS (Silliman et al., 2006). Qualitative analyses included determining the most frequently misspelled words and the most common error codes from the sample data. This study examined the qualitative error di fferences in grades 1 to 4 to determine whether the childrenÂ’s grade level and type of writing sample had an influence on the type of misspellings. The thre e questions addressed whether: 1. The number and type of spelling er rors (phonological, orthographic, morphological) differed as a function of grade level. 2. The genre, expository or narrative, and ge nder affected the number and type of misspellings. 3. Patterns of performance re garding specific errors di ffered across and within grades for the phonological, orthographi c, and morphological categories.
40 4. Additional information was revealed thr ough a qualitative analysis of features within phonological, orthographic, and morphological categories. Inter-Examiner Agreement Agreement was conducted regarding featur e error categorizati on. Of the total 800 writing samples (N = 2 per child x 400 children) 6 percent (N = 48) of the samples were randomly selected for the agreement analysis A second examiner was trained to recode the narrative and expository data for each se lected participant using the POMAS scoring system. Training consisted of reviewing the rules within each of the POMAS categories (phonology, orthography, and morphology) and then having the second rater analyze some of the current spelling samples along w ith the primary examiner. Once the second rater was comfortable with the scoring system, she was asked to rate randomly selected samples from the database. Due to the complexity of the coding for individual error features, the features were collapsed into specific error cate gories (phonology, orthography, and morphology) rather than specific error feature for agr eement analysis. The agreement was calculated using the following formula (Salvia & Ysseldyke, 2001): 100 2 agreements of number total agreements of number Agreement The results of the analysis determined an overall 75% agreement for specific error category (phonology, orthography, and mo rphology). A 75% agreement between examiners is a reasonable outcome due to th e complexity of the study. The coding system was complex in that several errors overlapped into two different major categories. Therefore, the category coding was left to th e examinerÂ’s discretion. In several instances in the coding system, the error feature wa s the same although the error feature varied.
41 Therefore, the examinerÂ’s interpretation of the childÂ’s representation of the misspelled word had an effect on the code selected. Fo r example, the coding system included three different codes for vowel errors. One code was represented as a phonological error regarding short vowels, while the other error was an orthog raphic vowel error. Given the word funny spelled as fany which was included in the random sample, the first examiner coded this error as OVE (orthographic error in which the substituted vowel error still represented a short vowel) while the s econd examiner coded the error as PSV (phonological error involving th e short vowel). Thus, althou gh both examiners agreed on the specific error feature, agreement was not revealed for this error because the exact error code or category match did not occur. Instances such as this example occurred throughout the study. Although two examiners may agree on the error type, but not necessarily the main category, the resulting agre ement appears to be lower than the actual coding represented. An additional barrier effecting a strong agreement between examiners occurred because the original writing samples for the s ubjects were not available for reference. Therefore, the context of the misspellings could not be determined and other correct spellings in the samples were not available fo r review. This information would be useful because the examiner could observe if any of the misspelled words were spelled correctly at any other point in the samp le and if a previously spelled word may have impacted the misspelling. Overview of Subject Performance The data then was analyzed to determine the total number of spelling errors for each of the four grade levels (1st-4th) included in this study. Each grade level included
42 approximately 100 equally represented male and female participants who contributed both a narrative and expository writing sample. For this analysis, the total number of spelling errors was analyzed regardless of the type of writing sample or sex of the child. To normalize the data across participants, each childÂ’s data was normalized as follows: the number of errors produced in e ach major category (phonology, orthography, morphology) was divided by the total num ber of errors produced by the child. Table 3.1 represents the means and standa rd deviations for the analysis. Table 3.1. Total Number of Errors for Each Grade Level (N= 3,264) Grade Level Mean Standard Deviation 1 7.88 5.903 2 5.77 5.456 3 4.73 4.663 4 4.99 4.921 Total 5.84 5.390 The data were also analyzed to de termine the influence writing sample (expository or narrative) had on the total numbe r of errors for each grade level. Table 3.2 represents the means and standard deviations for the analysis of the total number of errors based on writing type.
43 Table 3.2. Total Number of Errors in th e Two Writing Samples According to Grade Level. Type Mean Std. Dev. 1 Expository Narrative 7.83 7.93 6.055 5.776 2 Expository Narrative 5.27 6.26 4.897 5.946 3 Narrative Expository 4.98 4.49 4.868 4.458 4 Expository Narrative 4.96 5.02 4.684 5.172 An additional analysis was completed to determine the total number of errors according to grade level based on gender. Ta ble 3.3 presents the means and standard deviations for this analysis. Table 3.3. Total Number of Errors A ccording to Grade Level Based on Sex. Grade Gender Mean Std. Dev. 1 Boys Girls 8.16 7.60 6.244 5.558 2 Boys Girls 5.72 5.82 5.420 5.519 3 Boys Girls 4.46 5.01 4.711 4.620
44 Table 3.3 (Continued) 4 Boys Girls 4.68 5.30 4.304 5.474 The data previously described was analyzed in closer detail to answer the research questions of importance for this study. Th ese analyses are discussed in the following sections. Statistical Analyses of Data for the Research Questions Question 1: Grade Level Effect s on Number and Types of Errors A four-way MANOVA was conducted to analyze whether 1st through 4th grade typically developing children differed in th e number and type of spelling errors they produced as a function of writing sample and gender. Specifically, a 4 (grade level) x 4 (error type) x 2 (gender) x 2 (sample type) an alysis was completed. The results of the MANOVA indicated only one significant inter action, the interaction between grade and error type, F (9, 2352)=4.838, p <.001, p 2=.018. This finding would suggest that the differences between error types were depende nt on grade. However, the effect size suggested that this inte raction explained very little of th e variance in this analysis. Post hoc testing for across grade differences was co mpleted using t-tests, with a Bonferroni correction ( p =.002). Of the 18 pairwise comparison of interest, only 3 were shown to be statistically significant (see Appendix B for ttest results). As illustrated in Figure 3.1, and supported by the t-test results, 1st graders made more phonological errors than did the 2nd and 4th graders. All groups made similar num bers of orthographic errors, while 4th graders made more morphological errors than 2nd graders.
45 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1234 GradeError Type Ratio Morphological Orthographic Phonological UncodedFigure 3.1. Comparison of Category Error Ratios by Grade Level. Question 2: Did The Writing Sample or Gende r Affect Number And Type of Spelling Error? This question can be answered by con tinued explanation of the previously described four-way MANOVA. The results in dicated that none of the interactions involving writing type nor the main effect was significant, F (1,784)=.119, p =.730, p 2 <.001. This finding suggested that the differences between the types of errors were not dependent on the type of writing sample and that the number of errors within each writing sample type were not significantly different. In addition to the analysis completed to determine if writing sample had an effect on the number and type of spelling errors, an analysis to determine the effect of gender
46 on the number and types of errors was also completed. In this case, the four-way MANOVA revealed that none of the intera ctions involving gender were significant, F (1,784)=.247, p =.619, p 2<.001. In addition, the main effect of gender was not significant. In other words, boys and girls made similar types of errors. Question 3: Error Patterns within Grade Level Statistical analyses were completed for each of the major error categories, phonology, orthography, and morphology, to dete rmine differences across grade levels for particular error types and between error types within grade levels. The error types were selected for each category based on th e total number of occu rrences within the sample and those that occurred relative to stage theory. In other words, the phonological and orthographic errors were chosen because the total number of error occurrences for these error types decr eased across the grade levels a nd the morphological error types increased from first to fourth grade. Twoway ANOVAs were completed to determine relationships within and across grade levels while post hoc testing was completed to identify specific differences between error types. These results will be discussed by error category. Phonological errors. To determine if there were grade differenc es in the use of pa rticular phonological error patterns, a two-way MANOVA was run with grade and error type as the independent variables and error frequency as the dependent variable. The results revealed a two-way interaction between grade level and error type, F (9,2388) = 6.45, p < .001, np 2 = .024. This finding would suggest that the occurrence of e rror type was dependent upon grade level. To further analyze this interac tion, differences in errors will be considered
47 both across grades for a particul ar error pattern and within grades for differences in error pattern use. Across grade patterns. Post hoc testing with the Bonferroni procedure (a procedure which controls for family-wise erro r) revealed that 8/24 pairwise comparisons of interest were significant. (Appendix C lis ts the data associated with these post hoc comparisons). As illustrated in Figure 3.2, first graders made the most errors for all phonological error types with s econd through fourth grades following in a decreasing pattern. However, only PFPV (final positi on voicing) and PSON (sonorant clusters between first and second grades di splayed a significant difference. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 PFLPPFPVPSEPSON featuremean 1 2 3 4Figure 3.2. Decreasing Phonologi cal Error Feature Use by Grade Level. Phonological Codes Represented: Flap (FLP), Final pos ition voicing (FPV), Silent e (PSE) and Sonorant Clusters (PSON).
48 Within grade patterns. Post hoc testing using t-tests with a Bonferroni correction ( p = .002) revealed that 17/24 pairwise compar isons of interest were significant. As illustrated in Figure 3.4, children in all 4 grades demonstrated a similar pattern in that the PSE (silent e) and PSON errors were produced the most frequently. Frequency of pattern use was the largest for the first graders, w ith PSE being used significantly more often than PSON ( p =.0003). This was also true for the second and th ird graders ( p <.001 and p =.0001 respectively). A significant difference for the occurrence of the PSE and PSON error types was not found fo r the fourth graders ( p =.1871). Orthographic errors To determine if there were grade differenc es in the use of particular orthographic error patterns, a two-way MANOVA was run with grade and error type as the independent variables and error frequency as the dependent variable. The results revealed a two-way interaction between grade level and error type, F (12,3184) = 6.5, p < .001, np 2 = .024. This finding would suggest that the occurrence of e rror type was dependent upon grade level. To further analyze this interac tion, differences in errors will be considered both across grades for a particul ar error pattern and within grades for differences in error pattern use. Across grade patterns. Post hoc testing with the Bonferroni procedure (a procedure which controls for family-wise erro r) revealed that 5/30 pairwise comparisons of interest were significant. (Appendix C lis ts the data associated with these post hoc comparisons). As illustrated in Figure 3.3 first graders made the most errors for all orthographic error types. Orthographic errors decreased in quantity across second through
49 fourth grades. A significant difference was found for the OL N (letter name) error type when comparing first to second grade, first to third grade, and first to fourth grade. Similarly, a significant difference was also found for the OVr (vocalic r) error type when comparing first to third and first to fourth grades. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ODIOLNOLROLSOVr featuremean 1 2 3 4Figure 3.3. Decreasing Orthogr aphic Feature Use by Grade Level. Orthographic codes represented: Digraph (ODI), Letter name (OLN), Letter reversal (OLR), Letter sound (OLS), and Vocalic r (OVr). Within grade patterns. Post hoc testing using t-tests with a Bonferroni correction ( p = .001) revealed that 21/40 pairwise compar isons of interest were significant. As illustrated in Figure 3.6, all 4 grade levels comm itted the most orthographic errors in the OLN and ODI (digraph) categories. However, th e first graders had the most errors in the
50 OLN category followed by the ODI category wh ile the other 3 grad es displayed the opposite pattern. Frequency of pattern use was th e largest for the firs t graders, with OLN being used significantly more often than ODI ( p =.0001). The OLS (letter sound) and OVr error types were also common for all 4 grade levels occu rring in a decreasing manner across grade levels. However, no significan t differences were revealed. Significant differences were found for all four grade levels between ODI, OLN, OLS, OVr when compared to OLR (letter reversal). Morphological Errors To determine if there were grade di fferences in the use of particular morphological error patterns, a two-way MANOV A was run with grade and error type as the independent variables and error frequenc y as the dependent variable. The results revealed a two-way interaction be tween grade level and error type, F (2,1592) = 5.491, p < .001, np 2= .020. This finding would imply that the o ccurrence of error type was dependent upon grade level. To further analyze this in teraction, differences in errors will be considered both across grades for a partic ular error pattern and within grades for differences in error pattern use. Across grade patterns. Post hoc testing with the Bonferroni procedure (a procedure which controls for family-wise erro r) revealed that 4/18 pairwise comparisons of interest were significant. (Appendix C lis ts the data associated with these post hoc comparisons). As illustrated in Figure 3.4, an overall pattern for al l morphological error types was not determined. Morphological errors for the MCON (contractions) and MHOM (homonyms) error types were greatest in fourth grade. However, the MINF (inflections) error type was greatest for first graders with fourth graders having the
51 second most frequent occurrences. A signi ficant difference was found for the MHOM error type when comparing first to third grade, first to fourth grade, and second to fourth grades. The MINF error type also revealed a significant difference when comparing first to third grade. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 MCONMHOMMINF featuremean 1 2 3 4Figure 3.4. Morphological Error Feature Use by Grade Level. Morphological codes represented included: Contractions (MC ON), Homonyms (MHOM), and Inflections (MINF). Within grade patterns. Post hoc testing using t-tests with a Bonferroni correction ( p = .004) revealed that 8/12 pairwise compar isons of interest were significant. As illustrated in Figure 3.4, children in all 4 grad es demonstrated the most errors in the MHOM and MINF error categories. A pattern for error use was not determined for this
52 category. The first and second graders made more MINF errors than MHOM and the third and fourth graders made more MHOM than MINF errors. However, only the first graders were found to have significan tly more MINF errors than MHOM ( p =.0002). Significant differences were also determined for all four grade levels for the MCON and MINF error types, with second, third, and fourth graders demons trating a significant difference in MCON and MHOM errors. In summary, two-way MANOVAs fo r phonology, orthography, and morphology determined that the occurrence of error type was dependent upon grade level for all three error categories. Between and within grade patterns were de termined. The first graders made the most errors in a ll phonological and orthographic er ror categories. All children committed the most errors in the PSON ( phonological Â– sonorant clusters) and PSE (phonological Â– silent /e/) types for phonologica l errors and OLN (orthographic Â– letter name) and ODI (orthographic Â– digraph) types for orthographic errors. The first graders had significantly more instances of the PSE error type than the PSON error type and the OLN error type than the ODI error type. A common pattern was not determined for morphological errors. The MHOM (m orphological Â– homonyms) and MINF (morphological Â– inflection) errors were great est among all grade levels. The first graders made significantly more MINF than MHOM erro rs which is consistent with the finding of the Carlisle (1996) study discussed in Chapter 1. Significant differences were also determined for all grade levels for the MCON (morphological Â– contraction) and MINF error types. Second through fourth grades de monstrated significant differences between MCON and MHOM errors perhaps resulting fr om the second graders having less formal exposure to these language elements.
53 Questions 4: Qualitative Analysis of Error Patterns Within Grade Level A qualitative analysis was completed after the misspelled words were coded and statistically analyzed. This analysis was completed to determine common patterns among error types of misspelled word s. All of the misspellings of the same word were grouped together. Next, the same misspellings were th en sorted to determine if patterns emerged within a specific grade level or if children across all grade levels exhibited the same patterns of misspellings. Based on this info rmal analysis, common misspellings were revealed among different children. Table 3.4 provides a representation of the most commonly misspelled words (misspelled more than 25 times) throughout the writing samples. The table also provi des the number of occurrences for the specific misspelling and which grade levels the misspellings occurred. Table 3.4. Most Commonly Misspelled Words in The Spelling Samples. Most common Misspelling (occurring among different children) Word Number of times misspelled Word Number of occurrences/total children committing error Grade levels of children using the common misspelling because 113 becaus 7/7 1st,2nd,3rd recess 74 reses 13/12 1st,2nd,3rd,4th favorite 61 favrit 7/7 1st,2nd,3rd friend 55 frend 13/13 1st,2nd,3rd,4th when 51 wen 14/13 1st,2nd,3rd
54 Table 3.4 (Continued) Most common Misspelling (occurring among different children) Word Number of times misspelled Word Number of occurrences/total children committing error Grade levels of children using the common misspelling nice 46 nise 8/7 1st,2nd,4th teacher 45 techer 10/9 1st,2nd,3rd,4th school 43 scool 3/3 1st,2nd,4th too 39 to 39/36 1st,2nd,3rd,4th they 35 thay 12/10 1st,2nd,3rd,4th there 35 ther 8/8 1st,2nd,3rd played 34 plad 6/6 1st, 3rd theyÂ’re 33 there 10/10 1st,2nd,3rd,4th friends 33 freinds 7/4 3rd,4th pretty 27 prety 4/4 1st,2nd,3rd like 27 lik 9/9 1st,2nd,3rd didnÂ’t 27 didÂ’nt 4/3 2nd,4th with 26 wihe 2/2 2nd As indicated in the previous table, ma ny children misspelled the same words, and common misspellings were evident among many of them. It should be noted that when
55 analyzing this information, the common erro r patterns indicated occurred among different children. For example, if the same child used the same misspelling throughout the writing sample, that misspelling was considered as one instance of that error. As is evident in Table 3.4, the analysis revealed that the word because was most frequently misspelled; however only seven children, in 1st, 2nd, and 3rd grades, misspelled it the same way. Fourth graders did not have any instances of this spelling. Similar findings were evident in the words favorite and recess in which a small number of children agreed on a common mi sspelling when compared to the number of times the words were misspelled. The word favorite was commonly misspelled by 1st, 2nd, and 3rd graders and recess was commonly misspelled by all f our grades. Thus, from these numbers, it is evident that there were a large number of error discrepancies between spellings and grade levels. Children were not relying on the same phonological and orthographic strategies or knowledge and thus many different spellings occurred for the same word. In contrast to the previously descri bed misspelling patterns, the analysis of they (common among 1st-4th graders), theyÂ’re (common among 1st-4th graders), friend (common among 1st-4th graders), when (common among 1st-3rd graders), and like (common among 1st-3rd graders) revealed more common agreement among misspellings when comparing the number of children who sp elled the word the same way and the total number of times the word was used. Of utmost interest was the word too This word was misspelled 39 times throughout all 800 writing samples. In all 39 instances, 39 different children from all four grade leve ls agreed on the misspelling as to which is a homonym.
56 A qualitative analysis was also completed to determine the most commonly used error codes. Table 3.5 presents this inform ation (refer to Appendix A for a complete listing of all of th e error codes). Table 3.5. Most Commonly Used Error Codes for All Grade Levels. Code Number of uses Code Number of uses OVE 620 OLS 254 OLN 409 OUVP 234 ODI 367 POR 229 PSE 306 OVr 198 Key: OVE = orthographic vowel error, OLN = orthographic letter name, ODI = orthographic digraph, PSE = phonological silent /e/, OLS = orthographic letter sound, OUVP = orthographic unusual vowel pattern, POR = phonological-orthographic reversal, OVr = orthographic vocalic /r/ As illustrated in Table 3.5, the most fre quent error type throughout the data analysis was the OVE feature. The OLN and ODI error types were also used numerous times followed by the PSE error type. The OLS, OUVP, and POR error types were used relatively evenly throughout the samples. As indicated from the information in Table 3.5, the most commonly used error codes occurred in the orthographic and phonological error categories, with the orthographic error codes the most frequent. These error codes were found the most often because they represente d a broader category. For example, the OVE (vowel) error type could be a ny kind of vowel error (short or long). Rather than having an individual orthographic category for each, all vowel errors were classified into that category. This increased the number of possibl e occurrences for the error type because specific error categories such as one category for short vowels and a second category for
57 long vowels did not exist in which the ort hographic vowel errors could be assigned separately. Thus, it may have been appropriate to have had individua l error categories for each type of orthographic error. To conclude, a qualitative analysis revealed common spelling errors and patterns between error codes. The most commonly misspelled word was because which frequently occurred in 1st, 2nd, and 3rd graders. The commonly misspelled words showed varying patterns among grade levels. Some of the misspellings were committed by only one grade level while others were committed by all four grade levels. The most common error codes were also determined. Analyses indicated that the OVE error type occurred with high frequency, and that error types mo st often represented the phonological and orthographic categories.
58 Chapter 4: Discussion The primary objective of the current st udy was to determine whether differences in the quality of errors produced by children in grades 1-4 were influenced by their grade level. In other words, the major issue addres sed whether different pa tterns were evident when comparing the types of the errors rath er than strictly comp aring the quantity of errors. Secondary objectives were to dete rmine if the written genre (expository or narrative) influenced the number and types of errors and if gender influenced the number of errors. Results of the study indicated that the type of errors varied as a result of grade level. Analyses were completed for the er ror types in each major category (phonology, orthography, and morphology). Findings revealed that the most fre quent feature errors were clustered in the following categor ies: a) for phonology, final position voicing (PFPV), silent /e/ (PSE), and sonorant clusters (PSON); b) for orthography, letter name (OLN) and vocalic /r/ (OVr); and c) for morphology, homonyms (MHOM) and inflectional markers (MINF). The three categor ies were significantl y different from one other when comparing across grade levels. Gr ade level also influe nced the number of errors that children produced. Finally, the fi ndings indicated that ne ither genre type nor gender significantly contributed to the number of sp elling errors. First, these results will be discussed with a focus on pattern s found within error types and across grade levels. Strengths a nd limitations of the POMAS system of
59 classification will also be considered. Then, findings will be related to their educational and clinical implications fo r improved spelling intervention. Question 1: Grade Level Effect s on Number and Types of Errors The purpose of this first question was to de termine if the number of errors in each spelling category (phonology, ort hography, and morphology) differe d as a result of grade level. Results of the four-way MANOVA reveal ed an interaction between the number of error types and grade level, although the effect size ( p 2=.018) suggested that this interaction had minimal practical significance. Further analysis using t-tests revealed a significant difference in the occurrence of phonological errors between grades 1 and 2 and grades 1 and 4. Children in grade 1 made more errors. Moreover, a significant difference occurred in the fr equency of occurrence of mo rphological errors between second and fourth grades. In this compar ison, grade 4 children ma de more errors. As suggested by stage theory discus sed in Chapter 1, phonological knowledge develops early as children are exposed to the alphabet and gain letter knowledge. Therefore, these results are consistent with other studies (Reece & Treiman, 2001) that have found the number of phonological errors to decrease as children advance in grade level. A major difference between previous studies (e.g., Reece and Treiman (2001) and this study is the number of spelling samples th at were analyzed (N = 400), lending further support to previous findings that childr en become less dependent on the phonological route as their memory for sp ellings (sight words) incr eases (Ehri & Snowling, 2004). Stage theory (e.g., Bear et al., 2004) places the development of morphological skills in the later pha ses of spelling development. For example, inflectional morphology develops in the upper elementary and middle school grade levels before derivational
60 morphology, which may not emerge until middle school and continues into adulthood. As described by Bear et al., (2004), with th e spelling mastery of inflected morphology, children can already spell simple words and t hus move to include markers that denote past tense, plurality, or aspect (e.g., the progressive marker) (Bear et al., 2004). Other studies challenge the Bear et al. (2004) description of inflectional morphology development. Even young children u tilize characteristics of inflections. In the written narrative studies of Carlisle (1996) and Green et al. (2003) with children in grades 2 to 4, conventional use of inflectional marker s for the past tense, plurals, and the progressive increased significan tly between grades 2 and 4. The more frequent written inflections appeared to be relatively well ma stered in less demanding narrative writing by grade 4 (Green et al., 2003), with a transi tional period in more correct use spanning grades 2 to 3 (Carlisle, 1996). On the othe r hand, derived morphol ogical representations require a longer period of time for their conventional spellings to be mastered (Carlisle, 1987, 1988; Green et al., 2003; Nagy et al., 2003). Results from this study did not completely agree with these previous findings in that inflectional errors increas ed between grades 2 to 4. Based on the findings of Carlisle (1996) and Green et al. (2003), th ese errors should be infrequent in grade 4. In regard to derivational morphology, a greater number of error occurrences would have been expected; however, derivational errors rarely occurred and, as a consequence, were not a factor in these findings. Because the orig inal writing samples did not accompany the misspelled words, it was impossible to determine if the children correctly spelled derived forms or failed to use them at all.
61 Orthographic errors were also anticipated to decrease as a function of increasing grade level; however, significant differences for the number of error occurrences in this category were not evident. In fact, when pl otting the error type ra tio by major category, children in grade 4 exhibited more orthographi c errors than did child ren in grades 1 and 3. The increase in orthographic errors for the fourth graders was possi bly a result of using more complex and less common words. In this case, the fourth graders were able to represent the phonological skeleton of the word, but instructional factors, such as lack of attention to spelling refinements, may have contributed to increased errors. Examples included the spelling of compound words with a space, such as everybody as every body, or choosing the incorrect letter to represent a sound as in sament for cement. For the inflectional morphological and orthographic cat egories, these patterns may add support to repertoire theory (Sulzby, 1996) or the idea that multiple aspects of linguistic knowledge interact simultaneously. A second consideration accounting for th e orthographic and morphological errors may be that the academic ab ilities of each child were unknown. Based on participant information, children with possible dyslexia and/or a LLD were not excluded from the sample, although all participants had to meet inclusion criteria. For example, the fourth graders had more orthographic le tter reversal errors than did children in the lower grades. As discussed previously, children of this ag e would be expected to demonstrate fewer errors in all categories incl uding inflectional morphology (Green et al., 2003). Therefore, it is possible that the grade 4 sample wa s populated with children still experiencing significant decoding or other phonological pr ocessing difficulties characteristic of children with dyslexia (Bourassa et al., in press) or a LLD (Catts, Hogan, & Fey, 2003).
62 Question 2: Genre and Gender Effects on the Number and Type of Spelling Errors The second question asked if the type of written genre (narrative and expository) or gender affected the number of errors produc ed in each grade level. The results of the MANOVA did not reveal signifi cant differences either for the two written genres or gender. In regard to genre, children in each grade level did not display significantly more errors in one type of writi ng over the other, an unexpected finding. Based on the results of the Scott and Windsor (2000) study, expected outcomes would have favored considerably more errors in the expository wr iting samples than in the narrative samples. It should be pointed out that data from the current study were based only on listings of spelling errors by child. The actual writing samples were not available; thus, other variables, such as the productivity (l ength) of the narrative versus expository samples, could not formally be factored in to the results. Based on the mean number of words and clauses produced by grade level (s ee Table 2.2), productivity did seem to be a variable as the compositions were generally short. Given this important qualification, the similar patterns of error occurrence for narrat ive and expository writing may be attributed to several factors. First, as Carlisle (1996) points out, the fact that children in these grades were likely still Â“orchestratingÂ” the writi ng process (p. 70) for both genres may have contributed to the absence of differences. Even as late as the middle school years, marked individual differences exist in compositional fluency, or the rate of producing text, which is an important aspect of text generation (B erninger et al., 1996). In theory, the more fluent or more productive is the composition, the more opportunities that may exist to
63 misspell, at least for typically developing children who have progressed in their integration of the phonological, orthographi c, and morphological aspects of writing. A second factor that may account for the unexpected outcome is the fact that children may have avoided words that they were less certain about how to spell. Apel et al. (2004b) discuss the strengths and limitations of prompted writing samples. On the one hand, writing samples are the most representative measure of childrenÂ’s ability to spell in that spelling must be coordina ted with writing demands (e.g., genre, content, etc.). On the other hand, children tend to select words that are highly familiar rather than less familiar words. Another limitation of prompted writi ng samples is the lack of consensus on what comprises a representative example of writing ab ility, an issue that is not easily separated from the ability to spell. Finally, the absence of differences in errors between the two genres may be attributed to a combination of the prompts and the time allocated to produce a product. The narrative frame was Â“One day (fill in a person) had the (b est) (worst) day in school,Â” while the expository frame consisted of Â“I like (someone, someplace, or something) becauseÂ” (Berninger et al., 1992, p. 264). A total of 5 minutes was given for each. It is possible that a lack of interest in one or both topics and the short time frame for completion yielded spelling outcomes that were similar across the two genres. Question 3: Error Patterns within Grade Level The third question concerned the analysis of the spelling errors in relation to whether specific error patterns occurred as a result of grade level. Two-way MANOVAs comparing grade and error type were comp leted for each major error category, including errors in each category that decreased as grade level increased.
64 The MANOVA for phonological error types revealed a significant interaction between error type and grade level for flap s (PFLP), silent /e/ (PSE), final position voicing (PFPV), and sonorant clusters (PS ON). As described by Treiman, Cassar, and Zukowski (1994) and Bourassa et al. (in press), phonological e rrors in the flap, as in liddle for little would be expected to be higher in young children who are attempting to spell phonetically. Due to th is phonetic attempt at spelling flaps, young children showed a bias for using /d/ for /t/ for the flap due to the voiced sound heard when orally decoding the word or simply pronouncing flappe d words in everyday conversation. The current studyÂ’s finding for the PFLP erro r type was thus consistent with the findings from Treiman et al. ( 1994) and Bourassa et al. (in pr ess). Flap errors gradually improved across grade levels. Find ings were similar for the silent /e/ (PSE) error type. It seems likely that the same reason applies to the silent /e/ (PSE) rule as applied to the flap (PFLP) rule. Because this sound is not pronounced when decoding the word, younger children will be less likely to include it in a spelling. Orthographic errors, in particular letter name (OLN) and vocalic /r/ (OVr), also differed across grade level. The number of these orthographic errors significantly decreased from grade 1 to grade 4, consistent with the finding of Reece and Treiman (2001), who found more orthographic spellings as children developed early spelling skills. This finding may suggest that the ch ildren improved in orthographic skills from first to fourth grade possibly due to incr eased experience with reading and writing. The orthographic error findings also correla ted with the grade level expectations. According to Bourassa and Treiman (2001), letter name spellings should reduce over time as a result of increased experience with reading and writing and through formal
65 instruction. This erro r occurs because children have not yet developed an understanding of phoneme unit and, therefore, spell the unit with the single letter name. For example, young children would spell the word elephant as lefit or far as fr (Bourassa & Treiman, 2001). The results of the analysis for morphol ogical errors also revealed a significant interaction between error type and grade level. However, unlike the phonological and orthographic errors, this categ ory did not reveal a decreasing pattern for the number of occurrences across grade levels. For the homonym (MHOM) error type, the number of error occurrences increased from first to f ourth grades. However, for the inflectional morphology (MINF) error type, the grade 1 childr en actually made mo re errors than did the grade 4 children, a finding cons istent with Bourassa et al. (in press) and Green et al. (2003). This may suggest that the fourth grader s had more experience with this specific feature, whereas the grade 1 children had less experience. Thus, the first graders were attempting to include inflectional markers but misspelling them either due to the lack of direct teaching combined with their relian ce on phonetic (invented) spellings (Hauerwas & Walker, 2003). Although significant differences in specifi c inflectional morphological errors were reported, these errors increased with increased grade level. An increase in the morphological error MINF (inflected forms) contradicts the anticipated outcomes as described by Green et al. (2003) and Bourassa et al. (in pres s). In both of these studies, inflected morphological forms were found to gradually decrease as the children progressed academically. While it was expected that the oldest children in this sample
66 (4th graders) would be more like ly to utilize inflectional mark ers in their writing, it was not anticipated that this age group would generate the most errors. In regard to homonym errors, Simon & Simon (1973) suggested that children were likely to understand that two words were phonemically similar yet fail to infer the specific linguistic context in which to apply the appropriate form. It is possible that, although children in grade 4 implicitly unders tood phonemic similarity or the concept of homonyms, their errors may relate to minima l awareness of the semantic contexts that oblige the selection of one spelli ng form over another (Plessas, 1963). Comparisons of specific error types and grade level revealed significant interactions for phonology, orthography, a nd morphology. These results for phonology and orthography were consistent with previous research findings. Children in the younger grades would be expected to make more errors in these categories. Research findings also revealed that morphological e rrors should decrease as child ren increase in grade level. However, the findings of the current study c ontradicted prior research. Older children (4th graders) made significantly more mor phological errors than did younger children. Question 4: Qualitative analysis of Features within Phonological, Orthographic, and Morphological categories. The qualitative analysis revealed common patterns of spelling errors across groups of children and the most commonly used error codes. When considering the most commonly misspelled words, one might also consider whether these words are high frequency words in English. Since the words us ed in the writing samples were selected based on the childÂ’s vocabulary, a realistic conclusion is that children used familiar words acquired through conversational interactions especially in th e case of the younger
67 children who may not have read at all or were in the early stages of learning to read. Because children may not yet have a consolidat ed visual or orthographic representation of all of the words they accessed in their wri ting samples, it is expected that many errors resulted from pronunciation (how a word Â“soundsÂ”), resulting in a misspelling. With the previous information in mind, the five words most ofte n in error were: a) because spelled as becaus b) recess spelled as reses c) favorite spelled as favrit d) friend spelled as frend and e) when spelled as wen Interestingly, all four grade levels exhibited a common error pattern with recess and friend The first, second, and third graders commonly misspelled the words because favorite and when Treiman and Bourassa (2001) discuss how young spellers ofte n use letter name misspellings to represent liquid phonemes b ecause children have not yet developed an appreciation of the phoneme unit. This was evident in the spelling of favrit for favorite In this case, it seems as if children depended on letter-tosound knowledge to spell the word, not yet understanding that the phoneme sequence Â–or represented a unit. It is also possible that the children spelled the wo rd according to their pronunciation schema, disregarding the orthographic f eatures of the word. It also may be appropriate to assume, based on the incorrect spellings favrit and becaus, that children as old as grade 3 may delete final position silent /e/. However, th e grade 4 children did not exhibit either of these error features. Based on these findings, tw o conclusions are feasible. First, by grade 4, at least in this sample, children have a firm understanding of the silent /e/ pattern. Alternately, these two words, commonly misspe lled by children in grades 1 to 3, were high frequency words, and the fourth grader s had more adequate spelling experiences with high frequency words.
68 All four grade levels demonstrated the same misspellings for the words recess and friend When compared to the previously discus sed errors, it appear s that, although these words are commonly occurring English meani ngs, especially among school age children, they are more difficult words to master due to their less regular spellings. In this situation, pronunciation would not le ad to the correct spellings. Finally, it is interesting to observe th e pattern that existed between the most commonly misspelled words and the most co mmonly applied error codes. The most commonly misspelled words across children co ntained errors for the most commonly used error codes. For ex ample, the misspellings becaus and favrit both contain errors coded as OLN (orthographic Â– letter name) and PSE (phonological Â– silent /e/). The misspelling reses contains an error coded as ODI (orthographic Â– digraph) and frend contains the error code OUVP (orthographic Â– unusual vowel pattern). These error codes are among the eight most commonly occurring codes for all 800 of the writing samples. These results have important implica tions for improved spelling intervention. Because these words occurred most frequently in the childrenÂ’s writing, and these error codes occurred most frequently for all of th e misspelled words, an important goal is to place increased emphasis on the teaching of patte rns. Since the same errors were seen across all grade levels, it is insufficient to assume that these children simply have not been taught or have not fully grasped the c oncept. Considering the five most commonly misspelled words in this study alone, teaching one spelling pattern, such as when to place a silent /e/ at the end of a word, could ge neralize to correct at least two different commonly misspelled words.
69 Study Strengths and Limitations Strengths The strengths of the stud y are two-fold. Expanding the previously designed POMAS scoring system (Silliman et al., 2006) resulted in a new way of analyzing the misspellings of children. The misspellings were coded to determine the quality of the errors using specific error feat ures evident in writing. The information provided from this study with this expanded scoring system adds to the limited research on qualitative differences in spelling development. Previous research has focused on differences in the total number of errors evident in spelling an alyses (Bruck & Waters, 1988; Hauerwas & Walker, 2003), however, this approach does not necessarily contri bute to explaining patterns of errors and why these errors occur. The quantitative results revealed differences between groups based purely on the number of error occurrences, but these data were insufficient to determine si milarities in misspellings patterns. Strengths were also evident in the desi gn of the study. The participants comprised a large and generally representative sample of typically devel oping children selected from early elementary grade levels. As a re sult, qualitative advances in patterns of spelling development could be discerned. In contrast, other studies, while providing snapshots of spelling development, have focused on the comparison of typically developing children or adolescen ts with those experiencing either significant reading difficulties or language learning problems (e.g, Bruck and Waters, (1988); Carlisle, (1987), Hauerwas and Walker (2003)). Althoug h there was variability in the predicted outcomes for the categorization of spelling er rors, general patterns of development were discovered, providing valuable educational information for improved instruction, as well
70 as a potential direction for i nvestigating spelling as a di agnostic marker of subtle language impairment after grade 2 (Catts, Hogan, & Adlof, 2005). Limitations At least four factors may have impacted the results. The first variable affecting outcomes is that the spelling words were collected from writing samples from previous studies. Interpretations of the actual words children were attempting to spell depended therefore, on the previous examinerÂ’s discretion. Although inter-rater agreement was determined for the studies, discrepancies exis ted in the spelling data for grades 1-3. In fact, there were several instan ces in which several different words were written next to a misspelling because the examiner could not dist inguish the word based on the meaning or several words were appropriate for the mean ing. The same situation arose for the 4th grade children. Inter-rater agreement revealed discrepancies between examiners in that the agreement for the narrative writing task was 75 percent and the expository task was 60 percent. A second limitation was associated with th e absence of the actual writing sample. The influence of other words on the misspelled word could not be analyzed. For example, a word, such as bear may influence a misspelling of the word their as thear due to the vowel pattern. In this case, the child ma y have used knowledge of the spelling of bear to assist in spelling the word their even though the vowel spelli ng resulted in a spelling error. A third factor that may have influenced the outcomes concerned the sample composition. Although children were described as typically developing, it is feasible that children with more subtle dyslexia and/or LLD were included. For example, it would be
71 expected that the children w ith dyslexia would produce more spelling errors (Bourassa et al., in press); therefore, the possibility cannot be ruled out that one grade level may have had more children who were at the lower e nd of the Â“normalÂ” spectrum than another grade. This consideration may explain why th e fourth graders had a greater increase in orthographic errors than did th e lower grade levels. If the f ourth grade sample did have a greater number of Â“strugglingÂ” students, the result may have been an inflated number of errors (Simon & Simon, 1973) in the orthographic category. A final factor possibly affecting the resu lts is the fact that the misspelled words were collected from prompted writing samples ra ther than from a dictated spelling list. In this case, children could avoid using words th at they found challenging to spell. Word selection could increase or decrease the number and type of errors collected from sample Clinical and Educa tional Implications Clinical Implications The POMAS results may offer new opportunities for determining if a child is demonstrating difficulties characteristic of a LLD, particularly in grade 2 and beyond. This version of the POMAS was designed ba sed on 60 error codes. However, when completing the statistical analysis for all 400 participants, only 12 of the 60 codes were found to occur beyond chance levels. These e rrors were common am ong the children. Category patterns for phonology and orthography revealed that, as the grade levels increased from grade 1 to grade 4, the error occurrences in these categories decreased. The other error codes included in the POMAS o ccurred intermittently, with no significant pattern emerging. This evidence suggests the types of errors typi cally expected in childrenÂ’s spellings from grades 1-4. Thus, ch ildren who display significant numbers of
72 errors both in the common error categories an d across the span of the other 48 errors, may be exhibiting difficulties beyon d what is considered typical. The findings of the Silliman et al. (2006) study, which qualitatively compared spellings of children with a LLD with ch ronologically (CA) and spelling age (SA) matched groups, provided insight into errors th at were typical of these groups. The LLD group demonstrated similar numbers of featur e errors across all th ree major linguistic categories. In addition, when comparing the spelling errors of the three groups, the LLD group was found to have errors least similar to the correct spelling. The LLD group also exhibited significant difficulties represen ting inflected and derived morphological meanings. These findings in conjunction with the results from th e current study provide strong evidence for typical and untypical erro r patterns. With careful consideration of spelling performance using a qualitative scoring system, such as the POMAS, speechlanguage pathologists can determ ine patterns of development. Educational Implication Qualitative findings for the spellings of these young elementary-age children are also beneficial for educators. The quality of spelling instruction in too many American schools appears questionable. Because spelli ng impacts on reading and writing abilities (Apel & Masterson, 2001; Berninger et al., 2006 ; Goswami, 1992), the findings from this study may be a starting point for supplying t eachers with the interv ention strategies to adequately assess and teach spelling skills. Based on the findings of the most commonl y misspelled words in this study, it would be appropriate to encourage teacher s to target word lists containing common linguistic patterns. It has been implied that teachers often target story vocabulary that
73 does not relate linguistically (Apel et al., 2004a). Therefore, it would be beneficial to target vocabulary words classified as prim arily in the phonological category, such as short versus long vowels. Targ eting sight words with these similar patterns (many of which were evident in this study) would also lead to improved spelling abilities and the increased probability of applic ation to more fluent reading. Increasing awareness of spelling instructi on may also assist teachers in improving reading skills in their classroom. Knowi ng where a child experiences a breakdown in spelling skills may also assist in explaining r eading difficulties, and thus assist teachers in more accurate assignment of children to r eading groups. For example, reading skills, much like spelling, require the integra tion of phonologic, orthographic, and morphological knowledge. Children who dem onstrate phonologic and orthographic errors in their spellings will likely struggle with phonemic segmentation and identifying sight words automatically, which, in turn, affects the ease and accuracy of oral reading fluency. Understanding this information will assist in adequately identifying children according to those needing continued phonologi cal awareness and decoding instruction versus those who are at more advanced levels. An increased understanding of error types Â“typicalÂ” of children at specific grade levels helps to increase the educational valu e of spelling instruction. Increased teacher understanding about what can be Â“expected Â” will decrease the number of children Â“falling through the cracksÂ” in spelling develo pment, and allow teachers to teach spelling patterns. As suggested by Templeton (2004) teachers should avoi d teaching rules for misspelling and focus, instead, on the spelling knowledge that the child demonstrates but confuses. For example, based on the spelling erro rs evident in the writing samples for this
74 study, if the child has several words in which the long vowel in a word is misused, the teacher can examine the sample for instances in which it was used correctly. If there are instances of correct long vowel spellings, the teacher should focus on teaching accuracy for the long vowel rather than focusing strict ly on a spelling skill that was never used correctly by the child. This st rategy should not only lead to improvements in spelling skills, but also in writing and reading deve lopment because the child will be able to identify relationships between words with sim ilar patterns. In othe r words, if taught a pattern regarding long vowels, the child could apply this pattern to other words that may have originally o ccurred in error. To conclude, this study revealed valu able information regarding the spelling development of typically developing elementa ry school children. Because the children were compared according to error types, patter ns of spelling skills were revealed for each grade level. Thus, typical patterns of spelling development were hypothesized. Regardless of the inability to refer to the actual narratives from which the spelling errors were extracted, results from coding the lis tings of misspelled words revealed the progression of spelling development, whic h has potential import for both speechlanguage pathologists and teachers. These pa tterns achieved the goal of identifying the boundaries of typical patterns for a specific gr ade level. Awareness of these boundaries of typical variation should then lead to more individualized and effective intervention strategies in the classroom.
75 References Apel, K., & Masterson, J. J. (2001). Theoryguided spelling assessment and intervention: A case study. Language, Speech, and Hearing Services in Schools, 32 182-195. Apel, K., Masterson, J. J., & Hart, P. (2004a ). Integration of la nguage components in spelling: Instruction that maximizes st udents' learning. In L. Wilkinson (Ed.), Language and literacy learning in schools (pp. 292-318). New York: The Guilford Press. Apel, K., Masterson, J. J., & Niessen, N. L. (2004b). Spelling assessment frameworks. In B. J. Ehren (Ed.), Handbook of language and li teracy: Development and disorders (pp. 644-660). New York: The Guilford Press. Bear, D., Invernizzi, M., Templeton, S., & Johnston, F. (2004). Words their way: Word study for phonics, vocabulary, and spelling instruction Upper Saddle River, New Jersey: Pearson Education, Inc. Berninger, V., Abbott, R., Jones, J., Wo lf, B. J., Gould, L., Anderson-Youngstrom, Shimada, S., & Apel, K. (2006). Earl y development of language by hand: Composing, reading, listening, and speak ing connections, three letter-writing modes, and fast mapping in spelling. Developmental Neuropsychology, 29 61-92. Berninger, V., Cartwright, A. C., Yates, C. M., Swanson, H. L., & Abbott, R. D. (1994). Developmental skills related to writing and reading acquisition in the intermediate grades. Reading and Writing: An Interdisciplinary Journal, 6 161-196.
76 Berninger, V., Vaughan, K., Abbott, R. D., Br ooks, A., Abbott, S. P., Rogan, L., Reed, E., & Graham, S. (1998). Early intervention for spelling problems: Teaching functional spelling units of varying size with a multiple-connections framework. Journal of Educational Psychology, 4 587-605. Berninger, V., Whitaker, D., Feng, Y., Swanson, H. L., & Abbott, R. D. (1996). Assessment of planning, translating, a nd revising in junior high writers. Journal of School Psychology, 34 23-52. Berninger, V., Yates, C., Cartwright, A., Ru tberg, J., Remy, E., & Abbott, R. D. (1992). Lower-level developmental skills in beginning writing. Reading and Writing: An Interdisciplinary Journal, 4 257-280. Bourassa, D., & Treiman, R. (2001). Spelling development and disability: The importance of linguistic factors. Language, Speech, and Hearing Services in Schools, 32 172-181. Bourassa, D., Treiman, R., & Kessler, B. (i n press). Use of morphology in spelling by children with dyslexia and t ypically developing children. Memory and Cognition Bruck, M., & Waters, G. (1988). An analysis of the spelling errors of children who differ in their reading and spelling skills. Applied Psycholinguistics, 9 77-92. Carlisle, J. (1987). The use of morphologica l knowledge in spelling derived forms by learning-disabled and normal students. Annals of Dyslexia, 9 247-266. Carlisle, J. (1988). Knowledge of derivational morphology and spelling ability in fourth, sixth, and eighth graders. Applied Psycholinguistics, 9 247-266. Carlisle, J. (1996). An exploratory study of morphological errors in children's written stories. Reading and Writing: An Interdisciplinary Journal, 8 61-73.
77 Carlisle, J. (2003). Morphology matters in learning to read: A commentary. Reading Psychology, 24 291-322. Cassar, M., & Treiman, R. (1997). The beginni ngs of orthographic knowledge: Children's knowledge of double letters in words. Journal of Educational Psychology, 89 631-644. Catts, H. W., Hogan, T. P., & Adlof, S. M. (2005). Developmental changes in reading and reading disabilities. In A. G. Kamhi (Ed.), The connections between language and reading disabilities (pp. 25-40). Malwah, NJ: Erlbaum. Catts, H. W., Hogan, T. P., & Fey, M. C. (2003). Subgrouping poor readers on the basis of individual differences in reading related abilities. Journal of Learning Disabilities, 36 151-164. Dodd, B., & Carr, A. (2003). Young children's letter-sound knowledge. Language, Speech, and Hearing Services in Schools, 34 128-137. Ehri, L. C. (1986). Sources of difficulty in le arning to spell and read. In D. Routh (Ed.), Advances in developmental and behavioral pediatrics (Vol. 7, pp. 121-195). Greenwich, CT: JAI. Ehri, L. C., & Snowling, M. J. (2004). Deve lopmental variation in word recognition. In K. Apel (Ed.), Handbook of language and literacy: Development and disorders (pp. 433-460). New York: The Guilford Press. Gentry, J. R. (1982). An analysis of developmental spelling in GNYS AT WRK. The Reading Teacher, 36 192-200. Gentry, J. R. (2004). The science of spelling Portsmouth, New Hampshire: Heinemann.
78 Goswami, U. (1992). Phonological f actors in spelling development. Journal of Child Psychology and Psychiatry, 33 967-975. Graham, S., Harris, K. R., & Chorzempa, B. (2002). Contribution of spelling instruction to the spelling, writing, and reading of poor spellers. Journal of Educational Psychology, 94 669-686. Green, L., McCutchen, D., Schwiebert, C ., Quinlan, T., Eva-Wood, A., & Juelis, J. (2003). Morphological development in children's writing. Journal of Educational Psychology, 95 752-761. Hauerwas, L., & Walker, J. (2003). Spelling of inflected verb morphology in children with spelling deficits. Learning Disabilities Research & Practice, 18 25-35. Henderson, E. (1985). Teaching spelling Boston: Houghton Mifflin. Jastak, S., & Wilkinson, G. (1984). Wide range achievement test-revised Wilmington, DE: Jastak Associates. Leybaert, J., & Lechat, J. (2001). Variability in deaf children's spelling: The effect in language experience. Journal of Educational Psychology, 93 554-562. Masterson, J. J., & Crede, L. A. (1999). Lear ning to spell: Implica tions for assessment and intervention. Language, Speech, and Hearing Services in Schools, 30 243254. Miller, J., & Chapman, R. (1991). Systematic analysis of language transcripts [Computer program; A. Nockerts, Programmer]. Madison, WI: Language Analysis Laboratory, Waisman Center on Mental Retardation and Human Development.
79 Nagy, W., Berninger, V., Abbott, R., Vaughn, K., & Vermeulen, K. (2003). Relationship of morphology and other language skills to l iteracy skills in at-risk second-grade readers and at-risk fourth-grade writers. Journal of Educational Psychology, 95 730-742. Nunes, T., Bryant, P., & Bindman, M. (1997). Morphological sp elling strategies: Developmental stages and processes. Journal of Developmental Psychology, 33 637-649. Paul, R. (2001). Language disorders from infancy through adolescence: Assessment & intervention (2nd ed.). St. Louis: Mosby. Plessas, S. (1963). Children's errors in spelling homonyms. The Elementary School Journal 163-168. Reece, C., & Treiman, R. (2001). Children's spelling of syllabic /r/ and letter-name vowels: Broadening the study of spelling development. Applied Psycholinguistics, 22 139-165. Salvia, J., & Ysseldyke, J. E. (2001). Assessment in special an d remedial education (8th ed.). Boston: Houghton Mifflin. Scott, C. M., & Windsor, J. (2000). General language performance measures in spoken and written narrative and expository discourse of school-age children with language learning disabilities. Journal of Speech, Language, and Hearing Research, 43 324-339. Silliman, E. R., Bahr, R. H ., & Peters, M. L. (2006). Spelling patterns in preadolescents with atypical language skills: Phonological, morphological and orthographic factors. Developmental Neuropsychology, 29 93-123.
80 Simon, D., & Simon, H. (1973). Alternative uses of phonemic information in spelling. Review of Educational Research, 43 115-137. Singer, B., & Bashir, A. (2004). Developmenta l variations in writing composition skills. In K. Apel (Ed.), Handbook of language and literacy: Development and disorders (pp. 559-582). New York: The Guilford Press. Small, L. H. (1999). Fundamentals of phonetics: A practical guide for students Boston: Allyn and Bacon. Sulzby, E. (1996). Roles of oral and written lang uage as children approach literacy. In L. B. Resnick (Ed.), Children's early text construction (pp. 25-46). Mahwah, NJ: Erlbaum. Templeton, S. (2004). Instructional approach ed to spelling: The window on student's word knowledge in reading and wr iting. In L. Wilkinson (Ed.), Language and literacy learning in schools (pp. 273-291). New York: The Guilford Press. Treiman, R. (1991). Children's spelling erro rs on syllable-initial consonant clusters. Journal of Educational Psychology, 83 346-360. Treiman, R., & Bourassa, D. (2000a). The development of spelling skills. Topics in Language Disorders, 20 (3), 1-18. Treiman, R., & Bourassa, D. (2000b). Ch ildren's written and oral spelling. Applied Psycholinguistics, 21 183-204. Treiman, R., & Cassar, M. (1997). Spelling acq uisition in English. In M. Fayol (Ed.), Learning to spell: Research, th eory, and practice across languages (pp. 61-80). Mahwah, NJ: Lawrence Erlbaum Associates.
81 Treiman, R., Cassar, M., & Zukowski, A. (1994) What types of linguistic information do children use in spelling? The case of flaps. Child Development, 65 1318-1337.
83 Appendix A Â– Coding System Table A.1 Feature Errors. Category Code Description Example P PCD Consonant deletion beame | became P PCR Cluster reduction stuck | struck P PDIP True diphthongs vs Long vowels arond | around P PDV Devoicing pusels | puzzles P PECL 2&3 element clusters seet | street P PEP Epenthesis tolid | told P PFCD Final consonant deletion kee | keep P PFLP Flaps pride | pretty P PFPV Final position voicing becus | because P PFR Fronting graphits | graphics P PLV Long vowel error roop | rope P PNE Nasal error junp | jump P PSC /s/ clusters bes | best P PSE Silent /e/ Pale| pal P PSHW Schwa cristle | crystal P PSON Sonorant clusters (nasals, l, r, j) ad | and P PSR Syllable reduction maroni | macaroni P PSRS Schwa reduced syllable anmols | animals P PST Stopping teel | feel
84 Appendix A (Continued) Table A.1 (Continued) P PSV Short vowels kite | kit P PVO Voicing blay | play P PVOCR Vocalic /r/ cos | curls O OAA Apostrophe added getÂ’s | gets O OCD Consonant doubling terriffic | terrific O OCE Consonant error sogt | soft O OCL Capital letter california | California O ODI Digraphs sip | ship O OHY Hyphen fortytwo | forty-two O OHSV c/k Â– hard and soft velars Mace| make O OLD Letter doubling (syllable juncture) triped | tripped O OLN Letter name (l,s,r) cr | car O OLR Letter reversal (b /d, d/b) balls | dolls O OLS Letter sound (c/k, k/c, etc.) sereal | cereal O OLVP Long vowel pattern keep | kipe O OLWF Long vowel word families (Â“-oldÂ”, Â“ighÂ”) nite | night O ONA No apostrophe somebodys | somebodyÂ’s O OOW One word some times | sometimes O OPA Phoneme addition sradr | grade
85 Appendix A (Continued) Table A.1 (Continued) O OPE Plural error fris | fries O OSJ Syllable juncture Â– y to i cryed | cried O OSL Silent letter /h/ (w here, what, when) wen | when O OSY Syllabic /l/ terdals | turtles O OUVP Unusual vowel pattern cof | cough O OVE Vowel error stuped | stupid O OVr Vocalic /r/ (r/er, etc.) sistr | sister O OWB Word boundary (2 sep. wo rds) eachother | each other M MDER Derivation (root word) depasition | deposition M MDVM Derivational morphology brang | brought M MHOM Homonyms there | their M MINF Inflectional morphology bike | bikes M MPRE Prefixes Organize| reorganize M MSH Shifts Â– phonological change magishen | magician M MSUF Suffixes normal | normally PO POR Reversals tis | its PO POVDS Vowel dependent spellings (short vowels Â– tch, dge, ck/ch, ge) Baitch| batch PO POVM Vowels missing/deleted dble | double MO MCON Contraction wasnt | wasnÂ’t
86 Appendix A (Continued) Table A.1 (Continued) MO MOSP Mispelled root word resulting in phonologically-accurate spelling edgeucation | education MO MOV Overgeneralization losted | lost MP MPVS Visually similar erro r (i.e. are for car) are | car CQ Child started the word but failed to finish b | buy
87 Appendix B Â– t-tests for Grade Level Effects on Number and Types of Errors Table B.1 Phonology Comparisons Across Grade Levels Comparison t Degrees of freedom (df) p 1st compared to 2nd 3.766 396 .000 1st compared to 3rd 2.252 400 .025 1st compared to 4th 3.840 398 .000 2nd compared to 3rd -1.245 398 .214 2nd compared to 4th .121 396 .904 3rd compared to 4th 1.346 400 .179 Table B.2 Orthographic Comparisons Across Grade Levels Comparison t Degrees of freedom (df) p 1st compared to 2nd -2.515 396 .012 1st compared to 3rd .631 400 .528 1st compared to 4th -.744 398 .457 2nd compared to 3rd 2.886 398 .004 2nd compared to 4th 1.554 396 .121 3rd compared to 4th -1.249 400 .212
88 Appendix B (Continued) Table B.3 Morphological Comparisons Across Grade Levels Comparison t Degrees of freedom (df) p 1st compared to 2nd .460 396 .646 1st compared to 3rd -1.647 400 .100 1st compared to 4th -2.871 398 .004 2nd compared to 3rd -2.056 398 .040 2nd compared to 4th -3.240 396 .001 3rd compared to 4th -1.277 400 .202
89 Appendix C Â– Post Hoc Comparisons for Error Patterns across Grade Level Table C.1 Bonferroni Post Hoc Testi ng Results for Phonological Error Type Comparisons per Grade Level Comparison of Grade Levels Mean Difference Between Grades Std. Error p value 1 compared to 2 3 4 .13 .16 .21 .023 .023 .023 .000 .000 .000 2 compared to 1 3 4 -.13 .03 .08 .023 .023 .023 .000 1.000 .004 3 compared to 1 2 4 -.16 -.03 .06 .023 .023 .023 .000 1.000 .110 4 compared to 1 2 3 -.21 -.08 -.06 .023 .023 .023 .000 .004 .110
90 Appendix C (Continued) Table C.2 Bonferroni Post Hoc Testi ng Results for Orthographic Error Type Comparisons per Grade Level Comparison of grade levels Mean Difference between grades Std. Error p Value 1 compared to 2 3 4 .14 .23 .28 .039 .039 .039 .002 .000 .000 2 compared to 1 3 4 -.14 .08 .14 .039 .039 .039 .002 .180 .003 3 compared to 1 2 4 -.23 -.08 .05 .039 .039 .039 .000 .180 1.000 4 compared to 1 2 3 -.28 -.14 -.05 .039 .039 .039 .000 .003 1.000
91 Appendix D Â– Comparisons of Specific E rror Types Across Grade Level Table D.1 Comparison of Error Type PFLP Between Grade Levels. Error Type PFLP t Degrees of Freedom (df) p 1st compared to 2nd 1.654 396 .099 1st compared to 3rd 2.212 400 .028 1st compared to 4th 2.758 398 .006 2nd compared to 3rd .620 398 .536 2nd compared to 4th 1.297 396 .195 3rd compared to 4th .699 400 .485 Table D.2 Comparison of Error Type PFPV Between Grade Levels. Error Type PFPV t Degrees of Freedom (df) p 1st compared to 2nd 3.630 396 .000 1st compared to 3rd 3.955 400 .000 1st compared to 4th 3.935 398 .000 2nd compared to 3rd 1.010 398 .313 2nd compared to 4th 1.005 396 .315 3rd compared to 4th
92 Appendix D (Continued) Table D.3 Comparison of Error Type PSE Between Grade Levels. Error Type PSE t Degrees of Freedom (df) p 1st compared to 2nd 2.872 396 .004 1st compared to 3rd 3.072 400 .002 1st compared to 4th 7.048 398 .000 2nd compared to 3rd .499 398 .618 2nd compared to 4th 4.391 396 .000 3rd compared to 4th 3.148 400 .002 Table D.4 Comparison of Error Type PSON Between Grade Levels. Error Type PSON t Degrees of Freedom (df) p 1st compared to 2nd 3.368 396 .001 1st compared to 3rd 4.418 400 .000 1st compared to 4th 4.565 398 .000 2nd compared to 3rd 1.191 398 .235 2nd compared to 4th 1.333 396 .183 3rd compared to 4th .110 400 .912
93 Appendix D (Continued) Table D.5 Comparison of Error Type ODI Between Grade Levels. Error Type ODI t Degrees of Freedom (df) p 1st compared to 2nd .225 396 .822 1st compared to 3rd 1.393 400 .164 1st compared to 4th 2.435 398 .015 2nd compared to 3rd 1.035 398 .301 2nd compared to 4th 1.982 396 .048 3rd compared to 4th 1.146 400 .252 Table D.6 Comparison of Error Type OLN Between Grade Levels. Error Type OLN t Degrees of Freedom (df) p 1st compared to 2nd 4.380 396 .000 1st compared to 3rd 5.519 400 .000 1st compared to 4th 6.884 398 .000 2nd compared to 3rd 1.373 398 .171 2nd compared to 4th 2.997 396 .003 3rd compared to 4th 1.556 400 .121
94 Appendix D (Continued) Table D.7 Comparison of Error Type OLR Between Grade Levels. Error Type OLR t Degrees of Freedom (df) p 1st compared to 2nd 1.818 396 .070 1st compared to 3rd 1.986 400 .048 1st compared to 4th 3.114 398 .002 2nd compared to 3rd .237 398 .813 2nd compared to 4th 1.748 396 .081 3rd compared to 4th 1.417 400 .157 Table D.8 Comparison of Error Type OLS Between Grade Levels. Error Type OLS T Degrees of Freedom (df) p 1st compared to 2nd .544 396 .586 1st compared to 3rd 2.012 400 .045 1st compared to 4th 2.571 398 .011 2nd compared to 3rd 1.575 398 .116 2nd compared to 4th 2.191 396 .029 3rd compared to 4th .677 400 .499
95 Appendix D (Continued) Table D.9 Comparison of Error Type OVr Between Grade Levels. Error Type OVr T Degrees of Freedom (df) p 1st compared to 2nd 2.083 396 .038 1st compared to 3rd 4.733 400 .000 1st compared to 4th 4.811 398 .000 2nd compared to 3rd 2.711 398 .007 2nd compared to 4th 2.832 396 .005 3rd compared to 4th .217 400 .828 Table D.10 Comparison of Error Type MCON Between Grade Levels. Error Type MCON t Degrees of Freedom (df) p 1st compared to 2nd 1.349 396 .178 1st compared to 3rd 1.390 400 .165 1st compared to 4th -.745 398 .457 2nd compared to 3rd .035 398 .972 2nd compared to 4th -1.776 396 .077 3rd compared to 4th -1.813 400 .071
96 Appendix D (Continued) Table D.11 Comparison of Error Type MHOM Between Grade Levels. Error Type Â– MHOM t Degrees of Freedom (df) p 1st compared to 2nd -.467 396 .641 1st compared to 3rd -3.300 400 .001 1st compared to 4th -3.730 398 .000 2nd compared to 3rd -2.848 398 .005 2nd compared to 4th -3.297 396 .001 3rd compared to 4th -.549 400 .583 Table D.12 Comparison of Error Type MINF Between Grade Levels. Error Type MINF t Degrees of Freedom (df) p 1st compared to 2nd 2.292 396 .022 1st compared to 3rd 3.439 400 .001 1st compared to 4th .941 398 .347 2nd compared to 3rd 1.126 398 .261 2nd compared to 4th -1.297 396 .196 3rd compared to 4th -2.394 400 .017