UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The use of phonological and orthographic information for memory and spelling : an analysis of reading… Harrison, Gina Louise 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-714667.pdf [ 4.87MB ]
Metadata
JSON: 831-1.0054604.json
JSON-LD: 831-1.0054604-ld.json
RDF/XML (Pretty): 831-1.0054604-rdf.xml
RDF/JSON: 831-1.0054604-rdf.json
Turtle: 831-1.0054604-turtle.txt
N-Triples: 831-1.0054604-rdf-ntriples.txt
Original Record: 831-1.0054604-source.json
Full Text
831-1.0054604-fulltext.txt
Citation
831-1.0054604.ris

Full Text

THE USE OF PHONOLOGICAL AND ORTHOGRAPHIC INFORMATION FOR MEMORY AND SPELLING: AN ANALYSIS OF READING AND SPELLING SUBTYPES by GINA LOUISE HARRISON B.A., The University of British Columbia, 1992 M.A., The University of British Columbia, 1994 A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES [Department of Educational and Counselling Psychology and Special Education) We accept this dissertation as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September, 2001 © Gina Louise Harrison, 2001 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^A^jLC^ict^xJu-tCgQrt'bcilu^ Ib^ckalog^ ^SyczcAX**-The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT The present study was designed to examine differences between subtypes of readers and spellers in their performance on several phonological, orthographic, and memory tasks. A central question involved whether subtypes of readers and spellers could be distinguished based on their performance across the tasks administered. Based on their performance on a standardized achievement test, fourth and fifth grade children (N=50) were classified as having no difficulties with reading and spelling (good readers and spellers), difficulties with spelling, but not reading (mixed readers and spellers), or difficulties with both reading and spelling (poor readers and spellers). Each student was given a series of tasks to assess their use of phonological and orthographic information for memory and spelling. These tasks included: 1) rhyme judgment, 2) cued recall, 3) reading pronounceable pseudowords, 4) deciding which of. two pseudowords looks most like a real word, and 5) reporting on the kinds of strategies used to spell words. An error analysis was also conducted. Students with reading and spelling difficulties performed consistently lower than good and mixed readers and spellers on tasks assessing their use of phonological information. Good and mixed readers and spellers were not distinguishable on these tasks. Students with no reading and spelling difficulties or with spelling difficulties only performed better than poor readers and spellers on some tasks assessing orthographic processing. Specifically, mixed readers and spellers were distinguishable from good readers and spellers by their poorer recall of visually similar words. Good and poor subtypes were not distinguishable on this task. Poor readers and spellers also achieved comparable scores to the good and mixed readers and spellers on a measure of orthographic awareness. Overall results provided evidence supporting subtypes of reading and spelling ability groups. Students with no reading and spelling difficulties, or difficulties with spelling but not reading were similar in their use of phonological information. However, students with reading and spelling difficulties were more similar to the good readers and spellers in their use of orthographic information in memory. The findings from the present study have implications to subsequent research examining spelling ability, provide further evidence of the unique processing characteristics of the paradoxical good reader but poor speller, and suggest the possibility of unique programming needs to remediate spelling difficulties in mixed and poor readers and spellers. iii T A B L E O F CONTENTS ABSTRACT ii TABLE OF CONTENTS iii LIST OF TABLES : vi ACKNOWLEDGMENT : '. vii C H A P T E R 1 : INTRODUCTION 1 A. Rationale 1 B. Statement of the Problem and Overview of the Study 2 C. Significance of the Study 3 C H A P T E R 2: T H E SPELLING PROCESS 4 A. Introduction 4 B. Information Processing Models of Spelling 6 (1) Dual Routes: Orthography and Phonology 6 (2) Reciprocal Routes: Lexical Priming and Analogy 8 (3) Relationship Between Reading and Spelling Across Routes 10 C. Phonological and Orthographic Processes in Learning to Spell 13 (1) Stages of Emergence of Spelling Skills 13 (2) Beyond Stages: Phonology as the Framework for Orthography 15 D. Varieties of Spelling Competence : 17 (1) Analysis of Misspellings in the Composition of Subtypes 17 (2) Strategies and Processing Characteristics of Good and Poor Spellers 19 (3) Mixed Spellers: Good Readers but Poor Spellers 21 E. Phonological and Orthographic Codes in Memory 25 (1) Reading Disabilities and Phonological and Orthographic Codes in Memory .. 26 (2) Relationship Between Phonological Codes on Memory and Spelling Tasks .. 28 F. Summary of Background Research and Overview of the Present Study 30 G. Hypotheses ' 31 C H A P T E R 3: M E T H O D O L O G Y 32 A. Overview : 32 B. Participants 32 (1) ScreeningMeasures 33 a. WRAT-3 Reading Subtest 33 iv b. WRAT-3 Spelling Subtest 33 (2) Selection of Groups 33 C. Orthographic, Phonological, and Strategy Measures 34 (1) Rhyme Judgment Task 34 (2) Cued Recall Memory Task 35 (3) Pseudoword Reading Task " ,36 (4) Orthographic Awareness Task 37 (5) Strategy Reports 37 D. Error Analysis 38 (1) Phonological Similarity 38 (2) Visual/Orthographic Similarity ! ' 39 C H A P T E R 4: R E S U L T S 40 A. Overview . '• • • 40 B. Preliminary Results 40 C. Principal Analyses 42 (1) Rhyme Judgment Task : 42 (2) Cued Recall Memory Task 44 (3) Orthographic Awareness Task , •••• 46 (4) Pseudoword Reading Task 47 (5) Error Analysis : 47 (6) Verbal Report and Strategy Data 48 a. Correct Spellings ' 48 b. Misspellings '. 49 D. Summary • 50 C H A P T E R 5: SUMMARY AND CONCLUSIONS 53 A. Overview • • ••• 53 B. Discussion • '• • • 53 (1) The Use of Phonological Information by Reading and Spelling Subtypes ... 53 (2) The Use of Orthographic Information by Reading and Spelling Subtypes ... 53 C. Methodological Issues • 56 D. Directions for Future Research 58 E. Implications for Educational Assessment and Instruction 59 F. Conclusions • 60 REFERENCES : : 62 Appendix A: Parent Consent Form ..' 68 Appendix B: Cues and Targets Used in Rhyme Judgment and Cued Recall Tasks 70 Appendix C: Task Instructions: Rhyme Judgment '. 71 Appendix D: Task Instructions: Cued Recall Memory Task 72 Appendix E: Pseudoword Reading Word List 73 Appendix F: Orthographic Awareness Task Word List 74 Appendix G: Task Instructions: Strategy Report '. 75 Appendix H: Means and Standard Deviations on Dependent Measures for Gender 76 Appendix I: Means and Standard Deviations on Dependent Measures for Grade 79 vi LIST OF TABLES T A B L E 1 Initial Composition of Subtypes as a Function of Sex and Grade TABLE 2 Final Composition of Subtypes as a Function of Sex and Grade TABLE 3 Consistency of Teacher Nomination with Test Results TABLE 4 WRAT-3 Scores (percentiles) as a Function of Grade and Subtype TABLE 5 Means and Standard Deviations for Subtypes on Response Accuracy for the Rhyme Judgment Task T A B L E 6 Means and Standard Deviations for Subtypes on Recall Performance TABLE 7 Performance (total correct) on the Orthographic Awareness and Pseudoword Reading Tasks Across Subtypes T A B L E 8 Mean Percentages of Visual and Phonological Accuracy of Misspellings Across Subtypes T A B L E 9 Mean Percentages of Reports of Specific Strategies for Correct Spellings Across Subtypes T A B L E 10 Mean Percentage of Reports of Specific Strategies for Misspellings Across Subtypes vii ACKNOWLEDGMENT I wish to express my sincerest gratitude to the administrators, teachers, and students at the following elementary schools in Vancouver: University Hi l l , Carleton, Our Lady of Sorrows, and Norquay as well as the following elementary schools in Surrey: Royal Heights and Green Timbers, for their enthusiasm in welcoming me into their classrooms to collect data for this project. I would also like to thank Dr. Shelley Hymel for her insightful criticisms throughout the entire process of compiling this dissertation and for her academic guidance and support throughout my doctoral program. Dr. Linda Siegel and Dr. Jon Shapiro also provided me with valuable feedback for which I am grateful. I wish to acknowledge the University Graduate Fellowship that financially supported my doctoral training and this project. Finally, my heartfelt thanks goes to my family, particularly to my husband for his patience, encouragement, and support in helping me live my dreams. 1 C H A P T E R 1 INTRODUCTION A. Rationale Whether all students with spelling difficulties share similar processing characteristics is the key question motivating the present research. Two "routes" or sets of processes have been identified as central to spelling production-- phonological, related to letter-sound correspondences, and orthographic, related to the visual appearance of words (Barry, 1992, 1994; Seymour, 1992). These routes are separable, but interactive and reciprocal in the non-brain impaired population (Barry, 1992). In addition, spelling and reading are considered to have a symbiotic relations hip in literacy development each supporting and building from experiences with the other (Ehri, 2000; Snowling, 1994). For example, it is through their exposure to print that children associate speech sounds with symbolic albeit alphabetic forms. Children write spellings that become more phonologically and visually accurate the more exposure and practice they have with reading and spelling suggesting the gradual amalgamation, through print exposure, of phonological and orthographic processes (Snowling, 1994). • • < ? . • Even though the processes required for reading and spelling are related in important ways, it has also been suggested that spelling is a more demanding task than reading due to the "problem of asymmetry" (Bryant & Bradley, 1980) and the fact that there are many more (and inconsistent) sound-to-spelling patterns than spelling-to-sound relations. For example, we can read the word beef predictably since all words that end in -eef rhyme with beef. However, spelling the same word is less predictable because many words that rhyme with beef are spelled differently (e.g., leaf, chief) (Waters, Bruck, & Seidenberg, 1985). Differences have been found between students with and without reading and spelling difficulties in the way they use phonological and orthographic information in spelling (Bruck & Waters, 1988, 1990; Frith, 1980, 1985; Lennox & Siegel, 1993, 1994). For example, students who are good at reading and spelling or good at reading but not spelling appear to have better developed phonological processing skills than students who are poor at both reading and spelling (Frith, 1980; Lennox & Siegel, 1993, 1994). Moreover, poor readers and spellers appear to use more visual than phonological strategies in spelling and good readers but poor spellers seem to have difficulties remembering the visual appearance of words (Lennox & Siegel, 1993, 1994). There is also research that indicates that reading disabled students are more likely to remember orthographic than phonological information (Rack, 1985). Whether these memory processing differences can be replicated in groups differing in reading and spelling ability is the 2 main issue being addressed by the present study. If differences are evident, this research will augment the literature on spelling subtypes and support the distinctiveness of different subtypes when examining spelling difficulties. The findings from the present study may also provide evidence to educators that not all students with spelling difficulties are using the dual routes for spelling the same way, and hence may require different remedial strategies. B . Statement of the Problem and Overview of the Study Although there is strong research evidence indicating that good readers and spellers and good readers but poor spellers share comparable phonological processing skills (e.g., Frith, 1980; Lennox & Siegel, 1993, 1994), there is also evidence to suggest that the good reader but poor speller does not possess well developed phonological processing skills (Bruck & Waters, 1988, 1990; Waters, Bruck, & Seidenberg, 1985). Moreover, that poor readers and spellers exhibit better-developed orthographic than phonological skills has also been the subject of some controversy along with evidence that good readers and poor spellers experience difficulty with some aspects of remembering word patterns. The present study was therefore designed to examine how the three groups of readers and spellers would differ on tasks assessing their phonological and orthographic skills in memory and spelling. Of particular interest was whether findings could be replicated between past research and the present study- - providing more information about the processing characteristics of reading and spelling subtypes. For the purpose of examining the nature of phonological and orthographic processes in good and poor spellers who differ in reading decoding ability, several tasks thought to capture phonology and orthography were administered. Initially, students were identified as good at reading decoding and spelling (good), poor at reading decoding and spelling (poor) and good at reading decoding, but poor at spelling (mixed) by their classroom teachers. To verify these subtype classifications, these students then completed reading decoding and spelling dictation measures with subtype classification based on the decoding and spelling test scores. A misspelling analysis across phonological and orthographic criteria, based on Bruck and Waters (1988) criteria, was also conducted on the first 10 errors from each students' spelling test and verbal reports by students were collected while they completed a short spelling dictation task. Students in all three subtype groups completed a series of rhyme judgment and memory tasks adapted from Rack (1985) and Holligan and Johnston (1988) as well as tasks assessing orthographic awareness (Siegel, Share, & Geva, 1995) and pseudoword reading (Woodcock, 1987). 3 C. Significance of the Study There is some controversy over the ways in which subtypes of readers and spellers process information, and no research on the ways in which these subtypes differ in relation to the reliance on phonological and orthographic codes in memory. Moreover,, no research has been conducted to date that compares students' own reports of their spelling strategies with actual strategies of a phonological or orthographic nature inferred from cognitive tasks. The present research provides rich and converging evidence from multiple sources (i.e., direct and indirect measures) on the nature of three subtypes of readers and spellers and the kinds of strategies used to retrieve word spellings. The findings from this research provide insight into children's thought processes while they spell and have important practical implications for the focus and content of spelling instruction across ability subtypes. C H A P T E R 2 T H E SPELLING PROCESS A. Introduction Research on the cognitive processes involved in spelling as important to the development of reading and as a distinct area has increased rapidly over the last decade (Brown & Ellis, 1994). Although the processes involved in reading and spelling are related in important ways, such as knowledge of the sounds of letters in words (i.e., phonological awareness) and the ability to discriminate the visual forms of words (i.e., orthographic processes or "mental lexicon") the process of spelling a word is different from the act of reading because of the output required. Reception, or the taking in of information that is required for reading is generally considered to be easier than production which is required for spelling (Bryant & Bradley, 1980). Moreover, it has been suggested that spelling English words is a far more demanding task than reading (e.g., Barry, 1994; Seymour, 1992). The greater demands of the spelling process are due mostly to inconsistent spelling-to-sound (grapheme-phoneme) correspondences which are represented by many more correspondent mappings than in reading (Barry, 1994; Marsh, Friedman, Welch, & Desberg, 1980; Waters, Bruck, & Seidenberg, 1985). For example, while we read beef predictably since all words that end in -eef rhyme with beef, spelling the same word is less predictable since many words that rhyme with beef are spelled differently (e.g., leaf, chief] (Waters, Bruck, & Seidenberg, 1985). Traditionally, rote memorization was considered the cornerstone of spelling instruction (Heald-Taylor, 1998; Scott, 2000; Treiman, 1994) and is still relied upon to some extent today. Instruction from this approach focused on rehearsal and repeated copying of letters and words in isolation with limited transfer of these rote skills into essay writing. From this perspective, the student is regarded as a passive recipient of spelling knowledge (Heald-Taylor, 1998). However, current perspectives on the role of memory in spelling have stressed a reciprocal interaction between phonological and orthographic processes (Snowling, 1994). In order to retain word spellings, an association must be made between the way a word looks and the way a word sounds. According to Ehri (1980; Roberts & Ehri, 1983), it is the "amalgamation" of phonological and orthographic processes that is the key to a word's saliency in "lexical" memory. From a current instructional perspective, the student learning spelling is regarded as an active participant in the learning experience, not an "empty vessel" ready to be filled with spelling knowledge (Heald-Taylor, 1998). Thus, from an information processing approach words are actively received, encoded and retrieved by the student- - it is what is done with information that determines the strength of encoding and retrieval conditions (Schacter, 1996; Torgesen, 5 1985). Hence, the student is actively involved in encoding the phonology and orthography of words, and relies on numerous strategies to retain word spellings in memory (e.g., Barron, 1980; Gerber & Hall, 1987; Topping, 1995). To date, three main groups of spellers have been identified as processing spelling knowledge in distinct ways (Bruck & Waters, 1988; Frith, 1980; Lennox & Siegel, 1993, 1994): Good readers and spellers; poor readers and spellers; and good readers, poor spellers. The poor reader but good speller does not exist according to researchers (e.g., Aaron, 1982; Bruck & Waters, 1990), due to the greater complexities of the spelling as opposed to the reading task, as described earlier. Good readers and spellers tend to produce rich associations between phonological and orthographic information for spelling, but poor readers and spellers have deficient phonological skills and tend to rely on orthography, making more visually as opposed to phonologically accurate misspellings (e.g., Lennox & Siegel, 1993). Research on the so-called "mixed" group composed of good readers but poor spellers has revealed somewhat disparate findings. For example, some researchers have found similarities between good readers and spellers and the mixed group for their use of phonological processing, but deficient use of orthographic processing (as measured by the ability to detect and produce conventional letter patterns in words) (Frith, 1980; Lennox & Siegel, 1994). However, other researchers have found comparable phonological processing deficits and no differences due to orthographic processing between mixed and poor readers and spellers (Bruck & Waters, 1988; Waters, Bruck, & Seidenberg, 1985) leading some (e.g., Kamhi & Hinton, 2000) to conclude that mixed readers and spellers are not distinguishable from poor readers and spellers. There is substantial evidence, however, that the mixed group is unique and distinguishable from the poor group (e.g., Frith, 1980; Lennox & Siegel, 1993, 1994) and that the dissociation between reading and spelling skills in the mixed group is due, in part, to orthography and visual memory of letter sequences and whole words. In fact, most of the inconsistent findings in the literature regarding the mixed reader and speller is due to the varied measures used to define reading and orthographic processing. There is a dearth of research specific to the memory processes of groups distinctive in their use of phonological and orthographic processes. Although it has been found, that these three groups differ in the degree to which they rely on phonological or orthographic information in spelling (e.g., Lennox & Siegel, 1993, 1994), there have been no studies examining the links between these processes and memory across the three groups. The importance of such a study is emphasized by the fact that memory strategies, prolific in spelling curricula (e.g., Heald-Taylor, 1998; McNaughton, Hughes, & Clark, 1994), may be directed to the specific processing needs of each group. 6 In setting a foundation on which to study the relationship between phonological and orthographic processes and memory in spelling, this chapter begins by examining information processing models of spelling with emphasis on memory as a central process. Two main "routes" are described and the ways in which spelling and reading skills are related are examined from this dual route perspective. Extending from a review of current perspectives on spelling processes, literature and theory on spelling development is examined with emphasis on spelling skill as emerging in a continuous manner rather than through a series of hierarchical "stages". Next, evidence across varieties of spelling competence and specific sub-groups of spellers is examined, outlining the controversy over the existence of a group of spellers with mixed reading and spelling skills. Finally, the relationships between phonological and orthographic processes and memory are described. B. Information Processing Models of Spelling Models of spelling from an.information processing perspective stress the internal mechanisms somewhere within the brain that are responsible for spelling production. Whereas the act of writing as a physical process is believed to involve peripheral mechanisms, memory processes are considered "central" to this model in that they form the basis by which information is encoded, stored, and retrieved (Seymour, 1992). A consensus exists in the literature about the involvement of two other central processes involved in spelling, one based on the retention of word forms (i.e., lexical processes) and the other on the retention of a set of sound-to-spelling (i.e., phoneme-grapheme) correspondences (Barry, 1994; Seymour, 1992). Although evidence exists that these two central processes, termed "dual routes", function in distinct and separable ways, it is more likely that, in non-brain impaired adults and children, a reciprocal interaction more accurately describes their functional relationship (Barry, 1992; Brown & Ellis, 1994; Snowling, 1994). (1) Dual routes; Orthography and phonology. According to Barry (.1992) and Seymour (1992), there are two persuasive reasons why a dual route theory of spelling production is plausible. The first reason involves research conducted with patients who have acquired a spelling disorder (i.e., acquired dysgraphia), either through disease or injury (Barry, 1992; 1994). Compelling evidence for the viability of the dual route framework is that dissociations have been observed between either phonological or orthographic processes in these patients, depending on the type or localization of the impairment. Moreover, it is through this clinical evidence that the separability and distinctiveness of the two processes to spelling production is apparent. First, impairments in phonology (i.e., phonological dysgraphia) have been reported in the literature where the patient's capacity to use the phonological route has been impaired, but the lexical route remains intact (e.g., Bub & Kertesz, 1982; Roeltgen, Sevush, & Heilman, 1983). As 7 a result, the person's ability to spell nonwords (e.g., sim), for which they must rely on sound-to-spelling correspondences, is impaired, although their capacity to use lexical processes in memory is intact. A striking example of such a dissociation was described by Shallice (1981) in a seminal case study. The patient, PR, was unable to write nonsense syllables or nonsense words. Although PR could repeat a spoken nonword immediately and after a delay, he made many errors when spelling dictated nonwords (82%) as compared to words (6%). PR also had difficulty manipulating the sounds within words (i.e., phoneme segmentation), a critical pre-requisite skill for the phonological aspects of spelling (Cataldo & Ellis, 1988; Goswami & Bryant, 1990; Goswami, 1992). According to Shallice, and as also described by Barry (1992), PR reported using a strategy for spelling that appears to demonstrate his reliance on visual-orthographic processes in that he reported seeing the word on an "inner screen" (p. 418). Barry (1992) cites a case study reported by Beauvois and Derouesne (1981) describing the French patient, RG, who suffered from an inability to access visual-orthographic information in memory (i.e., lexical agraphia). Unlike PR whose most significant deficit involved nonword spelling, RG could spell nonwords, but showed increasing difficulty producing spellings for real words as the sound-to-spelling correspondences became more ambiguous (e.g., from 93% to 36%, respectively). Due to his impaired visual memory, RG tended to spell ambiguous words phonetically, having difficulty accessing alternate sound-spelling representations from memory (e.g., "pigeon" spelled "pigon"). Other researchers have reported similar findings with lexically impaired patients (e.g., Goodman & Caramazza, 1986; Goodman-Schulman & Caramazza, 1987; Roeltgen & Heilman, 1984) whose most striking characteristic is their tendency to spell every word as if it is a nonword (e.g., yot for yacht). Certainly, research conducted with brain-impaired populations provides strong evidence for the involvement of two functions separately located within the brain which are also independently vulnerable to neurological damage (Shallice, 1988). However, the question of whether the processes are really separable in the non-brain impaired population remains empirically unanswered. What has been posited in the literature is that populations (both children and adults) varying in spelling competence differ to the extent that they rely more on visual-orthographic or phonological strategies for spelling production (Barron, 1980; Boder, 1973; Lennox & Siegel, 1993, 1994). According to Barry (1992), evidence concerning the reliance on different processes as apparent in the strategies used for spelling production is considered the second reason for the plausibility of the dual-route model. Indeed, a common method for uncovering the strategies used for spelling is to examine the errors across regular and irregular words. In contrast to regular words (e.g., cat, greed) which are predictable in their sound-to-spelling relation, irregular words (e.g., said, colonel) have unpredictable sound-to-spelling correspondences and thus can 8 only be spelled by accessing the word's visual form from lexical memory (Boder, 1973; Ehri, 1980; Roberts & Ehri, 1983). In fact, Barron (1980) posited that spelling regular words accurately requires both phonological and visual-orthographic strategies, but that successful spelling of irregular words is the result of visual-orthographic strategies alone. Moreover, so called "slips of the pen" produce many signs of a reliance on phonology with misspellings that are plausible based on sound-to-spelling relationships (e.g., sensative; properganda), as Barry (1994) has also noted. The most compelling argument found within this literature on strategy use across the two processes, however, is that the routes work together in a reciprocal manner (Snowling, 1994). The best corpus of evidence for this reciprocal pattern of influence to spelling production comes from research on lexical priming and analogy. (2) Reciprocal routes: Lexical priming and analogy. As Snowling (1994) has indicated, spelling new words according to the visual-orthographic patterns of known words is termed spelling "by analogy". For example, if the known word leave was consulted to spell the word bereave, the latter word has been spelled by analogy. Indeed, studies examining spelling by analogy have been considered the single most persuasive line of evidence for the functional interaction of phonological and orthographic processes in spelling production (Snowling, 1994). A study by Campbell (1983) provided evidence of the use of analogy in spelling. Campbell demonstrated that it was possible to bias adult participants' spellings of rhyming nonwords by preceding each target nonword with a rhyming real word spelled with the same rime. For example, each target nonword had alternative orthographic representations (e.g., [fri:t] could be spelled/reat orfreet) and was preceded by a rhyming real word especially selected to prime one of the alternative spellings (e.g., cheat primed the spelling/reat). The fact that participants were more likely to spell the nonwords in a pattern that matched the orthography of the priming words led Campbell (1983) to suggest that phonological and visual-orthographic processes are implicated jointly in the spelling of nonwords. There is also evidence that this lexical priming effect involves associative (i.e., semantic) processes as well. For example, Seymour and Dargie (1990) in a study modeled after Campbell (1983) found that adults were more likely to spell the nonword [boup] as bope when the word Vatican preceded presentation and as boap after the presentation of the word soap. It was concluded that nonword spelling involves a system that is not distinct from orthographic and semantic processes. Moreover, it is also clear that processes (phonological, visual, and in this case semantic) are working in an interactive manner, depending on the information processing demands of the task, as Snowling (1994) has also noted. Furthermore, it has also been found that adults appear to be sensitive to the frequency of sound-to-spelling patterns in nonwords, particularly for vowel phonemes (Barry & Seymour, 9 1988). According to Barry and Seymour (1988), a high so und-to-spelling contingency is evident when a particular letter string is the most common way of representing a vowel sound (e.g., ai as in rain). In contrast, when a letter string is rarely used to represent a particular vowel sound (e.g., et as in rein) a low sound-to-spelling contingency is evident. Extending Campbell's (1983) study, Barry and Seymour (1988) manipulated not only the orthographic alternatives to prime rhyming nonwords, but also the contingencies of sound-to-spelling patterns in vowel phonemes. The results revealed that in addition to priming, contingency influenced the probability of letter choice in nonword spelling as adult participants produced more high than low contingency spelling patterns for the vowels in nonwords. For example, the nonword /pi:m/ was more likely to be spelled as peam orpeem than as piem or peme- - spelling patterns that are rarely or never used in English. Indeed, it appears that letter choice in nonwords spelling is influenced by the occurrence of lexical priming and the frequency of occurrence of sound-to-letter patterns. The functional interaction of phonological and orthographic processes as evidenced by lexical priming has also been found in children's spelling production. As Snowling (1994) has noted, the first study of this kind was reported by Marsh, Friedman, Welch, and Desberg (1980) who asked children (7-year-olds and 10-year-olds) to spell nonwords which were analogous to irregular real words (e.g.Jation, wength, zoldier as in nation, length, and soldier). It was presumed that each nonword would prime the students to use the letter pattern of the real word. Although none of the young children (7-year-olds) were able to spell the nonwords by analogy to the real words, about a third of the older children (10-year-olds) were successful, suggesting a possible developmental factor in the use of analogy strategies. However, this study has been criticized on the basis that there were no preliminary data indicating whether children knew how to spell the analogous real words in the first place (Goswami & Bryant, 1990) and hence target words were likely too difficult (Snowling, 1994). Campbell (1985) also extended the lexical priming study that was conducted with adults to 9- to 11-year-old children. A list of words and nonwords was dictated to children, and they were instructed to write down any nonwords they heard. Parallel to the adult study, nonwords were preceded by real words (e.g., Izoull preceded by pole or coal) which were intended to prompt the spelling pattern of the preceding prime (e.g., zole after prime pole). Similar to Marsh et al. (1980), younger children with a reading age below 11 years experienced a much weaker priming effect than children with a reading age of 11, whose performance was similar to that of adults. Although it appears that there are developmental constraints operating on the tendency for young children to be susceptible to lexical priming, a key factor appears to be whether young children's attention is directed to the orthographic structure of words (Snowling, 1994). In the studies noted above, only older children with more exposure to the printed word and hence 10 more sophisticated orthographic knowledge were susceptible to the priming effects. However, Goswami (1988) has found that children as young as 7 years can in fact spell words by analogy. Her methodology departed somewhat from the other studies described in that real words, as opposed to nonwords, were used, and processing demands of students were explicitly directed to the analogous word. For example, children were instructed how to spell a clue word such as beak. Clue words were written on cards that remained in sight and children were informed that clue words might help them to spell unfamiliar words, like bean or peak. Under these information processing conditions, young children were able to use the analogous cue words to spell target words. Thus, it appears that due to their less sophisticated knowledge of words and the lexical patterns of words, younger children are not as likely to automatically produce spellings based on an analogy to a known word because their store of known words is rudimentary (Snowling, 1994). However, when provided with explicit instruction, even young children can proficiently spell newly learned words by analogy. In summary, the findings on the effects of lexical priming and spelling by analogy provide evidence for the functional interaction of phonological and orthographic processes in spelling (Goswami & Bryant, 1990) across adult and child populations. Moreover, the fact that exposure to print, in the form of reading experience, is a factor in young children's ability to spontaneously use analogy strategies highlights the role reading plays in spelling production. In fact, the processes required for both reading and spelling may be considered complementary and interactive (e.g., Ehri, 1987;Goswami & Bryant, 1990). (3) Relationship between reading and spelling across routes. It has been suggested (e.g., Cataldo & Ellis, 1988; Ehri, 1989, 2000; Snowling, 1994) that reading serves to establish mappings between phonology and orthography and allows children to experiment with written words, going back and forth between their written attempts and the stored form of the word in memory based on having seen the word in reading. Certainly, print is the medium through which children come to know that speech sounds take on symbolic form. It is through exposure to print that letter sounds are associated with alphabetic forms and as such the dual processes of phonology and orthography are inherently intertwined. Although the processes required for reading and spelling appear to be complementary, it is also possible that the enactment of processes across the two routes differs somewhat between reading and spelling due to what Bryant and Bradley (1980) have termed the "problem of asymmetry" -- the fact that there are many more (and inconsistent) sound-to-spelling patterns than spelling:to-sound relations. A study by Waters, Bruck, and Seidenberg (1985) was conducted to uncover the processing requirements of reading and spelling. Additionally, of critical concern to these researchers were the differences between spelling ability groups based 11 on the degree to which phonological processing (i.e., spelling-to-sound and sound-to-spelling correspondences) was relied upon across spelling and reading tasks. In Waters et al. (1985), three groups of grade three students were instructed to spell and read words and nonwords that differed with respect to the number of alternate spellings or pronunciations associated with a particular spelling pattern. Indeed, evidence of sound-to-spelling and spelling-to-sound correspondences being used in spelling and reading, respectively, was hypothesized to be related to the likelihood of alternate spelling patterns for a word's pronunciation influencing spelling accuracy, and the likelihood of alternate pronunciations associated with a word's spelling pattern affecting reading accuracy: Accordingly, words and nonwords included the following five types: 1) regular words spelled and read as they are pronounced (e.g., dish), 2) regular words, read as they are pronounced but with alternate spellings (e.g., beef-- eef can be spelled as eef, eif, eaf), 3) ambiguous words with spelling patterns that have two or more associated pronunciations (e.g., bear, fear), 4) exception words that have common spelling patterns that are usually pronounced otherwise (e.g., have, -ave usually pronounced as in gave, save, pave), and 5) strange words that have irregular pronunciations with spelling patterns occurring in a very limited number of monosyllabic English words (e.g., ache). Based on standardized reading comprehension and spelling dictation scores, groups were composed of "good" spellers (good spellers and readers), "poor" spellers (poor spellers and readers) and "mixed" spellers (good readers, poor spellers) who did not differ statistically from the good group in terms of reading nor the poor group in terms of spelling. Error analyses revealed that across ability groups and for both spelling and reading, students relied on sound-to-spelling and spelling-to-sound correspondences, respectively. For example, in spelling words, "good" spellers did best on regular words and made the most errors on strange, exception, ambiguous, and regular words with alternate spellings. Similarly, the reading performance of "good" spellers was related to the frequency with which alternate pronunciations were associated with a particular spelling pattern. For example, exception words produced more errors than ambiguous words which produced more errors than regular and regular words with alternate spellings. "Good" spellers were also more competent in reading nonwords, another sign of their better-developed spelling-to-sound knowledge. However, the "mixed" and "poor" spellers showed evidence of poorer phonological processing in spelling and reading. For example, while these students produced more errors spelling exception and strange words compared to the "good" spellers, they also made more errors spelling nonwords with fewer phonetic misspellings than the "good" spellers. Moreover, there was no difference in "mixed" and "poor" spellers' ability to read ambiguous and regular 12 words, suggesting a poorer knowledge of sound-to-spelling correspondences compared to the "good" spellers. A main criticism of the Waters et al. (1985) study is that a measure of reading comprehension was the only measure used to distinguish students in terms of reading competence (Lennox & Siegel, 1993, 1994). It is widely accepted that phonological processing is central to reading ability (Stanovich, 1989; Siegel, 1993) and that there is a very strong association between deficits in reading skill and deficits in phonological processing. The most significant way to assess phonological processing is with pseudoword reading tasks which are more highly correlated with text reading and word recognition than with comprehension (Siegel, 1993; Siegel, 1998). Therefore, using pseudoword reading, text reading, br word recognition tasks to define reading ability provides a more optimal definition of reading ability than reading comprehension. Indeed, these methodological criticisms do not undermine the shared link that has been established between phonological processing in reading and spelling. Based on the findings from the Waters et al. study, it appears that children who are good at both reading comprehension and spelling have a better knowledge of spelling-sound correspondences than do children who are good comprehenders but poor spellers or children who are poor at both reading comprehension and spelling. Clearly, the phonological processing involved in formulating these spelling-sound or sound-spelling relationships is crucial for reading and spelling alike, as others have also reported (e.g., Cataldo & Ellis, 1988; Ehri, 1987; Goswami & Bryant, 1990; Snowling, 1994). Additionally, Ehri (1980; Roberts & Ehri, 1983) has suggested that both spelling and reading are reliant on the richness of the connections between phonology and orthography in the form of sounds and symbols. When this functional interaction between the two processes is poorly established, a break down in skill is the result, either for reading, spelling or for both. The Waters et al. (1985) study points to one crucial reason for a poor connection and hence the establishment of a "dysfunctional" interaction-- the lack of ability to make sound-spelling and spelling-sound correspondences. Therefore, the most pertinent pieces of information to emphasize from this section are: 1) that given the shared processing demands of both tasks, any study of spelling skills must also consider reading proficiency as well, and 2) that research evidence has uncovered three groups of spellers who differ in both reading and spelling ability. Research examining the characteristics of these groups will be presented in a subsequent section. Summary. Research has clearly established that there are two information processing routes for spelling, one for processing the sounds in words and one for the visual parts of words. Moreover, while these routes can work in isolation under conditions of neurological 13 insult, in "normal" adult and child spellers, the processes work with some degree of functional interdependence. Furthermore, conceptualizing the dual routes as interactive and composed of processes shared by both reading and spelling serves as the guiding theoretical framework for the proposed research. In the following section, a review of the literature on the development of the dual processes is presented, emphasizing the interaction of phonological and orthographic cues in learning to spell. C. Phonological and Orthographic Processes in Learning to Spell Research and theory in spelling development are relevant to the proposed research for two important reasons: 1) to elucidate the emergence of phonological and orthographic processing and the progressive fluency of processing with age and experience, and 2) to examine the relative importance of each of these processes to children's developing competence in order to compare processing differences between groups (e.g., good and poor spellers who vary in reading ability). As a movement away from traditional theories of spelling development which emphasized the rote memorization of letter sequencing and whole words, researchers posited new theories of spelling development based on a hierarchy of acquired skills (Gentry, 1982; Henderson & Templeton, 1986). From this perspective, development progresses from a sequence of stages stressing the relative importance of phonological skills and gradually including more lexical, albeit orthographic, proficiency (Lennox & Siegel, 1994). Although there is consensus that developmental changes do indeed occur in children's spelling knowledge, the fact that these changes are constrained within stages has been criticized (e.g., Lennox & Siegel, 1994; Snowling, 1994; Treiman, 1993, 1994; Varnhagen, McCallum, & Burstow, 1997). Rather, it has been argued that children enact strategies of a phonological or orthographic nature at any one time depending on the information processing demands of the task (e.g., Snowling, 1994) and that the progression of development across processing is more continuous and reciprocal than hierarchical. (1) Stages of emergence of spelling skills. Developmental models in which the acquisition of spelling skills progresses through a series of discrete stages have been described by a number of investigators (e.g., Brown, 1990; Gentry, 1982; Henderson & Templeton, 1986). Importantly, a theme shared by all of these models is the beginning emphasis in early stages on phonological processing to a more sophisticated use of orthographic processing in later stages. Thus, as children become progressively more fluent with handling the sound-symbol relationships of words, the use of strategies of a more orthographic nature (i.e., lexical analogies) will follow in the developmental sequence. 14 As Brown (1990) and Lennox and Siegel (1994) have reported, stage theorists (e.g., Gentry, 1978; 1982; Gentry & Gillet, 1993; Henderson & Templeton, 1986) posit a beginning stage where children experiment with language through its representative symbols. During this earliest stage, children may attempt to write "messages" that reflect a nascent understanding that speech can be represented symbolically, but with no clear sound-symbol correspondences evidenced (e.g., ship written as jjed). This is followed by a semiphonetic stage (grades 1 and 2) in which children learn about sound-symbol associations and about left-to-right progression. A child in this stage may write ru for are you. A phonetic stage (grades 3 and 4), and a fourth transitional stage (grades 4-6) follow in which spellers become more knowledgeable about spelling rules and develop a visual-orthographic strategy. In the phonetic stage, spellings tend to reflect an awareness of every letter sound detected in a word (e.g. said spelled as sed). The transitional stage is marked by greater fluency with sound-letter relationships coordinated with a burgeoning attempt to represent the visual appearance of words (e.g., yacht spelled yahct). Finally, theorists suggest that in stage 5 (grade 7 to adulthood) spellers demonstrate a sophisticated knowledge of English spelling and recognize spellings because they "look right" (e.g., Brown, 1990). Mature spellers can fluently detect and represent the sounds within words and have a vast store of what correct spellings look like. Similarly, Frith (1980, 1985) has proposed a three- stage model of spelling development that is particularly relevant to the proposed research because it posits the interplay of development and process between reading and spelling (Ellis, 1994). In the first togographic stage, spelling comprises some rote words, such as the child's own name. The second alphabetic stage reflects the emergence of decoding skills. In the third and final orthographic stage spelling lexical analogies may be used as spelling becomes more independent of sound. Moreover, according to Frith (1985) the relative use of strategies of a phonological or orthographic nature changes as children hone their literacy skills. Thus, through spelling children learn to apply phonological strategies which become more fluent with practice and hence transfer into reading. Reading provides children with print exposure that enables practice in processing and storing the visual forms of words which they can draw upon when spelling. Despite the apparent symbiosis between reading and spelling, however, the progression of development as reflected in Frith's stages moves from phonological to orthographic processing. Indeed, the progression from an early use of phonological cues to a more mature use of orthographic strategies in later stages is a typical theme underlying most stage theories of spelling development. Another model of spelling development, important to the proposed research for its description of the integration of phonological and orthographic processes'in memory, is one described by Marsh, Friedman, Welch, and Desberg (1980). According to these authors, in the 15 initial stages when the child is decoding basic consonant-vowel-consonant (CVC) patterns and enacting a sequential phonemic encoding strategy (i.e., where the sounds in words are stored in the order they appear) the entire word must be processed and, as a result, a visual representation of the word in memory is constructed. Once this basis of word knowledge has been established (i.e., the sounds and visual forms of words), the child then develops the use of an encoding strategy that is hierarchical and based on conditional rules. At this stage, for example, the more sophisticated speller moves back and forth between their store of word knowledge (encoded via phonological and orthographic processes) and the familiarity of spelling words (i.e., known words versus new words). It is also suggested that the more experienced speller might move between strategies, enacting a phonemic encoding strategy in spelling unfamiliar words to a strategy based on analogy with known words in visual memory. This ability to switch between strategies is considered a sophisticated skill and according to Marsh et al. (1980) is not apparent in young inexperienced spellers. However, what is critical in the Marsh et al. model to the proposed research is that both phonological and orthographic encoding occur from the earliest stages of spelling development. Moreover, as exposure to print expands, so too does the repertoire of words in memory. Although phonological skills are a crucial prerequisite to spelling development (e.g., Goswami & Bryant, 1990; Read, 1980) the fact that these skills precede orthographic processes in the developmental sequence has been challenged by many (e.g., Lennox & Siegel, 1993, 1994; Snowling, 1994; Treiman, 1994). Furthermore, while Marsh et al. (1980) describe a model that moves beyond these sequential constraints, the flexibility of processing between phonological and orthographic skills is still considered a mature skill. However, even young children demonstrate an ability to flexibly enact both processes at any one time, depending on the information processing demands of the task (Snowling, 1994). It is thus more appropriate, as Snowling (1994) has also described, to view the development of phonological and orthographic processes as reciprocal and interactive rather than as hierarchical. (2) Beyond stages: Phonology as the framework for orthography. Indeed, there is evidence that young children use orthographic strategies along with phonological cues at an early age. For example, Treiman (1994) found that young children (e.g., beginning grade 1) were more likely to use legal spellings, such as ck at the ends of words as opposed to the beginnings, and thus drew on orthographic processes from the outset. Further evidence of children's orthographic processing from early in spelling development is found in the lexical analogy studies conducted by Goswami (1988) as described in a previous section. To review briefly, Goswami found that when children as young as 7 years were provided with explicit instructions and an explicit prompt to use a previously learned word to help them 16 spell a new word (e.g., use beak to help spell bean), children spelled by analogy and hence drew on orthographic processing as well as older more experienced spellers. Similarly, Snowling (1994) describes a study where 8-year-olds were able to use orthographic processing after being primed to spell nonwords using an analogy strategy. Children were also sensitive to the likelihood of different spelling-sound relations (i.e., how common is the spelling pattern), demonstrating more priming for words with low-contingency or unusual spelling patterns than high-contingency, common spellings. However, it is difficult to judge the efficacy of this study as it is presented merely in descriptive form; no statistical information regarding analyses were described. Still, when considered along with the evidence provided by Goswami (1988) and Goswami and Bryant (1992), it does appear that children's use of orthographic processing in the form of lexical strategies develops alongside the use of phonological strategies. The fact that some researchers have demonstrated the early emergence of orthographic knowledge and processes compared to others (e.g., Campbell, 1985; Marsh et al., 1980) is likely due in large part to the information processing demands of the tasks, particularly working memory demands. As noted in an earlier section, one important difference between the studies conducted by Campbell (1985) and Goswami (1988) was the visibility of a cue word as an explicit prompt for deriving an analogous spelling (e.g., the word beak remained visible and children were instructed to vise the word to help them spell bean). Children's capacity to hold information in mind while carrying on some other form of processing (i.e., working memory) becomes progressively more efficient with age and experience (Siegler, 1991). As children build up their knowledge base through practice using phonological and orthographic processes, they develop a richer store of information to use for spelling. The richer the store of information, the more fluent children become in processing spelling information, and in handling working memory demands. It is for this reason that, as Snowling (1994) states, "[Children's] spellings will normally be better when writing single words than composing a story, their letter-sound translation skills will be better when tested in isolation than when used in a situation where they also have to segment and to memorize components of spoken words" (pp. 112). As noted earlier, memory is indeed central to the spelling process and hence the role of phonological and orthographic codes in memory will be examined in a subsequent section. Summary. Theorists advocating stages of spelling development have elucidated the important place of phonological skills in spelling development, but have neglected the early impact of visual-orthographic processes within a coordinated model. Indeed, there are important connections between the.use of phonological and orthographic processing in learning to spell. According to Lennox and Siegel (1993, 1994), through decoding, words are stored in visual memory which serves as a-knowledge base to construct spellings by analogy. 17 Analogy to a word in visual memory is "therefore dependent on earlier phonological skills and can only proceed if phonological skills have been developed successfully" (Lennox & Siegel, 1994, pp. 96). Moreover, early phonological skills affect not only the child's proficiency at sounding out individual letters, but also the ability to segment rhymes such as en in hen and pen using analogies between spellings (Goswami, 1988; Goswami & Bryant, 1992; Lennox & Siegel, 1994). Therefore, as Snowling (1994) has suggested, phonology is the framework upon which orthographic knowledge is built and this phonological framework is fundamental to the development of orthographic knowledge. D. Varieties of Spelling Competence Due to research examining the important relationship between reading and spelling (e.g., Ehri, 1987; Waters, Bruck, & Seidenberg, 1985), spelling competence has come to be defined both within its own unique set of processes and in relation to reading, as described previously. Early research dichotomized good and poor spellers based on spelling scores alone, disregarding reading performance (e.g., Lesiak, Lesiak, & Kichheimer, 1979). However, it is now widely accepted that the processes required for spelling are highly compatible with those required for reading, particularly phonologicalprocesses (Bryant & Bradley, 1980; Ehri, 1985, 2000; Waters & Bruck, 1985, 1988). While three main subtypes of spellers have been identified who differ in reading ability (Bruck & Waters, 1988, 1990; Frith, 1980, 1985; Lennox & Siegel, 1993, 1994), there is still considerable controversy over the precise nature and distinction between groups, particularly mixed spellers who read well but spell poorly (Frith, 1980, 1985). Mixed spellers are the most interesting to examine, from a processing perspective, in that they reflect a group with an asymmetrical level of proficiency between reading and spelling (Bruck & Waters, 1988). The subsequent section, which begins with a description of the methodological approach used to determine the spelling strategies employed by students, considers the controversy over subtypes of spellers, concluding with an analysis of issues left to resolve. (1) Analysis of misspellings in the composition of subtypes. A critical methodology in distinguishing groups is the analysis of students' misspellings (Brown, 1990; Cook, 1980; Lennox & Siegel, 1993, 1994). As Lennox and Siegel (1994) describe, the use of error analysis is based on the assumption that errors provide insight into the kinds of strategies of a . phonological or orthographic nature enacted by students and hence can provide clues of processing strengths and weaknesses. Based on such analyses, patterns of processing differences as reflected in the variety of misspellings produced by good, poor, and mixed spellers 18 have been uncovered (e.g., Bruck & Waters, 1990; Lennox & Siegel, 1993, 1994; Waters, Bruck, & Seidenberg, 1985). Accordingly, a general scoring criterion involves evaluating a misspelling on the basis of the degree to which it reflects the sound (i.e., evidence of the use of a phonological strategy) or the visual appearance (i.e., evidence of the use of an orthographic strategy) of the target word (Bruck & Waters, 1988; Lennox & Siegel, 1994). Bruck and Waters (1988) first described the rationale behind this scoring criterion which was also used by Lennox and Siegel (1994). According to these researchers, if the misspelling sounds like the target word (e.g., yot for yacht), then a phonological misspelling has been produced and indicates the use of a phonological strategy. Conversely, if there is limited evidence of a phonological match between the misspelling and the target word (e.g., ufch for yacht), then there is no indication of phonological strategy use. Similarly, the production of a misspelling that looks like the target word (e.g., yatch for yacht), is evidence of a visual misspelling and reflects an attempt to use orthographic strategies. If no visual match is apparent between the misspelling and the target word, there is no evidence of orthographic strategy use. Several advantages and disadvantages have been noted on the use of a misspelling analysis. First, it has been suggested that an error analysis that considers both phonology and orthography is critical in spelling research where the dual processes are involved (Treiman, 1993). More insights are uncovered as to how students produce spellings when clues of strategy use across phonology and orthography are considered. Similarly, analyzing student spelling products, albeit their misspellings, reflects a product of their thought process as they solved the problem of the spelling task (Gerber, 1984; Gerber & Hall, 1987). However, there are several criticisms of the use of misspelling analysis. First, it has never been empirically demonstrated that the assumptions regarding the strategies used to produce errors are the same as those strategies used to produce correct spellings as Lennox and Siegel (1994) have also pointed out. Moreover, an important element is omitted from the analysis: students' own report on the kinds of strategies they used. In fact, it has been suggested in regard to strategy instruction research generally (Kail & Bisanz, 1982), that an optimal way to examine strategy use by children is to include both a quantitative measure (such as an error analysis) along with student self-reports at the same time as they are producing spellings. As Steffler, Varnhagen, Friesen, and Treiman (1998) have recently noted, a thorough analysis of spelling strategies needs to consider children's own reports of strategy use across correctly and incorrectly spelled words. In response to these criticisms, the present study seeks to join these two methodologies, combining children's verbal reports while they spell along with an analysis of spelling errors. 19 (2) Strategies and processing characteristics of good and poor spellers. Much of the research across reading and spelling ability groups has provided contradictory results. The varied findings reported in this area, however, are likely due to two key methodological factors: a) differences in measures used to determine reading achievement (i.e., comprehension versus decoding), and b) sample age differences (Brown, 1990; Lennox & Siegel, 1994). Despite controversial evidence, research on the processing characteristics of good and poor spellers consistently indicates the important role of phonological and orthographic processes enacted by students with varying spelling abilities. Early work examining the differences between good and poor spellers was conducted by Bryant and Bradley (1980) and Frith (1980, 1985). These researchers emphasized the compatibility in processes between reading and spelling as it characterizes proficiency within each skill. Concerning the strategy use of poor readers and spellers, Bryant and Bradley (1980) posited a strategy selectivity model whereby there is a failure to transfer phonological strategies used in spelling to decoding in reading. These students were also considered to over-rely on a phonological strategy at the expense of orthographic processing. Hence, poor readers and spellers demonstrate, according to this model, some inflexibility in the way they use phonological and orthographic processing between reading and spelling tasks. Similarly, Frith (1980) supported this strategy selectivity model suggesting that groups of spellers may be distinguishable with respect to their use of "full cue" or "partial cue" strategies for reading and spelling. For example, students who read by a full cue approach consult spelling-sound patterns and they tend to read mostly "by ear" fully attending to every part of the word. As Ehri (1980, 1987) has also suggested, such a strategy promotes the amalgamation of phonological and orthographic information, and students build up a rich store of word knowledge. Conversely, students who read by partial cues deprive themselves of the richness of the underlying spelling system (Willows & Scott, 1994). Consistent with Ehri's theory, reading by partial cues weakens the association of sound-spelling patterns, and as a result, a more limited store of word knowledge is constructed. According to Frith (1980), the fact that most poor spellers read by partial cues limits the usefulness of orthographic, information. Although Biyant and Bradley (1980) and Frith (1980, 1985) have suggested that good and poor spellers differ in the kinds of strategies they use for reading and spelling, Waters, Bruck, and Seidenberg (1985) have found that students in fact rely on the same basic phonological process for reading and spelling albeit with varying degrees of success. In their study, as described earlier, Waters etal. (1985) instructed third grade children to read and spell nonwords and real words that differed in terms of their regularity for reading and spelling. Waters etal. found that students actually relied on spelling-sound and sound-spelling relationships for reading and spelling, respectively. However, the Waters etal. study may be 20 criticized with respect to the fact that reading comprehension defined reading performance to the exclusion of reading decoding skills. In order to address concerns regarding the type of measure used to determine reading performance (i.e., comprehension versus decoding) arid whether their earlier findings would be replicated with older students, Bruck and Waters (1988) extended their previous study to older children (grades 3 and 6) formerly classified as reading disabled (RD). Using a single word decoding measure of reading ability, the RD students' performance was compared to that of non-RD students of the same reading and spelling levels. Moreover, misspellings were scored according to their phonological and visual-orthographic accuracy to target words. The analysis of the phonological accuracy of misspellings was based on two systems: constrained and unconstrained. The unconstrained system required correct sound-spelling associations (e.g., kat for cat). The constrained system in addition to requiring sound-spelling association rules, involved letter pattern and positional rules considered essential to accurate sound-spelling knowledge. Within this system, for example, natcher would not be scored as a phonologically accurate spelling of the target word nature, but would have been scored correct in the less stringent unconstrained system (Lennox & Siegel, 1994). Visual-orthographic accuracy was scored on the basis of the proportion of bigrams (i.e., letter pairs) and letters included in the student's misspelling compared with the target word. The results indicated, in contrast to earlier findings by Frith (1980), that RD children did not use qualitatively different strategies to read and spell words. The knowledge and use of spelling-sound and sound-spelling correspondence rules were found to be the same essential phonological process underlying reading and spelling across ability groups. In order to resolve the controversy reported in the literature, Lennox and Siegel (1993) endeavored to uncover a clearer understanding of the nature of the processes and strategies used by good and poor spellers. These researchers used a test of single word decoding in isolation to define reading achievement and sampled a large number of students (N= 420) from a wide age range (ages 6 to 16), thus responding to some of the key methodological criticisms of earlier studies. One of the key questions motivating the study was to examine whether differences between students were due to a "developmental lag" or whether students enacted qualitatively different strategies, depending on their level of spelling proficiency. That is, these researchers investigated whether the strategies used for spelling by poor spellers reflected those used by younger good spellers (hence evidence of a lag in development for the poor spellers) or whether poor spellers actually drew on different processes to spell words regardless of their age. Misspellings were analyzed according to the phonological unconstrained, constrained, and visual-similarity scoring scheme designed by Bruck and Waters (1988) as previously described. 21 The results indicated, consistent with Waters etal.'s (1985) research, that good spellers had a better understanding of sound-spelling relationships and their application. Unlike the Waters etal. research, however, good spellers were also found to have a better knowledge and use of orthographic strategies in that misspellings were close visual matches to target words. Moreover, Lennox and Siegel (1993) also found that, developmentally, good and poor spellers made about the same proportion of visual errors spelling in grade 2, but that qualitative differences emerged by the third grade. Although the good spellers spelled phonologically, the poor spellers demonstrated more reliance on orthographic processing for spelling difficult words, and these differences persisted until about the sixth grade. At the sixth grade level, phonological and orthographic strategies were found to be used to a similar extent by good and poor spellers as indicated by the equal proportion of misspellings across both phonological and orthographic criteria at the 6th grade level. Thus, according to the Lennox and Siegel research, it appears that up to the sixth grade, poor spellers use different strategies from good spellers in their spellings. Arguably, the most compelling element of Lennox and Siegel's (1993) design is the broad age range included in the sample (i.e., 6 to 16 year-olds). This design uncovered a developmental course of strategy use based on a misspelling analysis which surpassed the scope of earlier findings as described by Waters etal. (1985) and Bruck and Waters (1988). Moreover, the finding that poor readers and spellers make use of orthographic information more than phonological information is consistent with findings reported by some reading researchers. Most notably, Rack (1985) and Holligan and Johnston (1988) found that reading disabled students were more likely to remember orthographically similar target words (e.g., wash-cash) than phonologically similar words (e.g., good-should), suggesting a greater sensitivity to the processing of visual as opposed to phonological information. (3) Mixed spellers: Good readers, but poor spellers. In earlier work, Frith (1980) compared the performance of good, poor, and mixed spellers (reading performance based on single word decoding) on a variety of skills involved in reading and spelling. Based on the classification of misspellings as phonetic or non-phonetic, the good and mixed spellers were found to be more similar in terms of phonetic processing compared to poor students. Frith concluded that while the mixed spellers could use spelling-sound correspondence rules, they appeared to have difficulty with the precise sequence of letters in a word. Consistent with the strategy selectivity model (Bryant & Bradley, 1980), Frith suggested that the strategies used by mixed spellers were qualitatively different than strategies used by either good or poor spellers. For example, mixed spellers were hypothesized to use a partial cue strategy for reading and spelling. Using this strategy, mixed spellers succeed at reading by not attending to the complete letter-by-letter structure of words, but use more economical strategies such as attending 22 partially to the letter-by-letter structure and analogy to known words. Although this strategy may be considered efficient for reading, the reliance on a partial cue for spelling is ineffective due to the irregularities and exceptions in the English language. Moreover, mixed spellers are not attaining a complete, amalgamated experience with the phonological and orthographic characteristics of words leading to their spelling difficulties. However, the contention that mixed spellers have adequate phonological skills in the form of spelling-sound correspondences has been contradicted by the findings of the Waters, Bruck, and Seidenberg (1985) and Bruck and Waters (1988) research, as previously described. Specifically, Waters etal. (1985) found that mixed and poor third grade spellers produced a smaller proportion of phonetically accurate errors than did the good group. Moreover, despite the fact that students in the mixed and good groups were matched on performance in reading comprehension, the mixed spellers' quantity and quality of misspelling were more similar to the misspellings of the poor spellers than the good spellers. However, in a subsequent study with reading performance determined by single-word decoding, Bruck and Waters (1988) found that sixth grade mixed spellers were distinguishable from poor spellers by their better developed visual skills, having produced more visually accurate misspellings. These results are consistent with Frith's (1980) findings. Additionally, Jorm (1981) used a reading comprehension measure of reading performance and found, similar to Frith (1980), that mixed spellers made more phonetically accurate misspellings than poor spellers, resembling more closely the error patterns of good spellers. However, when mixed spellers (as defined by their reading comprehension performance) completed a single word recognition test, their performance was similar to the poor spellers. Jorm concluded, similar to Bruck and Waters (1985, 1988), that the same phonological processes underlie spelling difficulties, regardless of reading ability. However, Jorm's sample included only four students in the mixed speller group, greatly limiting statistical power and the generalizations that may be made about the processing characteristics of the mixed group. Clearly, how reading performance is determined is central to the definition of spelling subtypes and is the source of some conflicting findings in the literature as described thus far. To summarize these findings, Frith (1980) used a single word reading decoding measure in grade 6 students and found that mixed subjects were more like good spellers in their tendency to make phonetic errors. Waters and Bruck (1985) used a reading comprehension measure with grade three students, and found that there were no differences between the mixed and the poor spellers in terms of phonetic accuracy-- both groups made the same proportion of phonetic errors compared to the good group-- suggesting weaker phonological skills shared by mixed and poor subtypes when reading comprehension is used to define subtypes. .23 However, when reading is defined as single word decoding, older mixed students (grade 6) in the Bruck and Waters (1988) study were more similar to good spellers in terms of phonetic accuracy, consistent with Frith's (1980) earlier findings. Moreover, sixth grade mixed spellers (who were good reading decoders) were more likely to make visually accurate misspellings similar to good readers and spellers, providing evidence of a distinct subtype of poor spellers who are relying on orthographic processes to a greater extent. In summary, when mixed groups are defined as good comprehenders-poor spellers, no differences in phonological or orthographic processes emerge between mixed and poor spellers, raising doubt about the existence of a distinct subtype of "mixed" speller. However, when mixed groups are defined as good decoders-poor spellers, differences across phonological and orthographic processes are apparent. Bruck and Waters (1988) also reported that for the mixed students in their study, all but two performed proportionately well on both the reading comprehension and reading decoding measures. Hence, an asymmetry in the mixed students' reading skills was evident, There is no easy explanation for this asymmetry. One suggestion raised by Bruck and Waters is drawn from models of reading which stipulate that reading recognition is necessary but not sufficient for reading comprehension and that other skills such as reasoning, vocabulary, and general knowledge play an important role (Cunningham & Stanovich, 1991; Curtis, 1980). What is important to note for the purposes of the proposed study is the critical nature of defining reading ability. Since it is the good decoders and poor spellers who demonstrate distinctive differences in phonological and orthographic processes as described in the literature, these students were considered as the "mixed" subtype in the present study, and reading decoding was emphasized as the critical index for reading skill. Lennox and Siegel (1994) have conducted the most extensive study to date on the distinctive differences between good decoders but poor spellers (i.e., the mixed group). Misspelling analysis across phonological and orthographic criteria, as devised by Bruck and Waters (1988), was performed and students also completed eight additional tasks assumed to capture phonological and orthographic processing. These tasks were: 1) reading regular (e.g., nice) and exception words (e.g., police); 2) reading pseudowords (e.g., bim); 3) phonological real word judgment (e.g., kake-dake, which sounds like a real word?); 4) reading regular and irregular pseudowords (e.g., bave read as bau- like have- ovbave- like gave); 5) analogy reading (e.g., puscle ,fody); 6) spelling nonwords (e.g., imbaf); 7) orthographic real word judgment (e.g., fdv-fdk, which looks more like a word?); 8) spelling recognition (e.g., rain-rane). Their results indicated that good and mixed spellers performed better than the poor spellers on the phonological tasks that involved reading regular and irregular words, pseudowords and in spelling pseudowords. Moreover, good and mixed spellers produced more misspellings that were closer phonological matches to target words than did poor spellers, 24 consistent with earlier findings reported by Frith (1980) and Bruck and Waters (1988). However, mixed spellers performed similarly to poor spellers on the orthographic real word judgment task where a decision had to be made about which non-word looked more like a real word. The fact that performance on this task requiring orthographic processing was similar for mixed and poor spellers suggests that both groups have some difficulty with remembering the visual appearance of words. Whether this difficulty is due to problems representing the word visually in memory in the first place or in retrieving an already visually encoded word remains to be examined. Also unanswered is the question of why mixed spellers tend to make more visually accurate misspellings (as reported by Bruck & Waters, 1988 and Lennox & Siegel, 1993) suggesting some general reliance on orthographic processing for spelling, yet have specific difficulty detecting the visual patterns of real words from pseudowords. Summary. The following conclusions may be drawn from the research on the processing characteristics of spelling subtypes to date: 1) When measures of reading comprehension are used to define reading ability and spelling performance is assessed across dictation tasks varying in regularity, mixed spellers are more like poor spellers in their use of processing across phonological and orthographic routes. That is, regardless of reading ability, good and poor spellers rely on both processes for spelling but poor spellers are hampered in their use of these processes. 2) When single word decoding is used to distinguish groups, mixed spellers perform like good spellers in terms of phonological processing, but are more similar to poor students on some aspects of orthographic processing. Although they tend to make more visually accurate misspellings, mixed spellers have some difficulty judging the visual resemblance of pseudowords to real words. 3) Poor spellers tend to over-rely on orthographic strategies relative to good and mixed spellers (good decoders/poor spellers) due to their poor phonological skills. There appears to be some attempt to compensate, at least in part, for their phonological processing difficulties. 4) These distinctions appear to be apparent up to a spelling grade of 6, according to Lennox and Siegel (1994). At about the sixth grade level, students across groups appear to use an equal proportion of strategies across routes due to some flexibility in processing achieved through practice. This flexibility in strategy use with age has also been reported by Marsh, Friedman, Welch, and Desberg (1980) as described earlier. A critical issue that may explain the inconsistencies in the literature regarding subtypes of readers and spellers is the type of reading measure used to define reading ability. Findings in the literature have been based on defining reading performance as reading comprehension or reading decoding. It is therefore difficult to compare results across studies which consider rather different subgroups of readers and spellers. Furthermore, an important area that remains unresolved within this literature is the nature of the differences across spellers due to orthographic processing and the different ways 25 in which orthographic processing has been measured (e.g., misspelling analysis, orthographic awareness task, etc.). Indeed, there may be different definitions of orthographic processing depending on the processing nature of the task used. There is also the issue of whether subtypes of spellers differ in the extent to which they rely on phonological and orthographic codes in spelling. For example, it has been consistently found that older good decoders-poor spellers make more phonetically and visually accurate misspellings, compared to poor decoders-poor spellers (e.g., Bruck & Waters, 1988; Lennox & Siegel, 1993). However, this mixed group of good decoders-poor spellers has some difficulty on other tasks tapping orthographic skills. As has been reported in the literature, good spelling involves the integration of phonology and orthography to produce a rich and "amalgamated" store of words in lexical memory (Ehri, 1987). Whether the difficulties experienced by mixed and poor subtypes of spellers have to do with the way in which information is "coded" in memory (i.e., through a phonological/orthographic code or both) remains to be examined. Therefore, an important area worth pursuing is the nature of both processes across subtypes of spellers (differing in reading decoding) in relation to memory. E. Phonological and Orthographic Codes in Memory As noted earlier, memory is considered a central process to spelling through which phonological and orthographic information is encoded, stored, and retrieved for the spelling task (Ehri, 1987). It is also important to establish that memory is not a unitary phenomenon, but is composed of a myriad of encoding, storage, and retrieval processes across a broad spectrum of retention tasks (Schacter, 1996). Moreover, a theme integral to current memory research is that performance differences, across tasks and individuals, are a function of different processing (encoding and/or retrieval operations) being applied to information as opposed to different "capacities" or "stores" of memory systems (Schacter, 1997). In short, it is what we do with information that makes it memorable, not where it goes. One of the most reliable findings reported in the literature on reading disabilities is that students exhibit memory processing problems (Torgesen, 1985). Specifically, a compelling finding is that reading disabled (RD) students have difficulties encoding phonological information (Shankweiler, Liberman, Mark, Fowler, & Fischer, 1979; Torgesen, 1985). Moreover, other researchers have found that although RD students have difficulty processing phonological information, they appear to be able to encode orthographic information to a greater degree (e.g., Rack, 1985). The product of the encoding of phonological and orthographic material is referred to as a phonological and orthographic "code", respectively. Methodologically, evidence that students have used such codes is found in the proportion of information of a phonological or orthographic nature retrieved on retention tasks (e.g., Holligan & Johnston, 1988; Rack, 1985; Swanson & Ramalgia, 1992). 2 6 A description of the research conducted with RD students across the two processing routes is presented next. For all of the studies reviewed, reading ability is defined as performance on reading decoding tasks. Only one study to date has examined the relationship between phonological codes for memory and spelling tasks with RD students. This study is reviewed in detail. There is currently no research on the phonological and orthographic coding differences in memory between students who show an asymmetry in their reading decoding arid spelling skills. The present study addressed this issue. (1) Reading disabilities and phonological and orthographic codes in memory. Shankweiler, Liberman, Mark, Fowler, and Fischer (1979) demonstrated that reading disabled (RD) 8-year-olds make less use of a phonological code in memory compared to same-age non-disabled students. Shankweiler et al. presented groups of disabled and non-disabled readers with letter sequences for subsequent recall. Letter names in these sequences were either phonetically confusable (B C G T P V D) or were phonetically distinct (H Q S L W R K). Non-reading disabled students made more errors with phonetically confusable sequences exhibiting what Shankweiler et al. termed the "phonetic confusability effect". The RD students showed this effect but to a significantly lesser degree. Moreover, this effect was obtained regardless of whether stimuli were presented auditorily or visually. Based on this reduced "confusability effect" in poor readers, the researchers concluded that memory deficits in poor readers are due to degraded phonological processing. In order to examine whether RD students in fact make more use of orthographic information at encoding given their deficient use of phonological processing, Rack (1985) conducted a study with RD and non-RD students (mean age 11.8). Students were instructed to make rhyme judgments about word pairs and to complete a cued recall task immediately proceeding the rhyme judgment tasks. Importantly, four types of word pairs were devised which captured the shared visual appearance of word pairs, or the shared sound or phonological characteristics of word pairs. For example, pairs were either rhyming and visually similar (tall-wall); rhyming only (tall-crawl); visually similar only (tail-shall); or there was no relationship between the items forming control pairs (tall-thunder). In the first experiment, students were instructed to signal by making a tick in their booklet if the pair of words read by the experimenter rhymed, or by writing a cross if they thought the pair of words did not rhyme. Immediately following the rhyme judgment task, students were presented with one of the items in each word pair as a cue and were instructed to recall the other word in the pair. In the second experiment, a reaction time measure was taken in the rhyme judgment task. The results indicated that RD students were slower than non-disabled students in making rhyme decisions, not surprising given their reading difficulties. What is most noteworthy, however, is that RD students also took longer to say that pairs rhymed which were 27 visually distinct (head-said) than pairs which were visually similar (head-dead) compared to non-disabled students across auditory and visual presentations. In an earlier study, Seidenberg and Tanenhaus (1979) found the same "orthography effect" in adult good and poor readers. Consistent with this research, Rack concluded that, unlike non-disabled readers, the RD students made use of an orthographic code in rhyme judgment. Thus, RD students were more likely to attend to the visual as opposed to the phonological characteristics of words when making rhyme judgments. Moreover, RD students also recalled more targets from cues that were visually similar than not visually similar despite the fact that auditory presentation was used. This effect, termed the " ortho graphic similarity effect" by Rack was not demonstrated by the non-disabled readers. Thus, based on Rack's (1985) research it appears that RD students show some tendency to rely on an alternate code for reading, one based on orthography as opposed to phonology. In another series of experiments, Holligan and Johnston (1988) found phonological processing differences between RD and non-RD 8-year-old students, consistent with Shankweiler et al. (1979), with some evidence of differences in the use of orthographic codes, consistent with Rack (1985). Most notably, in Experiments 2 and 3, students made timed rhyme judgment decisions for word pairs across the same four types of categories as in the Rack study: orthographically dissimilar rhyming (rude-food); orthographically similar rhyming (town-down); orthographically dissimilar nonrhyming (boil-safe); and orthographically similar nonrhyming (lost-post). Memory was assessed using a recognition test where students were shown one of the target words as a cue, and were instructed to decide which word from a set of four additional words (foils) was the word they had previously seen paired with the cue. All words were read aloud by the experimenter. The recognition test was given immediately following the rhyme judgment task. The findings indicated that on the rhyme judgment task, all students tended to make "yes" responses to pairs of visually similar words, and "no" responses to visually dissimilar ones, leading to an increase in errors on both visually similar non-rhyming pairs (such as lost-post) and visually different rhyming pairs (rude-food). Moreover, RD students showed evidence of poor phonological processing in that they were less accurate at rhyme judgment generally than non-disabled students. Performance on the recognition memory task indicated that poor readers were using a qualitatively different form of memory coding from their reading-age controls. Specifically, poor readers were more likely to remember an item matched to a visually similar non-rhyming word (e.g., post-lost) than to an item matched to a visually dissimilar rhyming word (food-rude); whereas, the opposite pattern was found for reading age controls. Moreover, while Rack (1985) suggested that RD students were relying more on an orthographic code for reading, Holligan and Johnston (1988) suggested that what was most apparent from 28 their findings was not the reliance on orthography perse but rather the difficulty with phonological processing experienced by RD students. Holligan and Johnston interpreted their findings as suggesting that although RD students were more likely to remember visually similar non-rhyming word pairs, this apparent reliance on orthography is not so much a function of the visual similarity, but rather due to the phonological distinctiveness of non-rhyming words. Clearly, there is evidence of qualitative differences in phonological and orthographic coding experienced by RD and non-RD students. RD students, perhaps due to their phonological coding problems, appear to code "something" in memory through an alternate orthographic code. These findings are consistent to those reported earlier by Lennox and Siegel (1993; 1994) regarding the preference for orthographic processing by poor readers and spellers. However, whether subtypes of reading decoders and spellers differ in the extent to which they rely on phonological or orthographic codes in memory remains to be examined. To date, one study by Swanson and Ramalgia (1992) has researched the role of phonological codes in memory and spelling across groups of RD students compared to younger students of the same reading level. This study is described next. (2) Relationship between phonological codes on memory and spelling tasks. Swanson and Ramalgia (1992) examined whether older RD students differed in the extent to which they relied on phonological information for memory and spelling tasks, compared to younger, non-RD students. Another main issue addressed by this research was whether the phonological information required for memory and spelling was correlated. RD students were matched with younger reading-age controls either on spelling and reading or on reading alone based on measures of reading decoding and spelling dictation. This grouping allowed for a comparison between students whose reading and spelling skills were equally poor, compared to those students whose spelling skills were more deficient than their reading skills. Given that research has shown (as described in an earlier section) how the processes required for reading and spelling are compatible, such a grouping controlled for the development of skills, and hence the proficiency in processing, across reading and spelling. All students were presented two lists (one for visual and one for auditory presentation) of 36 words, 12 words across each of three kinds of encoding conditions: phonetically dissimilar, but semantically related (cat, dog, mouse, bird), consonant similar (e.g.,, B_T: but, bit, bat, bet), and rhyme (un: run, fun). Memory performance was assessed by free recall and recognition tests, in that order, immediately following the presentation of word lists. Also, in order to gather information on the ways in which the phonological codes used for spelling and memory may be related, all students completed a 60- word spelling dictation test and a misspelling analysis was conducted that emphasized the extent of phonological processing used by students. Consequently, errors were scored across semi-phonetic (use of rudimentary spelling-sound 29 patterns in a word with medial and vowel sounds usually ignored, e.g., lidl for little), phonetic (over-generalization of spelling-sound patterns, e.g., new for knew; loos for lose), and morphemic (word-as opposed to sound-based spelling patterns, e.g., tier for tire; wadding for wading) criteria. The results indicated that the recall performance of RD students with comparable reading and spelling deficits was 49% on the phonologically dissimilar list, 37% on the consonant similar list, and 29% on the rhyming list. In contrast, the recall performance of RD students whose spelling deficiencies were much worse than their reading problems was 44% on the phonologically dissimilar list, and 24% and 29% on the consonant and rhyming lists, respectively. Accordingly, the more the phonological processing demands of the task increased, the worse the students with poorer spelling than reading skills performed. Furthermore, the results of the misspelling analysis revealed that phonetic errors were more common for the RD students with comparable deficits in reading and spelling, and that their performance was like the performance of the good spellers of the younger control group. Thus, the students with comparable reading and spelling deficits were at least attempting to draw on phonological information in their spellings compared to the students with more pronounced spelling deficits. Additional analyses also indicated a relationship between the phonological coding processes of memory and spelling for RD students with comparable reading and spelling deficits. However, for RD students with more deficient spelling than reading skills, no relationship was found between the phonological coding processes of memory and spelling. These findings prompted the researchers to conclude that although RD students with comparable spelling deficiencies rely on phonological codes in memory, the memory processing of RD students with poorer spelling skills appears to be independent of phonological processing. Consequently, some aspect of the asymmetry in reading and spelling skills (in this case, poor reading decoding and poorer spelling skills) suggests the possibility that an alternate code is being used to remember phonological information. This suggestion is somewhat consistent to findings reported earlier by others (e.g., Lennox & Siegel, 1993, 1994; Waters, Bruck, & Seidenberg, 1985; Waters & Bruck, 1985, 1988), however it is difficult to draw any direct comparisons given the distinctive make-up of the groups in the Swanson and Ramalgia (1992) study. For example, Waters and colleagues (Waters & Bruck, 1985; Waters, Bruck, & Seidenberg, 1985) found that good readers-good spellers, poor readers-poor spellers, and good readers-poor spellers all relied on phonological processing, albeit at varying degrees of precision, but that older students (11 -year olds) with asymmetrical reading and spelling skills tend also to rely on visual processing in their spellings (Waters & Bruck, 1988). The fact that Swanson and Ramalgia (1992) found that students with an asymmetry in deficient reading and spelling skills seemed to rely on some other code besides phonology in memory and spelling may indicate that 30 an alternate code-- an orthographic code-- was being used by these students, consistent with earlier reading research as reported previously (e.g., Holligan & Johnson, 1988; Rack, 1985; Shankweiler, Liberman, Mark, and Fowler, 1979). The link between the reliance on orthography in spelling and memory across subtypes of reading decoders and spellers has yet to be established, however. F. Summary of Background Research and Overview of the Present Study According to the research reviewed, there are two routes- - phonological and orthographic-- involved in spelling production. These routes are separable, but interactive and reciprocal in the non-brain impaired population. In addition, development of spelling skill is viewed more from contemporary theory as emerging in a continuous manner with greater proficiency across phonological and orthographic processes with experience. It has also been established that the processes for reading and spelling are compatible and interactive and that children move between reading and spelling to gain greater proficiency with skills across both tasks. The research on the compatibility between reading and spelling processes is crucial in that it tells us that spelling performance should be examined in relation to reading. Finally, there is a discrepancy in findings reported in the literature over the precise nature of spelling subtypes, particularly the processing characteristics of the good reader but poor speller (i.e., mixed speller). The most current evidence on this issue, presented by Lennox and Siegel (1993, 1994) and earlier by Bruck and Waters (1988) points to qualitative differences in phonological and orthographic processing between spelling subtypes. Moreover, research in the area of reading disabilities has indicated some differences between good and poor readers in terms of phonological and orthographic codes in memory. Whether the qualitative differences in phonological and orthographic processing between subtypes of readers and spellers can be extended to show a relationship between these processes and memory was examined by the present study. Students were initially identified by their teachers as good at reading decoding and spelling (good), poor at both reading decoding and spelling (poor) and good at reading decoding, but poor at spelling (mixed). To verify subtype classifications, these students then completed standardized reading decoding and spelling dictation measures with subtype classification based on the reading and spelling test scores. Students who met the criteria for each subtype also completed a series of tasks including rhyme judgment, cued recall, pseudoword reading, orthographic awareness, and reported on their spelling strategies. A misspelling analysis based on Bruck and Waters (1988) criteria was also conducted to examine the phonological and visual accuracy of students' misspellings. Of primary interest was how subtypes of readers and spellers would differ across the phonological and orthographic measures. 31 G. Hypotheses The main question addressed by the present study was how wil l the performance of reading and spelling subtypes differ on several tasks of phonological and orthographic processing? If good and mixed readers and spellers have well developed phonological processing skills, as research has indicated (e.g., Frith, 1980; Lennox & Siegel, 1993, 1994) then their performance on the tasks assessing phonological processing will be higher than the performance of the poor readers and spellers, due to a phonological core deficit (Stanovich, 1989), on the same tasks. These tasks include: making more rapid and more accurate rhyming decisions, remembering more phonologically similar words, reading more pseudowords accurately, making more phonologically accurate misspellings, and reporting the use of more phonological strategies for spelling than the poor readers and spellers. If the mixed readers and spellers are not distinguishable from the poor readers and spellers as some research has indicated (e.g., Waters & Bruck, 1985) then no differences will be detected between the mixed and poor readers and spellers on the tasks assessing phonological processing. If poor readers (i.e., reading disabled) and poor readers and spellers have better developed orthographic than phonological processing skills, as research has indicated (e.g., HoUigan & Johnston, 1988; Lennox & Siegel, 1993, 1994; Rack, 1985; Siegel, Share, & Geva, 1995) then the poor readers and spellers in the present study will show a reliance on an orthographic code in memory, will make more visually than phonologically accurate misspellings, and will report the use of visual rather than phonological strategies for spelling. No differences will be detected on these tasks if poor readers and spellers experience comparable difficulties with both phonological and orthographic processing. If mixed readers and spellers, due to their reliance on partial cues for reading and spelling as Frith (1980) has suggested, are attaining incomplete information about the visual appearance of words, then the mixed readers and spellers will remember fewer visually similar words on the cued recall task, will produce fewer visually than phonologically accurate misspellings, will report less of a reliance on visual strategies for spelling, and will have difficulties detecting the correct visual patterns of words as assessed by the orthographic awareness task compared to the good and poor readers and spellers. 32 CHAPTER 3 METHODOLOGY A. Overview The study consisted of two phases. The first phase involved an initial screening for reading and spelling group membership based on the administration of two subtests (reading and spelling) from a standardized measure. The second phase required the completion of phonological, orthographic, memory, and strategy tasks by those students screened as meeting criteria for each reading and spelling ability group. B. Participants A total of 61 grade 4 and 5 students from six urban schools for whom English was their primary language participated in the study. No students with sensory, physical, or intellectual disabilities participated in the study. Students were nominated by their teachers as having good reading decoding and spelling skills; poor reading decoding and spelling skills; or good reading decoding, but poor spelling skills. Teachers were specifically requested to choose students who were good or poor readers in that students had a good or poor grasp of letter-sound relationships. Teachers were also requested to choose students who were good or poor spellers based on spelling performance as reflected in students' writing samples and on dictated spelling tests. Teachers were also informed of the specific percentile cut-offs that would be used to define subtype groups. Students were selected from both grades 4 and 5 because developmentally they have attained some degree of flexibility in strategy use across phonological and orthographic processes (e.g., Marsh, Friedman, Welch, & Desberg, 1980). The sample included 59% boys and 41% girls. Moreover, 48% of the total sample were grade 4 students while 52% were students from grade 5. All participation was voluntary and written parental consent was obtained for each child (See Appendix A). All students completed the first phase of the study which involved the administration of screening measures described next. The author administered all tasks in a non-blind testing format. Table 1 presents the number of students nominated by their teachers across each of the three subtype groups. Table 1 Initial Composition of Subtypes as a Function of Sex and Grade Sex Grade Total boys /girls 4/5 good readers/good spellers (2) (7) (5) (4) 9 poor readers / poor spellers (20) (6) (14)(12) 26 good readers/poor spellers (15) (11) (10) (16) 26 33 (1) Screening Measures. The 61 students nominated by their teachers were individually administered the reading and spelling subtests from the Wide Range Achievement Test-Third Edition (WRAT-3) as the first phase of the study. The administration of the reading and spelling subtests from the WRAT-3 served to confirm teacher nominations and provided a screening instrument to identify students comprising the reading and spelling ability subtypes. Certainly, a main criticism with research conducted in the area of reading and spelling subtypes is the lack of comparability between groups (e.g., reading comprehension used to define reading ability; different measures used to define reading and spelling, etc.). Therefore, the choice of measures and the scoring criteria used for selecting reading and spelling subtypes was based on the research conducted by Lennox and Siegel (1993, 1994), as described in more detail below. a. WRAT-3 Reading subtest. Students were individually asked to read a list of words of increasing difficulty. Testing took place in a quiet room in the school. The test's standardized procedure for starting and stopping points was followed and responses were scored according to the scoring procedures described in the WRAT-3 test manual. Reliability estimates as reported in the manual for this subtest are high, with coefficients in the upper .80's and .90's across items, ages, and forms. Evidence supporting the content and construct validity of the reading skills assessed is also reported. b. WRAT-3 Spelling subtest. After completing the reading subtest, students were individually asked to produce, in writing, the correct spellings of increasingly more difficult words. Words were orally presented by the examiner once in isolation, then in the context of a sentence, then again in isolation. As with the reading subtest, starting and stopping points were determined based on standardized administration procedures and items were scored based on procedures described in the test manual. Estimates for reliability are also high for the Spelling subtest with coefficients in the upper .80's to .90's across items, ages, and forms. Evidence supporting the content and construct validity of the spelling skills assessed is also reported. (2) Selection of Groups. Based on scores obtained on the Reading and Spelling screening measures, students were classified as 1) good readers and spellers (good decoders-good spellers) if they scored at the 35th percentile or higher on both the reading and spelling subtests, 2) poor readers and spellers (poor decoders-poor spellers) if they scored at the 25th percentile or lower on both the reading and spelling subtests, and 3) mixed readers and spellers (good decoders-poor spellers) if they scored at the 35th percentile or higher on reading and at the 25th percentile or lower on spelling. Average performance falls within the 25th and 75th percentiles. These cut-offs were used to maintain comparability between past research and the composition of the reading and spelling subtypes of the present study. Moreover, the fact that only nine students were nominated as good readers and spellers by their teachers is due to the 34 sampling procedure used. Twenty students in the good group were identified quickly when assessed with the WRAT-3 and to ensure relatively equal sample sizes across subtypes, teachers were subsequently asked to nominate poor and mixed subtypes. Based on their performance on the reading and spelling measures, the final sample consisted of 50 students, 56% boys and 44% girls, 42% from grade 4 and 58% from grade 5. Eleven students were excluded from the final sample because their scores fell outside of the specified percentile ranges1. Table 2 presents the number of students in the final sample across each of the three subtype groups. Table 2 Final Composition of Subtypes as a Function of Sex and Grade Sex Grade Total boys /girls 4/5 good readers/good spellers (8) (12) (9) (11) 20 poor readers/poor spellers (12) (5) (8) (9) 17 good readers/poor spellers (8) (5) (4) (9) 13 C. Orthographic, Phonological and Strategy Measures Students in each of the groups completed five additional measures as the second phase of the study. These measures were selected to assess students' processing of visual or orthographic information, phonological information, and their report of the kinds of strategies used to spell words. Each measure is described in detail below including the process being assessed, the materials needed, what students were required to do, and how scores were attained. Students were assessed individually by the author in a quiet room in the school. (1) Rhyme judgment task. This task was included to assess students' ability to make decisions about whether or not pairs of words rhymed when they differed in orthographic and phonological similarity. To be successful on this task, students needed to inspect the variations in visual and phonological similarity between words. Thus, this task is thought to capture students' phonological and orthographic processing of lexical (whole word) information. The study required 64 word pairs, matched for word length, counterbalanced into 4 conditions which varied in relation to phonological and orthographic similarity (see Appendix B). The four conditions were: Rhyme only (coal-pole): Rhyming and visually different Rhyme-orthographic (weed-need): Rhyming and visually similar Word orthographic (wash-cash): Nonrhyming and visually similar Dissimilar (best-card): Nonrhyming and visually different 1The mean score for reading was 38.54 (SD = 11.57) and for spelling 25.0 (SD = 9.81) for these students. 35 All words, with the exception of a few changes due to pronunciation differences (i.e., British English pronunciation as opposed to North American) were adapted from Holligan and Johnston (1988) who used this methodology to examine differences in phonological and orthographic processing between reading disabled and non-reading disabled students. Words were also of medium frequency, based on Carroll, Davies, and Richman (1972) norms, and all word pairs were semantically unrelated. In addition, four grade 5 classroom teachers reviewed the lists and verified that they were words with which most fourth and fifth graders were familiar. Although three grade 4 teachers were asked to comment on the appropriateness of the lists for fourth and fifth graders, none of these teachers provided feedback. In order to capitalize on the reciprocal use of orthographic and phonological processes across students, word pairs were presented together visually (i.e., displayed on the computer screen) and read aloud by the experimenter at the same time. The word pairs were randomized and presented successively in lower case on the monitor of a Macintosh Powerbook computer. Students were introduced to the task by being asked to give their own examples of rhyming words. They were then given a chance to practice identifying rhyming words with word pairs not included in the main experiment. Word pairs appeared on the computer and at the same time were read aloud by the researcher. Students were instructed, "You're going to see two words appear on the screen and you have to decide, as quickly and as accurately as you can, whether the two words rhyme. If you think they do, then press the "yes" key, if you think they don't then press the "no" key." (see Appendix C). "Yes" and "no" response keys were clearly labelled for students. The practice set was administered until students achieved 100% accuracy; students then continued with the rest of the task. The entire task, including the practice set, took about 15 to 20 minutes to complete. Words stayed on the screen until students had made a response and a measure of response latency was recorded in milliseconds. The computer recorded the accuracy and latency of student responses and raw scores were calculated for each student across each condition for correct responses only. For each condition, accuracy scores represented the number of items for which students made a correct decision regarding whether or not word pairs rhymed. Latency scores reflected the amount of time students took to make a decision regarding whether word pairs rhymed. Thus, a total of eight scores were achieved. (2) Cued-recall memory task. One way to assess how students processed the word pairs and their attention to the phonological and orthographic qualities of the words is to ask students to remember as many of the word pairs as they can. Thus, cued recall, as opposed to free recall, was selected as the memory assessment method for one critical reason-- the encoding conditions lend themselves well to a cued-recall task. Since encoding tasks consisted of word 36 pairs, it was critical for memory assessment generally and cued-recall performance in particular (e.g., Watkins & Gardiner, 1982) that encoding and retrieval conditions were compatible. For the cued memory task, students were given one word from a previous pair they had seen on the rhyme judgment task and were asked to recall the appropriate paired target word. Cue words consisted of half of the targets from the rhyme judgment tasks-- 32 cues were randomly selected, 8 cues from each of the 4 presentation conditions. Cues were presented to students in a counterbalanced order, visually on screen and at the same time read aloud by the examiner. Immediately following the rhyme judgment task, students were informed that they would now see only one of the two words on the computer screen and that they should "Try and remember the other word that went with it." They were also instructed not to worry if they did not remember many since they were not asked to remember the words at the start of the rhyme judgment task (See Appendix D). Each word appeared in the center of the screen in the same font and size as presented in the rhyme judgment task and remained on the screen until students generated a response. Students were encouraged to "just guess" if they could not remember the appropriate word that was seen earlier with the target during the rhyme judgment task. Students responded orally and the examiner recorded the responses by hand. In keeping with the importance of context effects on cued recall performance (e.g., Watkins and Gardiner, 1982) students completed the memory task on the lap top computer and in the same order by which they completed the rhyme judgment task--thereby equating retention intervals between tasks and students. This task took about ten minutes to complete. For each student, scores were calculated for the number of correct words recalled across each of the four different conditions. Thus, total scores on this task reflected the number of words students remembered from the rhyme judgment task across the four conditions varying in visual and phonological similarity. (3) Pseudoword reading task. In order to assess the use of phonological processing, students were asked to "read" or decode a series of nonwords. A total of fifty nonwords were used from the Word Attack subtest of the Woodcock Reading Mastery Test (WRMT, Woodcock, 1987) (See Appendix E). Psychometric data are not reported as this test was not administered in a standardized manner and norms were not used. Following the cued recall task, students were shown the list of nonwords including two practice trials and were instructed to "Go ahead and read these words as if they were real words. Do the best that you can and if you're not sure just guess." Students attempted to read all of the words and the researcher recorded responses on a record sheet. This task took about five minutes to complete. Correct responses matched the pronunciation requirements described in the test manual. Raw scores were calculated for each student reflecting the number of correctly pronounced nonwords. 37 (4) Orthographic awareness task. To assess students' use of orthographic or visual processing, students were presented a series of word pairs following the pseudoword reading task, and were asked to decide which one of the pair "Looks like a real word or could be a real word" (See Appendix F). This is the task used by Siegel, Share, and Geva (1995) and required 17 pairs of pronounceable nonwords. As Siegel et al: described, one word from each pair contained a bigram (i.e., two consecutive letters) that never occurs in English (e.g., wolg) and the other word from the pair contained an orthographically legal bigram (e.g., wolt). This task took about three minutes to complete. The researcher recorded students' responses and raw scores were calculated for correct answers. Scores thus reflected the number of words on which students made accurate judgments about the word's visual similarity to actual English words. (5) Strategy reports. Following the orthographic awareness task, students were asked to report on the kinds of strategies they used for spelling to assess the degree that strategies of a phonological and visual nature were used for spelling. This was the final measure students completed. For each student, the last five items correctly spelled and the last five items misspelled from the previous administration of the Spelling subtest of the WRAT-3 served as the target words for a total of ten spelling words. The highest range of items was selected to control for the difficulty level for each student since items at higher levels were more difficult based on test standardization procedures. Thus, each student's strategy report was based on correct and incorrectly spelled words at comparable levels of difficulty. All of the students spelled previously correct spellings accurately on the subsequent administration. However, in cases where students spelled previously misspelled words correctly, the next misspelled item from the spelling test was used. Words were orally presented to students one at a time and in random order (correct vs. incorrect), used once in a sentence, and presented again individually (e.g., "Book. I read the book. Book"). Once students had spelled the word, they were asked to tell the researcher how they knew to spell the word that way and what they were thinking as they were spelling the word (see Appendix G) resulting in an immediate retrospective report or a "think aloud" protocol (Kail & Bisanz, 1982) . This task took about 15 to 20 minutes for students to complete. Strategy reports were tape recorded and later transcribed. Consistent with the technique employed by Steffler, Varnhagen, Friesen, and Treiman (1998), the initial coding involved 10 categories, and when several reports were given for one word (e.g., "I sounded it out, and it looks right"), the coding for each report was included. For the second phase of coding, five of the most reported strategies in terms of percentage of reporting were included. These reports were: phonetic, practice/exposure, visual, direct retrieval, and explicit rule. A phonetic strategy was coded when students reported having "sounded the word out". When students reported having "seen the word a lot" in print or 38 "spelled the word a lot" in writing the strategy was scored as practice/exposure. This strategy was not reported in the Steffler et al. study but was frequently reported by students within the present study. Visual strategies were coded when students reported that the spelling did or did not "look right" and when they spelled by analogy (e.g., "book is like look, but with a 'b' not an T"). When students reported just "knowing" how to spell the word, a direct retrieval strategy was coded, and for words that students expressed a specific spelling convention (e.g., "'i' before 'e' except after 'c'; the 'e' at the end makes the 'a' say its name") the specific rule strategy was scored. Most students reported multiple strategies for any one spelling (e.g., "I knew it and I've seen it before lots"; "It looks and sounds right") and multiple reports were coded separately. For correct and incorrect spellings, the total number of strategies reported were calculated for each student. These strategies consisted of five distinct categories, as described above. Scores for each student were recorded that reflected the proportion of strategy reports across the five categories out of each students' total report of strategies (e.g., student A reported 37% of their strategies as reflecting a phonological strategy). Thus, across students these scores represented the degree to which different strategies were relied upon by students for spelling correctly and incorrectly spelled words. D. Error Analysis Along with the scores collected on the previously described measures, an error analysis was also conducted to examine the degree of phonological and visual similarity of misspelled words to target words. This analysis provided further information on how students used phonological and orthographic processes while they spelled. The first ten spelling errors from the Spelling subtest of the WRAT-3 were analyzed for each student yielding a total of 500 items to analyze. Analysis procedures were based on the Bruck and Waters (1988) scoring system which yields scores for the phonological and visual similarity of misspelled words to target words described next. (1) Phonological similarity. Students were scored for the phonological accuracy of their misspellings if it sounded like the target word through the application of grapheme-phoneme correspondence rules. Using the same scoring method as Walton (1997) in an unpublished manuscript researching phonological development in beginning spellers, scores were calculated based on the maximum number of phonemes accurately represented and in the correct order in relation to the total number of phonemes in the words. Within this scheme, spelling "bt" for "bat" is scored as 2/3 or .67 since two of the three phonemes are represented accurately. Similarly, spelling "dres" for "dress" would yield a score of 1.00 since all four phonemes are represented accurately. Mean scores across students were calculated with a high inter-rater 39 reliability estimate, r = .92 based on the examiner's and another rater's independent assessment2. Raters resolved disagreements through discussion. (2) Visual/orthographic similarity. This system yields a score based on the percentage of bigrams (letter pairs) and individual letters that students' misspellings shared with the target word. For example, the word bat has two bigrams ("ba" and "at") and three letters for a total of five. The spelling "bt" has no bigrams and two letters that match the target word, for a total of two. Thus, the misspelling "bt" would yield a visual accuracy score of 2/5 or .40. Likewise, the word dress has four bigrams ("dr", "re", "es", "ss") and five letters for a total of 9. The spelling "dres" has three bigrams and four letters that match the target word for a total of seven. Thus the misspelling "dres" would yield a visual accuracy score of 7/9 or .77. Mean scores were calculated across students with a strong inter-rater reliability estimate, r = .96 based on the examiner's and another rater's independent assessment. Disagreements were resolved through discussion. 2The second rater was a teaching colleague, unconnected to the schools sampled, who received a full day's training in the scoring method. 40 CHAPTER 4 RESULTS A. Overview Preliminary analyses considered the identification of the three reading and spelling subtype groups. First considered was the overlap between teacher nominations and standardized test scores. The principal question addressed by this research was how reading and spelling subtypes differed in the way they used phonological and visual information in memory and spelling as well as the nature of the spelling strategies reported by students. Moreover, for each student, scores were obtained across the six measures: rhyme judgment, cued recall, orthographic awareness, pseudoword reading, strategy reports, and error analysis. The following chapter describes the preliminary and principal findings from the present study. B. Preliminary Results First, no significant differences were found between subtypes for gender and grade across the dependent measures (see Appendices H and I for means and standard deviations). Moreover, in phase 1 of the study teachers were asked to nominate those students who were good readers and good spellers (good subtype), poor readers and poor spellers (poor subtype), and good readers, but poor spellers (mixed subtype). A preliminary question of interest was the consistency of teachers' selections for subtype group membership with reading and spelling test results. Standardized reading and spelling achievement subtests were administered to these students to see if they actually met the percentile cut-off criteria for each subtype. Table 3 Consistency of Teacher Nomination with Test Results Group a Teacher n o m i n a t e d Did not meet criteria Good Poor Mixed Total Good 0 9 0 0 9 Poor 6 2 14 4 26 Mixed 5 9 3 9 26 Total 11 20 17 13 61 a Group membership based on WRAT-3 test scores As indicated by Table 3, teachers were most consistent in identifying students with good reading and spelling skills. They were the least consistent in identifying those students with mixed reading and spelling skills who were good readers but poor spellers. Although teachers anticipated that students displayed asymmetrical skills, spelling performance was not as poor as anticipated when assessed with the standardized measure and more students who were 41 nominated as having "mixed" reading and spelling skills actually were good readers and spellers when assessed with standardized measures. Again, so few good readers and spellers were nominated by their teachers due to the fact that twenty students in the "good" group were identified quickly when assessed with the WRAT-3. In order to ensure relatively equal sample sizes across subtypes, teachers were subsequently asked to nominate poor and mixed subtypes. Moreover, two students who were identified as having poor decoding and spelling skills actually had good decoding skills when assessed with the standardized measure. These discrepancies in subtype selection highlight the imprecision of teacher selection alone for identifying reading and spelling difficulties, at least for the mixed subtype. However, this imprecision is likely due to the percentile criteria used to define subtypes rather than teachers' perceptions of subtype differences. Also of preliminary interest was how students across the three subtypes scored on the phase 1 reading and spelling standardized measures (WRAT-3). Table 4 lists the mean scores (measured in percentiles) and standard deviations across grade and subtype. Scores within the 25th to 75th percentiles are average. Ranges are presented in parentheses. Table 4 WRAT-3 Scores (percentiles) as a Function of Grade and Subtype Reading Spelling M SD M SD Good readers/Good spellers 4 * 80.1(50-99) 17.8 66.4(37-99) 18.8 5 ' 72.4(55-91) 10.8 70.6(37-93) 19.5 Poor readers / Poor spellers 4 .16.7(9-23) 5.9 .17.8(10-23) 6.0 5 14.8(5-23) 6.5 14.1(7-23) 7.1 Good readers / Poor s pellers 4 56.0(37-75) 19.2 21.7(18-23) 2.5 5 56.5(37-90) 17.0 19.2(21-23) 1.9 Two one-way ANOVAs, one for reading and one for spelling performance, were conducted to verify that subtypes differed significantly on the reading and spelling subtests. Results indicated significant differences between groups for the reading F [2, 47J = 95.155 p < .000 and spelling F (2, 47} = 98.858 p < .000 measures. Tukey post hoc analyses indicated that all subtypes were distinguishable on the basis of their WRAT-3 spelling test scores. As presented in Table 3, the good subtype achieved significantly higher scores than the mixed and poor subtypes whose scores did not significantly differ. Tukey HSD post hoc analyses also indicated that although the good and mixed subtypes achieved significantly higher reading scores than the poor subtype, the good subtype also achieved significantly higher scores than the mixed subtype. Thus, it appears that although the subtypes included students who were good or poor 42 spellers, it also included students who were very good (good subtype), good (mixed subtype) and poor (poor subtype) readers. Implications for this composition of groups will be considered in the interpretation of results discussed in Chapter 5. C. Principal Analyses Of primary interest was how the three reading and spelling groups (i.e., good, poor, mixed) varied in the degree to which they relied on orthographic and phonological processes. Accordingly, a series of repeated measures and one-way analyses of variance were conducted and significant effects were followed up with least significant difference (LSD) comparisons for repeated measures analyses and Tukey HSD post hoc analyses for one-way ANOVAs. The following section describes findings across reading and spelling groups and the five categories of dependent measures. 1) Rhyme judgment task. A main question motivating the present research was whether subtypes of readers and spellers differed in their accuracy and response time in making rhyming decisions on words differing in visual and phonological similarity. To briefly review, words were rhyming and visually different (rhyme only; coal-pole), rhyming and visually similar (rhyme-orthographic; weed-need), non-rhyming and visually similar (word orthographic; wash-cash), and non-rhyming and visually different (dissimilar; best-card). Differences between subtypes may indicate the reliance on different processes (e.g., visual vs. phonological). It was anticipated that due to their better phonological processing skills, students with good reading and spelling skills and good reading but poor spelling skills (i.e., mixed subtype) would generally be quicker and more accurate at making decisions about whether words rhymed across word type conditions compared to students with poor reading and spelling skills. To test this hypothesis, a series of two 3 X 4 repeated measures ANOVAs were conducted with subtype as the between subjects factor and response time and accuracy as the within subjects factors for different word types. For response time, results indicated a main effect for word type F(3,46) = 28.34 p < .01. There was no significant group by response time interaction indicating that all subtype groups took as long to make rhyming decisions across the four word types. A test of the simple main effects using least significant difference (LSD) comparisons with Bonferroni correction indicated that all students were significantly slower responding to the word-orthographic (wash-cash) words (M = 2.76 SD = .87) than to the rhyme-orthographic (weed-need) (M = 2.50 SD = .93), the rhyme only (pole-coal) (M = 2.17 SD = .71) and the dissimilar (best-card) words (M = 2.24 SD = .67). Moreover, although students were significantly slower responding to the rhyme only words (pole-coal) (M = 2.17 SD = .71) than to the rhyme-orthographic (weed-need) (M = 2.50 SD = .93), and dissimilar (best-card) words (M = 2.24 SD = .67).words, they were also significantly slower responding to the dissimilar words than to the 43 rhyme-orthographic words. Thus, students required the most time to make a rhyming decision for words that did not rhyme but looked the same, possibly confused by the words' visual similarity and they made the fastest decisions on words that were phonologically and visually similar or that were distinctive with no phonological or visual similarity. There was also a significant main effect of word type for accuracy of deciding whether word pairs rhymed F (3,46) = 20.128 p < .01 modified by a significant group by word type interaction F (6,43) = 6.488 p < .01 on the accuracy of students' responses. Table 5 shows the means and standard deviations for the reading and spelling subtypes on their accuracy of responses across the word conditions varying in phonological and orthographic similarity. Accuracy is based on raw scores with a maximum of 16 correct across each word type. Table 5 Means and Standard Deviations for Subtypes on Response Accuracy for the Rhyme Judgment Task Word Type Group M SD rhyme only good 15.55 .82 (coal-pole) poor 12.53 3.92 mixed 14.46 3.61 rhyme -orthographic good 16.00 .00 (weed-need) poor 15.12 1.72 mixed 15.77 .44 word orthographic good 14.5 1.47 (wash-cash) poor 8.82 5.62 mixed 14.46 2.26 dissimilar good 15.75 .55 (best-card) poor 15.29 .85 mixed 15.85 .38 In follow-up analyses, four one-way ANOVAs were conducted with Bonferroni correction to determine the nature of the differences between groups for their accuracy of responses for each word type. Specifically, these analyses examined variations across reading and spelling subtypes for each of the four word types. As shown in Table 5, the results indicated the most notable difference between groups was for the accuracy of responses for visually similar, non-rhyming words (e.g., wash-cash) F (2,47) = 13.884 p < .001. Tukey HSD post hoc analyses revealed that the good and mixed subtypes were significantly more accurate responding to the phonologically similar words (coal-pole; weed-need) than the poor subtype, as anticipated due to the better phonological processing skills of the good and mixed subtypes. Rhyme judgment accuracy for phonologically similar words did not differ significantly between the good and mixed subtypes. Tukey HSD post hoc analyses also revealed that good and mixed subtypes 44 were more accurate on the visually similar, non-rhyming words (wash-cash) than the poor subtype. The poor subtype was therefore more likely to mistakenly respond that two visually similar non-rhyming words rhymed (wash-cash) than the other subtypes, showing a tendency to over-rely on the visual appearance of these words for rhyming decisions. Additional analyses were conducted to examine how the accuracy of each subtype group differed, relative to their own performance, across the four word types. Specifically, a one factor repeated measures ANOVA was conducted for each group with word type as the repeated measure. The results indicated significant differences for the good F (3, 47) = 12.75 p_ < .001 and poor F (3, 47) = 13.68 p < .001 subtypes only. The mixed subtype's accuracy did not differ significantly across the four word types. Follow-up least significant difference (LSD) comparisons with Bonferroni correction indicated that good and poor readers and spellers showed a similar pattern of accuracy across word types, relative to their own performance. For example, and as shown in Table 5, good and poor subtypes were more accurate on rhyme-orthographic (weed-need) than rhyme only (coal-pole) words, more accurate on rhyme only than word orthographic (wash-cash) words, more accurate on rhyme-orthographic than word orthographic words, and more accurate on dissimilar (best-card) than word-orthographic words. Thus, good and poor subtypes were each making relative use of the phonological rather than the orthographic similarity of word types in making accurate rhyming decisions. Importantly, the poor subtype appeared to be using phonology in making their rhyming decisions, but were not as effective as the good subtype. Moreover, the poor readers and spellers were more accurate on the dissimilar (best-card) words than on the rhyme-only (coal-pole) words suggesting that it may have been easier for the poor subtype to make a distinctively no rhyme compared to a distinctively rhyme decision. (2) Cued recall memory task. Of primary interest was how subtypes of readers and spellers would differ in their recall of words across the four word types, providing more evidence of the kinds of processes relied upon by different subtypes of readers and spellers. Differences were anticipated between reading and spelling groups on the recall of certain types of word pairs. Specifically, it was expected that the poor subtype would remember fewer phonologically similar word pairs compared to the good and mixed subtypes due to their weaker phonological processing skills. It was also hypothesized that the poor subtype would remember more visually similar than phonologically similar words, consistent with previous research suggesting greater reliance on orthographic cues in memory (e.g., Holligan & Johnston, 1988; Rack, 1985). A critical question motivating the present research was whether there would be differences for the mixed subtype, since no research on the rhyme judgment and cued recall performance has been conducted with this group to date. If the mixed subtype has greater difficulty with visual 45 memory, as others (e.g., Lennox & Siegel, 1993, 1994) have suggested, then the mixed group was expected to recall fewer visually similar words than the good and poor subtypes. A 3 X 4 repeated measures ANOVA was conducted with group as the between subjects factor and recall performance as the within subjects factor for the different word types. Results indicated a main effect for word type F (3,46) = 69.29 p < .001 but this effect was modified by a significant interaction between reading and spelling groups and the type of word recalled F(6,43) = 3.88 p < .001. Table 6 lists the means and standard deviations for recall performance (based on raw scores with a maximum of 16 correct per word type) on words differing in visual and phonological similarity across subtypes. Table 6 Means and Standard Deviations for Subtypes on Recall Performance Word Type Group M SD rhyme only good 2.55 1.90 (coal-pole) poor 1.94 1.30 mixed 1.62 1.50 rhyme - orthographic good 5.55 2.43 (weed-need) poor 2.88 1.22 mixed 3.38 1.76 word orthographic good 1.90 1.65 (wash-cash) poor 1.41 1.32 mixed .615 .77 dissimilar good .20 .70 (best-card) poor .06 .24 mixed .08 .28 To examine the nature of subtype differences across the four word conditions, four one-way ANOVA's with Bonferroni correction were conducted. Specifically, these analyses considered whether the good, poor, or mixed subtype groups differed in their ability to recall each of the four different word types. As the means in Table 6 indicate, results revealed significant differences between subtype groups for the recall of the rhyming, visually similar (rhyme-orthographic) words (e.g., weed-need) F (2,47) = 9.99 p < .01 and the non-rhyming, visually similar (word orthographic) words (e.g., wash-cash), F (3,46) = 3.51 p < .01. Follow-up Tukey HSD post hoc analyses revealed that good readers and spellers recalled more rhyme-orthographic (e.g., weed-need) word targets (M = 5.55, SD = 2.43) than the poor readers and spellers (M = 2.88, SD = 1.23) or the mixed subtype of good readers, poor spellers (M = 3.38, SD = 1.76). Tukey post hoc analyses also indicated that the mixed subtype recalled 46 significantly fewer word-orthographic (e.g., wash-cash) (M = .62, SD = .77) words than the good (M = 1.90, SD = 1.65) subtype. Good and poor subtype groups did not differ in their recall of word-orthographic words. In order to examine how recall performance differed across the four word types for each of the reading and spelling groups, an additional one factor repeated measures ANOVA was conducted for each subtype group with word type as the repeated measure. Results indicated significant differences between word types for the good F (3, 47) = 40.202 p < .0001, poor F(3, 47) = 57.955, p < .0001, and mixed F(3, 47) = 29.841 p< .0001 subtypes for recall performance. Follow-up comparisons using the least significant difference (LSD) with Bonferroni correction indicated that mixed readers and spellers recalled more rhyme only (coal-pole) than word orthographic (wash-cash) words. Apparently the phonological similarity was more salient to the mixed subtype than the orthographic similarity or, conversely, the mixed subtype did not use the orthographic similarity effectively in memory consistent with previously described results. Good and poor readers and spellers recalled as many rhyme only (coal-pole) words as word orthographic (wash-cash) words indicating that for these subtypes both the phonological and orthographic characteristics of words were equally salient. This finding also provides some evidence for the capacity of the poor group to make use of phonological as well as orthographic information in memory, despite their observed phonological processing difficulties making accurate rhyming decisions, as previously described. (3) Orthographic Awareness Task. This task was included to further examine students' use of orthographic or visual processing. On the orthographic awareness task, students were shown a list of word pairs and asked to identify the word that "Most looks like a real word or could be a real word" from the pair. This task is considered to assess students' attention to and processing of the visual characteristics of words. A one-way ANOVA with subtype as the independent variable and raw scores on the orthographic awareness task as the dependent variable was conducted. Consistent with previous results (e.g., Lennox & Siegel, 1994) it was anticipated that students with good reading and spelling skills would perform significantly better than students with poor and mixed reading and spelling skills. However, results indicated no significant differences between subtype groups on the orthographic awareness task. All students achieved comparable scores on this task, regardless of reading and spelling subtype, as the means and standard deviations in Table 7 indicate. 47 Table 7 Performance (total correct) on the Orthographic Awareness and Pseudoword Reading Tasks Across Subtypes Orthographic Pseudoword Reading Awareness M SD M SD Good readers /Good spellers 15.1 1.4 42.2 9.6 Poor readers/Poor spellers 14.4 1.7 26.4 9.3 Good readers/Poor spellers 14.1 1.3 36.9 6.1 (4) Pseudoword reading task. Students were required to read a list of pseudowords using appropriate letter-sound correspondence rules in order to assess students' phonological processing. A one-way ANOVA with subtype as the independent variable and raw scores (i.e., number correct) from the pseudoword reading task as the dependent variable was conducted. The results indicated significant differences between the reading and spelling groups F (2,47) = 15.37 p < .001. Tukey HSD post hoc analyses indicated that good and mixed readers and spellers performed better than poor readers and spellers (see Table 7 for means and standard deviations) in using phonological processing skills to decode pseudowords. consistent with previous results (e.g., Lennox & Siegel, 1994). No significant differences were detected between good and mixed readers and spellers on this task. (5) Error Analysis. Students' misspellings from the spelling subtest of the WRAT-3 were analyzed for the degree to which misspellings approximated the correct visual and phonological characteristics of the words. Two one-way ANOVA's were conducted, with Bonferroni correction, for the visual and phonological accuracy of misspellings across the reading and spelling subtypes. Subtype was the independent variable and visual or phonological accuracy mean scores, respectively, were the dependent variables. The results indicated significant differences between subtype groups on both the visual, F (2,47) = 6.68 p < .003, and phonological, F (2,47) = 26.49 p < .001, similarity of word spellings to target spelling words as indicated by the means and standard deviations in Table 8. Table 8 Mean Percentages of Visual and Phonological Accuracy of Misspellings Across Subtypes Visual Accuracy .Phonological Accuracy M SD M SD Good readers /Good spellers 67.4 4.8 91.0 3.8 Poor readers/Poor spellers 61.9 6.5 79.5 6.0 Good readers/Poor spellers 68.2 4.4 88.9 4.6 48 Tukey HSD post hoc analyses revealed that the good and mixed readers and spellers made more visually accurate misspellings than the poor readers and spellers. Although the poor readers and spellers relied on orthography in memory, they did not rely on this process for spelling words. There were no differences between the good and mixed readers and spellers on the visual accuracy of their misspellings. Post hoc analyses also indicated that the good and mixed readers and spellers produced significantly more phonologically accurate misspellings than the poor subtype. There were no differences for the phonological accuracy of misspellings between the good and mixed subtypes. (6) Verbal report and strategy data. The inclusion of a self report measure provided critical additional data, from the students themselves, on the kinds of strategies they used to spell words. Of primary interest was whether there were any differences between the reading and spelling subtypes on reports of strategies used for correct and incorrect spellings. As Steffler, Varnhagen, Friesen, and Treiman (1998) also observed in their study, the students in the present study were motivated to report the kinds of strategies they used to spell and had no difficulty communicating their spelling behaviour. A number of different strategies were reported and coding of verbal reports was conducted in two stages, as described in Chapter 3. To review, for correct and incorrect spellings, the total number of strategies reported were calculated for each student. These strategies comprised the five distinct categories as presented in Table 8. Scores for each student were recorded that reflected the proportion of strategy reports across the five categories out of each students' total report of strategies (e.g., student A reported 37% of their strategies as reflecting a phonological approach). Results will be presented separately for correct and incorrect spellings. a. Correct Spellings. A 3 X 5 repeated measures ANOVA was conducted for correct spellings with subtype as the between subjects factor and strategy type as the within subjects factor. Although there was no subtype by strategy interaction, indicating that all students reported a similar proportion of strategies across categories, there was a main effect for the kinds of strategies reported F (4,45) - 16.52 p < .001. Table 9 show the mean percentages and standard errors of strategies reported by students for their correct spellings. 49 Table 9 Mean Percentages of Reports of Specific Strategies for Correct Spellings Spelling Strategy M SE Phonological 25.65 2.82 Practice 28.89 3.37 Visual 13.26 2.39 Direct Retrieval 28.15 3.01 Rule 3.07 1.48 A test of the simple main effects using least significant difference (LSD) comparisons with Bonferroni correction indicated that all students reported a significantly greater use of a phonological and an exposure/practice strategy than a visual or rule-based strategy for spelling words correctly. They also reported significantly more direct retrieval than phonological or visual strategies, and significantly more visual than rule-based strategies. No significant differences were found between their report of using an exposure/rule strategy and a phonological or direct retrieval strategy. Thus, for spelling words correctly, students frequently reported three main approaches: sounding the word out, having seen or spelled the word before, and just knowing how to spell the word. b. Misspellings. Another 3 X 5 repeated measures ANOVA was conducted with subtype group as the between subjects factor and proportion of specific strategies reported across misspellings as the within subjects factor. Although there was no subtype by strategy interaction, indicating that all students reported a similar proportion of strategies across categories, there was a main effect for the kinds of strategies reported F (4,45) = 73.62 p < .001. Table 10 shows the mean percentages and standard errors of reported strategies for misspellings. Table 10 Mean Percentages of Reports of Specific Strategies for Misspellings Spelling Strategy M SE Phonological 55.78 3.13 Practice 8.74 1.77 Visual 28.98 1.25 Direct Retrieval 1.59 1.25 Rule 5.44 1.57 50 Follow-up least significant difference (LSD) comparisons with Bonferroni correction indicated, as presented in Table 10, that students reported using a phonological strategy significantly more than any other strategy for words they misspelled. They also reported using a visual strategy significantly more than direct retrieval, practice/exposure, and rule-based strategies. Thus, students were more likely to report sounding the word out, or that the word "didn't look right" for spelling difficult words. Despite their difficulties with phonological processing, poor readers and spellers appear to be attempting to use phonological approaches to spell difficult words. D. Summary Preliminary results indicated that teacher nominations of subtypes of readers and spellers overlapped moderately with standardized test results confirming subtype groups. Nominations were most consistent with test results for the good readers and spellers and the least consistent for the poor and mixed readers and spellers. Moreover, the criterion used to distinguish readers and spellers was found to discriminate poor spellers adequately. Specifically, the poor spellers attained significantly lower scores than the good and mixed subtype groups who were not distinguishable in terms of spelling ability. Although the reading measure discriminated the poor from the good and mixed subtype groups (i.e., the poor readers and spellers achieved significantly lower reading scores) the good readers and spellers also achieved significantly higher reading scores than the mixed readers and spellers. Thus, subtype groups consisted of very good readers and good spellers, good readers and good spellers, and poor readers and poor spellers. On the rhyme judgment task, all students were generally slower in responding to word-orthographic words (wash-cash). They were also slower responding to rhyme only (coal-pole) than rhyme orthographic (weed-need) and dissimilar (best-card) words. Good and mixed readers and spellers made more accurate responses to the phonologically similar words (coal-pole; weed-need) than the poor readers and spellers and there were no accuracy differences between the good and mixed subtypes on these phonologically similar words. Good and mixed readers and spellers also made more accurate responses to the non-rhyming, visually similar words (wash-cash) than the poor readers and spellers. In examining how each subtype's accuracy performance differed due to the type of word presented, good and poor readers and spellers showed a similar pattern of accuracy across word types, relative to their own performance. Specifically, these subtype groups made more use of the phonological rather than the orthographic similarity of word types in making accurate decisions suggesting that poor readers and spellers do attempt to use phonology, but are not as effective (based on their lower overall accuracy) as the good readers and spellers in applying this strategy. Mixed readers and spellers 51 showed no differences in the relative use of phonology or orthography in making accurate rhyme decisions. On the cued recall task, good readers and spellers recalled more rhyming, visually similar words (weed-need) than the mixed or poor readers and spellers. An important finding was that good and mixed subtypes were distinguished by the mixed readers and spellers' poorer recall of non-rhyming, visually similar (wash-cash) words. Good and poor readers and spellers were similar in their recall performance on these words. Each subtype's recall performance also differed due to the type of word they were required to recall. For example, mixed readers and spellers recalled more rhyming, visually different (coal-pole) words than non-rhyming, visually similar (wash-cash) words, perhaps due to the greater saliency of phonology or conversely the reduced saliency of orthography for this subtype. It is also interesting that good and poor readers and spellers recalled as many rhyming, visually dissimilar (coal-pole) words as non-rhyming, visually similar words (wash-cash) suggesting that for the good and poor subtype groups, both the phonological and orthographic characteristics of words were equally as memorable despite the poor subtype's difficulties using phonology to make accurate rhyme decisions. Poor readers and spellers performed as well as the good and mixed readers and spellers on the orthographic awareness task suggesting that the poor subtype group possesses intact memory for the appropriate appearance of words. However, the good and mixed readers and spellers performed better than the poor readers and spellers reading pseudowords reflecting the less well developed phonological processing skills of the poor group. No differences were found between the good and mixed readers and spellers for reading pseudowords suggesting well developed phonological processing skills for these subtype groups. The examination of spelling errors and student report of spelling strategies indicated some differences between subtype groups. Specifically, good and mixed readers and spellers made more visually and phonologically accurate misspellings than the poor readers and spellers. There were no differences between the good and mixed readers and spellers for the phonological or visual accuracy of their misspellings. Moreover, all students, regardless of reading and spelling subtype, reported that phonetic strategies, experience and practice with a word (either through reading or spelling), and direct retrieval helped them spell words correctly. Similarly, all students reported that they used a phonetic strategy proportionately more often to spell difficult words, followed by a visual strategy. Thus, no differences were detected between the subtypes in their report of spelling strategies. The present results have provided evidence for processing differences between subtypes of spellers; namely, that good and mixed spellers demonstrated comparable performance using phonological information in memory and spelling and that good and poor spellers demonstrated 52 similar use of orthographic information in memory. The fact that differences were found between subtypes of spellers adds to the growing research on specific subtypes and supports the notion that not all poor spellers use the dual routes of phonology and orthography the same way. The results will be discussed in the next chapter in terms of the dual route theory and in relation to the processing characteristics of reading and spelling subtypes. CHAPTER 5 SUMMARY AND CONCLUSIONS 53 A. Overview The present study examined the use of phonological and orthographic information in memory and spelling by subtypes of students who differed in reading decoding and spelling ability. These subtypes included students who were good at decoding and spelling (good), poor at decoding and spelling (poor), and good at decoding but poor at spelling (mixed). Students were given a series of tasks to assess their phonological, orthographic, and memory processing. The following chapter examines the findings presented in Chapter 4 organized around the phonological and orthographic processing of subtypes of readers and spellers in the present study. Methodological issues and directions for future research will also be presented along with educational assessment and instructional implications for spelling derived from the present study. B. Discussion (1) The use of phonological information by reading and spelling subtypes. The results of the present study showed that poor readers and spellers performed significantly more poorly than the good or mixed subtypes in most aspects of phonological processing, consistent with a wealth of previous research on the phonological processing difficulties of poor readers and spellers (e.g., Bryant & Bradley, 1980; Lennox & Siegel, 1993; Porpordas, 1999; Waters, Bruck, & Seidenberg, 1985). Although there was evidence that the poor readers and spellers attempted to use phonological skills in memory and spelling, they did not use these skills as effectively as the good and mixed readers and spellers. Poor readers and spellers were less accurate than good or mixed readers and spellers when deciding whether two similar sounding words rhymed, remembered fewer rhyming words, achieved lower scores on the pseudoword reading task, and made fewer phonologically accurate misspellings. Generally, mixed readers and spellers did not differ significantly from students with good reading and spelling skills on the phonological measures, indicating phonological processing skills that are as well-developed as students within the good subtype consistent with findings from other studies (e.g., Bruck & Waters, 1988; Frith, 1980; Lennox & Siegel, 1994). Importantly, good and mixed subtypes were not distinguishable by their use of phonological processing. (2) The use of orthographic information by reading and spelling subtypes. Remembering orthographic information was the single area that distinguished good and mixed subtypes. Mixed readers and spellers recalled fewer visually similar non-rhyming words compared to good readers and spellers who were similar to poor readers and spellers in their recall performance. 54 These results are consistent with previous studies conducted with the rhyme j udgment task and reading disabled students (e.g., Holligan & Johnston, 1988; Rack, 1985) and extends these findings to include reading and spelling subtypes. Moreover, good and poor readers and spellers showed a similar pattern of recall performance across phonologically and orthographically similar words whereas the mixed readers and spellers did not recall as many orthographically similar as phonologically similar words. Others have also reported on the mixed subtypes' pattern of processing orthographic information (e.g., Frith, 1980; Lennox & Siegel, 1994). Frith explained these processing differences by suggesting that mixed readers and spellers read by partial cues and therefore, given the relationship between the dual routes in reading and spelling, attain an incomplete visual representation of words in memory. An important contribution of the present study is the finding that although mixed readers and spellers were more accurate than poor spellers in responding to rhyming decisions on the visually similar, non-rhyming words (wash-cash) due to their better-developed phonological processing skills, they were much less successful in coding the orthographic information for subsequent recall. Poor readers and spellers, despite making inaccurate rhyme decisions, used this orthographic information effectively in memory. These results are also highly consistent with Lennox and Siegel's (1993) assertion that mixed readers and spellers have similar phonological skills and more deficient visual memory skills than normally achieving students. Certainly, the mixed subtype was also indistinguishable from the good subtype on the basis of phonological processing in the present study. An important contribution of the present study, however, is the direct evidence of a deficient code in memory by mixed readers and spellers, and that poor readers and spellers share as well-developed an ability to process and remember orthographic information as good readers and spellers. The fact that the poor subtype performed as well as the good subtype in using orthographic information in memory provides evidence to support intact visual memory skills by poor readers and spellers despite deficient use of phonological information. It is particularly interesting that differences were found between subtypes on some of the other tasks assessing orthographic processing. Firstly, poor readers and spellers were less accurate than the good or mixed readers and spellers in correctly determining whether two visually similar non-rhyming words rhymed (e.g., bowl-howl). These results are consistent with results previously reported by Rack (1985) and Holligan and Johnston (1988) on the rhyme judgment accuracy of normally achieving and reading disabled students. The fact that students with poor reading and spelling skills made more errors than the good or mixed readers and spellers on the visually similar, non-rhyming words in the present study may indicate an over-reliance on orthography at the expense of phonology, as Rack suggested. This over-55 reliance may be due to the poor subtype's weak phonological skills, as Holligan and Johnston indicated, rather than to their better-developed orthographic processing skills. However, there were no differences detected on the orthographic awareness task between subtype groups, suggesting that poor readers and spellers possess an ability to identify appropriate letter patterns in words that is as well-developed as the students with good and mixed reading and spelling skills. From a dual route perspective, these findings suggest that poor readers and spellers possess as well developed visual memory skills for orthographic patterns as the good and mixed readers and spellers, supported also by the poor subtype's well developed recall of visually similar words, as described earlier. These conclusions are also consistent with reading disabled students' better-developed orthographic awareness skills, as reported by Siegel, Share, and Geva (1995). In light of the findings from the recall task, it is interesting that the mixed readers and spellers performed as well as they did on this task, given their deficient recall for visually similar words. Also, Lennox and Siegel (1994) reported that good readers and spellers performed better than the poor and mixed readers and spellers on the orthographic awareness task who were similar in their performance. It is possible that the much smaller sample size in the present study as compared to Lennox and Siegel's large sample accounted for the different findings. What is noteworthy from the present study is the similarity of performance on this orthographic processing task between the poor readers and spellers and the other subtypes, compared to the poor subtype's much weaker phonological skills. Moreover, poor readers and spellers in the present study did not rely on orthographic processing for spelling and actually made fewer visually accurate misspellings than the good and mixed readers and spellers. These results are inconsistent with findings described by Lennox and Siegel (1993) who reported that by third grade continuing through to grade six, the poor readers and spellers tended to rely on orthographic processing for spelling difficult words as indicated by a greater proportion of visually accurate than phonologically accurate misspellings. Good readers and spellers reportedly made more phonologically accurate than visually accurate misspellings. Sample sizes were very different between the Lennox and Siegel study and the present study which may account, in part, for the disparate findings. It is also possible, due to the reciprocal nature of reading and spelling development as Ehri (2000) has described, that poor readers and spellers are exposed to fewer words generally in print (by reading less) and fewer "correct" visual representations of words (due to their misspellings). Their lack of reliance on visual approaches to spelling unfamiliar words may therefore be the direct result of their lack of exposure to print in general. The assessment of strategies reported by students while they spelled included the important component of asking students to report not only on their misspellings, but also how they correctly spelled words as others have suggested (e.g., Lennox & Siegel, 1994; Steffler et al., 5 6 1998). Students in the present study emphasized the use of different strategies depending on whether or not the word was spelled correctly. Specifically, all students reported three main approaches to spelling words correctly: they sounded the word out (e.g., phonological strategy), they had read or spelled the word before (e.g., practice or exposure strategy), or they just knew how to spell the word (e.g., direct retrieval strategy). It is interesting that students' strategies were the same, despite their reading and spelling skills differences and their observed phonological and orthographic processing differences. Moreover, the fact that students reported having used a practice or exposure strategy is consistent with Ehri's (2000) contention that reading and spelling are not that different; that we read to spell, and spell to read. We check word spellings by reading them, for example, and we attain practice with phonology and the appropriate visual appearance of words through being exposed to these patterns in print. For misspellings, all students reported a reliance on a phonological strategy (e.g., sounding the word out) to spell unfamiliar words despite the fact that the poor subtype do not possess strong phonological skills. Students also frequently reported a visual strategy (e.g., judging that the word does not "look" right) for spelling difficult words. While it is unclear why no significant differences were detected between groups, as anticipated, it is possible that students' strategy reports were more a reflection of the kinds of strategies they have been explicitly taught to use for spelling rather than a genuine indication of how they are actually spelling words. More research is certainly needed to clarify these issues. C. Methodological Issues Several limitations to the present study are acknowledged. First, the sample size was relatively small compared to other studies (e.g., Lennox & Siegel, 1993, 1994), particularly for the mixed subtype of good readers and poor spellers which may have affected the power of the statistical analyses used to uncover actual differences between subtypes. Moreover, the fact that, teachers' nominations of subtypes differed for the mixed and poor students when compared to WRAT-3 test results may likely have been an artifact of the percentile cut-offs used to define groups, rather than teachers' actual perceptions of the characteristics of mixed and poor readers and spellers. Similarly, good readers and spellers achieved significantly higher scores on the reading subtest of the WRAT-3 than the mixed readers and spellers suggesting that the good and mixed subtype groups were composed of "good" and "average" readers, respectively. If the difference in reading skill between the good and mixed group had been critical, however, then the mixed and poor students would have been expected to perform similarly on tasks that differentiated good from mixed readers and spellers, which was not the case. Specifically, since the main differences between the good and mixed subtypes were found on tasks assessing recall of 57 orthographic information, and since the poor subtype performed similarly to the good group on these tasks, the fact that the good subtype in the present study was composed of very good readers is not considered to have confounded the results. It was also not possible in the present study, due to the small sample size, to verify that students within the top (e.g., at or above the 76th percentile) and bottom (at or below the 10th percentile) of the reading and spelling distribution performed similarly to their good and poor counterparts at the 35th to 75th percentiles and the 11th to 25th percentiles, respectively. Since the criterion cut-offs are crucial in defining subtype groups, future research should include a check to verify their utility. There was also no specific information about the literacy development of the students included in the present study. A key concern would be whether the mixed group actually had difficulties acquiring early literacy skills (e.g., phonological processing) like the poor group, but for whatever reason, surpassed the poor group in acquiring these skills at a later date. The residual effect, however, of this literacy delay may be the poor spelling ability. Moreover, detailed information about current teaching practices in the children's classrooms was not included in the present study, although teacher reports indicated that all students had weekly dictated spelling tests. Another methodological consideration involves the definition of subtypes of spellers based on reading decoding. Although identification of good and poor readers on the basis of decoding tasks is used by the most extensive research to date (e.g., Lennox & Siegel, 1993, Siegel, 1994), there is still the unresolved issue of where reading comprehension factors into the subtype groupings. The present study did not include a measure of reading comprehension as part of an analysis of reading skill for the following reasons: First, much research has been conducted using only decoding. Second, decoding skills provide a better assessment of students' use of phonological information, part of the main focus of the present research. An important statistical consideration involves the possibility that ceiling effects on the rhyme judgment and orthographic awareness tasks and a floor effect on the cued recall task confounded the results. By observing the means of performance on these tasks, there does appear to be some scale attenuation problems (Elmes, Kantowitz, & Roediger, 1989). However, initial statistical checks on the normality of the data indicated normal distributions across word types on the rhyme judgment and cued recall tasks, and for the orthographic awareness task (based on the Shapiro-Wilk test of normality). Thus, scale attenuation problems are not considered to have confounded the results and analysis of variance procedures along with subsequent interpretations were appropriate for the data. Although the error analyses did not reveal any differences between good and mixed subtypes, it is possible that this procedure does not effectively measure students' spontaneous 58 strategies while spelling. Also, the mixed subtype in the present study demonstrated spelling strategies that were as effective across phonological and orthographic scoring criteria as the good subtype despite the fact that they performed more poorly remembering orthographically similar words. It is possible that the error analysis methodology does not reliably capture students' spelling strategies and spontaneous use of phonological and orthographic information while spelling. Instead, the error analysis may only provide a measure of the kinds of strategies students have been instructed to use when spelling hard words, namely a phonological approach. Students of varying ability will rely on this approach with differing levels of success. It is still likely that the mixed subtype's spellings are being impacted by their tendency not to use orthographic information effectively in memory, but that an error analysis does not capture this subtle processing difficulty. Instead, a technique that required students to visually recognize their own or others' spelling errors may have more effectively captured these orthographic processing skills. Similarly, it is also possible that while no differences were detected between subtypes on their report of strategies used for spelling, students were actually reporting the kinds of strategies they had been taught to use regardless of whether or not they used these strategies effectively. An improved interview eliciting actual strategy use may have been more effective. D. Directions for Future Research There are several intriguing extensions of the present research. One main issue is the memory processing of the mixed subtype and whether other aspects of orthographic coding are deficient (i.e., is the difficulty only attributed to visual memory for linguistic information?). Comparing the three subtypes' performance on a series of linguistic and non-linguistic visual memory tasks may provide further evidence of any different uses of orthographic information in memory by these groups. Similarly, future research may be directed at predicting what students are at risk for developing problems with spelling, particularly the mixed group of good reading decoders who have strong phonological processing skills who may be at risk due to an ineffective use of orthographic information in memory. It is well established that phonological awareness is the single most important predictor of reading success and that early intervention that targets the development of phonological processing will enhance students' literacy skills. It will be interesting to see if future research will find that orthographic processing can predict spelling performance in some students, given its importance in distinguishing subtypes, (e.g., those with asymmetrical reading decoding and spelling skills) and that early interventions that build visual memory skills in the context of reading and spelling instruction, for example, may improve students' (particularly mixed readers and spellers') early spelling skills. 59 Yet another unresolved and intriguing area for future research is the role of reading comprehension and subtypes of readers and spellers and performance on phonological and orthographic tasks. An extension of the present study may be to compare groups of reading decoders and comprehenders to see whether the processing characteristics of the three subtypes that have been established in the literature remain robust. Extending the view of spelling to include semantic and morphological as well as orthographic and phonological processes and spelling subtypes, as others have suggested (e.g., Kamhi & Hinton, 2000), would provide a comprehensive view of spelling as a linguistic process. More research is also needed on students' own reports of the kinds of strategies they use while spelling, rather than only inferring these strategies from analyses of students' errors. Separating students' actual approach to spelling from the strategies they have been instructed to use in spelling will continue to be a challenge for future strategy research. E . Implications for Educational Assessment and Instruction The present study has emphasized the importance of understanding spelling ability in relation to reading skills. As Ehri (1994, 2000) and Snowling (1994) have noted, reading and spelling are inherently interrelated, and spelling ability has a direct influence on reading skills. Indeed, there is research to support the benefits of spelling instruction to reading (e.g., Ehri & Wilce, 1987; Uliry & Shepherd, 1993). Viewing spelling as a skill that can be accommodated for students with learning difficulties via portable spell checkers and the word processor undermines the importance of spelling to literacy development generally (Ehri, 2000). Therefore, teaching and remediating spelling within the context of literacy instruction generally should be an important objective for all students. The fact that the present study, as well as previous research, has found processing differences between spelling subtypes also provides strong support for the existence of the good, mixed, and poor subtypes. Understanding the differences between these subtypes in relation to the use of phonological and orthographic information will be critical in the evaluation and subsequent instructional programming for students with spelling difficulties. The addition of phonological and orthographic tasks such as the rhyme judgment, cued recall, and orthographic awareness tasks of the present study may also serve to augment the assessment information gathered. Another worthwhile topic for future research is the nature of spelling instruction and interventions that will be most effective for the mixed and poor subtypes. Providing all poor spellers with the same strategies may not effectively meet their unique processing needs. Whether or not strategies that capitalize on the mixed subtypes' strong phonological skills while building their visual memory or similarly capitalizing on the poor subtypes' strong visual 60 memory for words, but poor phonological skills will lead to improvement in spelling ability remains to be examined. It is also possible that instructional approaches that capture the poor subtypes' well developed use of orthographic information in memory will effectively teach spelling to these students. In fact, Siegel (1994) describes the Bridge Reading Program, investigated by Biemiller and Siegel (1990) where severely reading disabled students learned to read through a visual route. Icons are used in this approach as picture cues for words and the assumption is that due to students' better developed visual processing skills, pictorial information will be linked more readily to students' lexical memory. The picture cues are gradually withdrawn as supports as students become more fluent identifying the words. This approach is reported by Siegel to have successfully increased first graders' scores on a word reading test after eight months of instruction. Whether such an approach could be adapted and used with the poor subtype of spellers would be an interesting extension. Mixed spellers, because of their tendency to not use orthography as effectively in memory, may benefit more from instruction that captures their better-developed phonological processing skills (i.e., strategies that emphasize breaking the word up into sound segments to spell). Indeed, there is still much to learn about the kinds of strategies that will be effective for different reading and spelling subtypes. F. Conclusions The present study has added empirical support for distinctive processing characteristics of mixed readers and spellers who are good at reading but poor at spelling. Certainly, the measure used to define reading (decoding subtest of the WRAT-3) is limited and does not provide a comprehensive assessment of students' reading ability and the many interactive elements that reading entails (Kibby, 1995). However, it does capture two important parts of this interactive process-- word recognition and phonological processing and the present results are best viewed within this context. Indeed, some researchers (e.g., Bruck & Waters, 1985; Kamlii & Hinton, 2000) contend that the mixed subtype is not distinguishable from the poor subtype. For example, Kamhi and Hinton have stated that "individual differences in spelling ability are primarily caused by differences in the knowledge and use of sound-spelling information rather than differences in some non-linguistic factor" (p. 48). However, findings from the present study have provided support that in most aspects of phonological processing, the good and mixed subtypes were distinguishable from the poor subtype. Moreover, the mixed subtype was distinguishable from the good subtype in using an orthographic code in memory, and the good and poor subtypes were not distinguishable by their use of orthography in memory. Moreover, the poor subtype was comparable to the good and mixed subtypes in identifying the appropriate visual appearance of words. These findings are consistent with the theory suggested by 61 Pennington et al. (1995) that students with poor reading and spelling skills experience a dissociation in the acquisition of phonological and orthographic skills. According to these researchers, although the development of phonological skills appears arrested, orthographic skills develop appropriately. Findings from the present study support this perspective for the poor readers and spellers. The mixed subtype, in the present study, demonstrated less use of orthographic information in memory. Extending Pennington et al.'s theory to include the mixed group of good readers and poor spellers, it is possible that this group experiences a more subtle dissociation in the reverse direction than poor readers and spellers-- that phonological skills are acquired appropriately, but that some aspect of the development of orthographic skills (e.g., memory for orthographic information) is delayed. These subtle difficulties in acquiring orthographic skills may be due to an incomplete representation of words in memory from reading affecting spelling performance, as Frith (1980) originally suggested. Therefore, contrary to Kamhi and Hinton's claims, there is more to subtype differences than phonology alone. R E F E R E N C E S 62 Barry, C. (1992). Interactions between lexical and assembled spelling. In C. M. Sterling and C. Robson (Eds.), Psychology, Spelling, and Education (pp. 71- 86). Avon, UK: Multilingual Matters Ltd. Barry, C. (1994). Spelling routes (or roots or rutes). In G. D. A. Brown and N. C. Ellis (Eds.), Handbook of spelling: Theory, research, and intervention (pp. 27-50). New York: John Wiley & Sons. Barry, C., & Seymour, P. H. K. (1988). Lexical priming and sound-to-spelling contingency effects in nonword spelling. Quarterly Journal of Experimental Psychology, 40A, 5-40. Barron, R. W. (1980). Visual and phonological strategies in reading and spelling. In U. Frith (Ed.), Cognitive Processes in Spelling, (pp. 195-213). London: Academic Press. Beauvois, M. F., & Derouesne, J. (1981). Lexical or orthographic agraphia. Brain, 104, 21-49. Biemiller, A. & Siegel, L. S. (1990). A test of an experimental reading curriculum with a disadvantaged population. Unpublished manuscript. Boder, E. (1973). Developmental dyslexia: A diagnostic system based on three atypical reading-spelling patterns. Developmental Medicine and Child Neurology, 15, 663-687. Brown, A. (1990). A review of recent research in spelling. Educational Psychology Review, 2, 365-397. Brown, G. D. A. & Ellis, N. C. (1994). Handbook of spelling: Theory, process, and intervention. New York: John Wiley & Sons. Bruck, M., & Waters, G. (1988). An analysis of spelling errors of children who differ in their reading and spelling skills. Applied Psycholinguistics, 9, 77-92. Bruck, M., & Waters, G. (1990). An analysis of the component spelling and reading skills of good readers-good spellers, good readers-poor spellers, and poor readers-poor spellers. In T. H. Carr and B. A. Levy (Eds.), Reading and its development: Component skills approaches (pp. 161-206). New York: Academic Press. Bryant, R. E., & Bradley, L. (1980). Why children sometimes write words which they do not read. In U. Frith (Ed.), Cognitive Processes in spelling (pp. 355-370). London: Academic Press. Bub, D., & Kertesz, A. (1982). Evidence for lexicographic processing in a patient with preserved written over oral single word naming. Brain and Language, 17, 146-165. Campbell, R. (1983). Writing nonwords to dictation. Brain and Language, 19, 153-178. 63 Campbell, R. (1985). When children write nonwords to dictation. Journal of Experimental Child Psychology, 40, 133-151. Carroll, J. B., Davies, P., & Richman, B. (1971). The American heritage word frequency book. Boston: Houghton Mifflin. Cataldo, S., & Ellis, N. (1988). Interactions in the development of spelling, reading, and phonological skills. Journal of Research in Reading, 11, 86-109. Cook, L. (1980). Misspelling analysis in dyslexia: Observation of developmental strategy shifts. Bulletin of the Orton Society. 31, 123-134. Curtis, M. E. (1980). Development of components in reading skill. Journal of Educational Psychology, 72, 656-669. Cunningham, A. E., & Stanovich, K. E. (1991). Tracking the unique effects of print exposure in children: Associations with vocabulary, general knowledge, and spelling. Journal of Educational Psychology, 83, 264-274. Ehri, L. (1980). The development of orthographic images. In U. Frith (Ed.), Cognitive processes in spelling (pp. 311 -338). London: Academic Press. Ehri, L. C. (1987). Learning to read and spell words. Journal of Reading Behavior, 19, 5-31. Ehri, L. (1989). The development of spelling knowledge and its role in reading acquisition. Journal of Learning Disabilities, 22, 356-366. Ehri, L. (2000). Learning to read and learning to spell: Two sides of a coin. Topics in Language Disorders, 20, 19-36. Ehri, L. & Wilce, L. (1987). Does learning to spell help beginners learn to read words? Reading Research Quarterly, 22, 47-65. Ellis, N. C. (1994). Longitudinal studies of spelling development. In G. D. A. Brown and N. C. Ellis (Eds.), Handbook of spelling: Theory, process, and intervention (pp. 155-177). New York: John Wiley & Sons. Elmes, D. G., Kantowitz, B. H., & Roediger, H. L. (1989). Research methods in psychology. St. Paul: Houghton Mifflin. Frith, U. (1980). Cognitive processes in spelling. London: Academic Press. Frith, U. (1985). Beneath the surface of developmental dyslexia. In K. E. Patterson, J. C. Marshall, and M. Colteheart (Eds.), Surface dyslexia (pp. 301-330). London: Erlbaum. Gentry, J. R. (1978). Early spelling strategies. The Elementary School Journal, 79, 88-92. Gentry, J . R. (1982). An analysis of developmental spelling in GNYS at WRK. Reading Teacher, 36, 192-200. 64 Gerber, M. M. (1984). Orthographic problem-solving ability of learning disabled and normally achieving students. Learning Disability Quarterly, 7, 157-164. Gerber, M. M., & Hall, R. J. (1987). Information processing approaches to studying spelling deficiencies. Journal of Learning Disabilities, 20, 34-42. Goodman, R. A., & Carmazza, A. (1986). Dissociation of spelling errors in written and oral spelling: The role of allographic conversion in writing. Cognitive Neuropsychology, 3, 179-206. Goodman-Schulman, R., & Carmazza, A. (1987). Patterns of dysgraphia and the nonlexical spelling process. Cortex, 23, 143-148. Goswami, U. (1988). Children's use of analogy in learning to spell. British Journal of Developmental Psychology, 6, 21-33. Goswami, U., & Bryant, P. (1990). Phonological skills and learning to read. London: Erlbaum. Goswami, U., & Bryant, P. (1992). Rhyme, analogy, and children's reading. In P. Gough, L. Ehri, & R. Treiman, (Eds.), Reading acquisition (pp. 35-48). Hillsdale, NJ: Erlbaum. Heald-Taylor, G. (1998). Three paradigms of spelling instruction in grades 3 to 6. The Reading Teacher, 51, 404-413. Henderson, E. H., & Templeton, S. (1986). A developmental perspective of formal spelling instruction through alphabet, pattern, and meaning. The Elementary School Journal, 80, 305-316. Holligan, C , & Johnston, R. S. (1988). Theuse of phonological information by good and poor readers in memory and reading tasks. Memory and Cognition, 6, 522-532. Jorm, A. F. (1981). Children with reading and spelling retardation: Functioning of whole-word and correspondence-rule mechanisms. Journal of Child Psychology and Psychiatry, 22, 171-178. Kail, R. V., & Bisanz, J. (1982). Cognitive strategies. In G. Puff (Ed.), Handbook of research methods in human memory and cognition (pp. 229-255). New York: Academic Press. Kamlii, A. G., & Hinton, L. N. (2000). Explaining individual differences in spelling ability. Topics in Language Disorders, 20, 37-49. Kibby, M. (1995). Practical steps for informing literacy instruction: A diagnostic decision-making model. Newark, Delaware: International Reading Association. Lennox, C , & Siegel, L. S. (1993). Visual and phonological spelling errors in subtypes of children with learning disabilities. Applied Psycholinguistics; 14, 473-488. Lennox, C , & Siegel, L. S. (1994). The role of phonological and orthographic processes in learning to spell. In G. D. A. Brown and N.,C. Ellis (Eds.), Handbook of spelling: Theory, process, and intervention (pp. 93-109). New York: John Wiley & Sons. 65 Lesiak, J. , Lesiak, J., & Kirchheimer, J. (1979). Auditory and visual factors related to spelling success. Psychology in the Schools, 16, 491-494. Marsh, G., Friedman, M., Welch, U., & Desberg, P. (1980). The development of strategies in spelling. In U. Frith (Ed.), Cognitive processes in spelling (pp. 339-353). Toronto: Academic Press. McNaughton, D., Hughes, C. A., & Clark, K. (1994). Spelling instruction for students with learning disabilities: Implications for research and practice. Learning Disability Quarterly, 17, 169-185. Pennington, B. F., McCabe, L. L., Smith, S. D., Lefly, D. L., Bookman, M. O., Kimberling, W. J. , & Lubs, H. A. (1986). Spelling errors in adults with a form of familial dyslexia. Child Development, 57, 1001-1013. Porpodas, C. D. (1999). Patterns of phonological and memory processing in beginning readers and spellers of Greek. Journal of Learning Disabilities, 32, 406-416. Rack, J. P. (1985). Orthographic and phonetic coding in developmental dyslexia. British Journal of Psychology, 76, 325-340. Read, C. (1980). Children's use of phonetic cues in spelling, parsing, and--maybe--reading. Bulletin of the Orion Society, 30, 209-224. Roberts, K. T., & Ehri, L. C. (1983). Effects of two types of letter rehearsal on word memory in skilled and less skilled beginning readers. Contemporary Educational Psychology, 8, 375-390. Roeltgen, D. P., Sevush, S., & Heilman, K. M. (1983). Phonological agraphia: Writing by the lexical-semantic route. Neurology, 33, 755-765. Roeltgen, D. P., & Heilman, K. M. (1984). Lexical agraphia: Further support for the two-system hypothesis of linguistic agraphia. Brain, 107, 811-827. Scott, C. M. (2000). Principles and methods of spelling instruction: Applications for poor spellers. Topics in Language Disorders, 20, 66-82. Schacter, D. L. (.1996). Searching for memory: The brain, the mind, and the past. New York: Harper Collins. Seidenberg, M. S., & Tanenhaus, M. K. (1979). Orthographic effects on rhyme monitoring. Journal of Experimental Psychology: Human Learning and Memory, 5, 546-554. Seymour, P.H.K., & Dargie, A. (1990). Associative priming and orthographic choice in nonword spelling. The European Journal of Cognitive Psychology, 2, 395-410. Seymour, P. H. K. (1992). Cognitive theories of spelling and implications for education. In C. M. Sterling and C. Robson (Eds.), Psychology, Spelling, and Education (pp. 50- 70). Avon, UK: Multilingual Matters Ltd. 66 Sha l l i ce , T. (1981). Phonologica l agraphia and the lex ica l route i n wr i t ing . B r a i n , 104, 413-429. Shankwei l e r , D . , L iberman , I. Y . , Mark , L. S., & Fowler, C . A . , & Fischer , F . W. (1979). The speech code a n d lea rn ing to read. J o u r n a l of Exper imenta l Psychology: H u m a n Lea rn ing a n d Memory , 5, 531-545. Siegel , L . S. (1993). Phonologica l deficits as the basis of a reading d isab i l i ty . Developmenta l Review, 13, 246-257. Siegel , L . S. (1994). Phonologica l processing deficits as the basis of developmenta l dys lex ia : Impl icat ions for remediat ion. In G . Humphr i e s and J . R i d d o c h (Eds.), Cognitive Neuropsychology a n d Cognit ive Rehabi l i ta t ion (pp. 379-400). Siegel , L . S. (1998). Phonologica l processing deficits and reading d isabi l i t ies . In J . L . M e t s a l a a n d L . E h r i (Eds.), Word Recogni t ion i n Beg inn ing Li teracy (pp. 141-160). M a w a h , N J : E r l b a u m . Siegel , L . S., Share, D . , & Geva, E . (1995). Evidence for super ior or thographic sk i l l s i n dys lexics . Psychologica l Science, 6, 250-254. Siegler, R. S. (1991). Chi ldren ' s T h i n k i n g (2nd ed.). Englewood Cliffs, N J : Prentice H a l l . S n o w l i n g , M . (1994). Towards a model of spe l l ing acqu is i t ion : The development of some componen t s k i l l s . In G . D . A . B r o w n and N . C. E l l i s (Eds.), Handbook of Spe l l ing , (pp. 111-128). New Y o r k : J o h n Wiley & Sons . S t anov ich , K . E . (1989). E x p l a i n i n g the differences between the dyslexic a n d the garden-variety poor reader: The phonological-core variable-difference model . J o u r n a l of L e a r n i n g Disabi l i t i es , 2 1 , 590-604. Steffler, D . , Varnhagen , C , K. , Friesen, C. K. , & Tre iman, R. (1998). There's more to ch i ldren ' s spe l l ing t h a n the errors they make: Strategic and automat ic processes for one-syl lable words. J o u r n a l of Educa t iona l Psychology, 90, 492-505. S w a n s o n , H . L . , & Ramalgia , J . M . (1992). The re la t ionship between phonolog ica l codes o n memory a n d spe l l ing tasks for s tudents wi th and wi thout l ea rn ing d isab i l i t i es . J o u r n a l of Lea rn ing Disab i l i t i es , 25 , 396-407. T re iman (1993) Topping , K . J . (1995). C u e d spel l ing: A powerful technique for parent a n d peer tu tor ing. The Read ing Teacher, 48, 374-383. Torgesen, J . K . (1985). Memory processes i n reading d isabled ch i ld ren . J o u r n a l of L e a r n i n g Disab i l i t i es , 18, 350-357. Tre iman , R. (1993). Beg inn ing to spel l . New York: Oxford Univers i ty Press. 67 Treiman, R. (1994). Sources of information used by beginning spellers. In G. D. A. Brown and N. C. Ellis (Eds.), Handbook of Spelling: Theory, Process, and Intervention (pp. 75-91). New York: John Wiley & Sons. Uhry, J . & Shepherd, J. (1993). Segmentation/spelling instruction as a part of a first-grade reading program: Effects on several measures of reading. Reading Research Quarterly, 28, 218-233. Varnhagen, C., McCallum, M., & Burstow, M. (1997). Is children's spelling naturally stage-like? Reading and Writing, 9, 451-481. Walton, C. (1997). The development of phonological awareness and spelling competence in beginning spellers. Unpublished Master's Thesis, University of British Columbia. Waters, G. S., Bruck. M., & Seidenberg, M. (1985). Do children use similar processes to read and spell words? Journal of Experimental Child Psychology, 39, 511-530. Watkins, M. J. , & Gardiner, J. M. (1982). Cued recall. In G. Puff (Ed.), Handbook of research methods in human memory and cognition (pp. 173-195). New York: Academic Press. Wilkinson, G. S. (1993). The Wide Range Achievement Test-Third Edition. Wilmington, DE: Wide Range Inc. Willows, D., & Scott, R. (1994). Spelling processes of the reading disabled. In G. D. A. Brown and N. C. Ellis (Eds.), Handbook of Spelling: Theory, Process and Intervention (pp. 193-210). New York: John Wiley & Sons. Woodcock, R. W. (1987). Woodcock Reading Mastery Test-Revised. Circle Pines, MN: American Guidance Service. Appendix A Parent Consent Form 6 8 PARENT/GUARDIAN INFORMED CONSENT FORM (Parent/Guardian Copy) Research Project: The Sounds and Appearance of Words in Memory: Implications for Reading and Spelling Skills Principal Investigators: Shelley Hymel [Professor/Faculty Advisor to G. Harrison, Dept. of Educational & Counselling Psychology & Special Education (ECPS) , phone: 822-6022) & Gina Harrison, M.A. [Doctoral Student, E C P S ; project conducted to fulfill Ph.D. dissertation requirement] University of British Columbia Purposes of the Project: The study is aimed at understanding the skills children use in reading versus spelling. We are especially interested in children's memory skills and whether or not they rely more on the sounds of the words or visual appearance of words. We hope this information will help us to find better ways to teach spelling and reading skills. Project Procedures: The project involves two sessions, both conducted at your child's school during the regular school day and at a time your child's teacher selects in order to avoid missing important school work. The first session takes about 30 minutes and the second session takes about 45 minutes. In the first session, students will be asked to read a list of individual words and to spell a list of dictated words. During the second session, students will be asked to spell a few more difficult words and to report on how they go about spelling these words. They will also be asked to complete several other tasks: (1) reading a list of non-words or pretend words (e.g., "rox"), (2) choosing which of two words in a pair of words "looks like a real word", (3) deciding whether two words rhyme or not, and (4) remembering a list of words. All of these tasks have been developed for children your son or daughter's age. Confidentiality: The study will be conducted by Ms. Harrison, working with your child's teacher, and all information obtained from this project will be treated as strictly confidential. This form, as well as the student and teacher consent forms will be the only documents containing identifying information, and these forms will be stored in a locked file cabinet in Ms. Harrison's U.B.C. office. All other forms will be numerically coded and access to these forms will be restricted to Ms. Harrison or Dr. Shelley Hymel. Only those children who receive parent permission and who themselves agree to participate will be involved. Student participation is voluntary, and your child has the right to withdraw from the study at any time without any type of penalty. Students who do not participate will simply work on regular school work. Participation in this project (or not participating) will have no effect on your child's grades or schoolwork. Appendix B Cues and Targets used in Rhyme Judgment and Cued Recall Tasks (Adapted from: Holligan & Johnston, 1988) Rhyme Only Rhyn e- Word-Orthographic Dissimilar Rhyming and Orthographic Nonrhyming and Noiirhymi lg and Orthographically Rhyming and Orthographically Orthographically Different pairs Orthographically Simil ar pail's Different pairs Similar pairs coal pole wash cash yarn rock pond wand coin join both mo til pest lace hawk walk fish wish bowl howl card boil come plum hang gang work fork lump shut nose goes rack tack paid said trip mask • bull wool sink link none cone cage send leak seek fall tall bead dead tape safe bait date meat seat farm warm fill camp bean teen blow flow post lost dish sand moan bone kiss hiss dull pull pile horn Joke soak bake rake root foot limp hush doom tomb drip grip fear wear free sick food rude bite kite rush bush shot ripe head shed gown town love move care rich glue shoe clue blue hose lose club sort near peer peak beak hood mood pipe glad tide hide Appendix C Task Instructions: Rhyme Judgment 71 Researcher: We're doing a research project on reading and spelling arid are interested in looking at the way people make decisions about how words rhyme-- how they sound the sariie--which is considered important to reading and spelling. Give me an example of two rhyming words. Student: Provides an appropriate example (If student doesn't provide an example, researcher provides one (e.g., "give"'and "live", they both sound the same, they both rhyme). Researcher: Now we're going to do a rhyming task on the computer. There are a few practice items, and then we'll start with the actual task. Let's try the practice ones. You're going to see two words appear on the screen and you have to decide, as quickly and as accurately as you can, whether the two words rhyme. If you think they do then press the "yes" key, if you think they don't then press the "no" key. [administer practice set] Researcher: You did really well on those. We're finished practicing now. Let's do some more, but for these ones I can't help you. Make sure .you decide "yes" or "no" whether the words rhyme. [administer task] Appendix D Task Instructions: Cued Recall Memory Task 72 Researcher: Now you're going to see one word from each pair, appear on the screen and you're going to try and remember the other word that went with it. Don't worry, I didn't ask you to try and remember the words so I don't expect you to get very many of them. Some people only get one or two. Just try and remember as many as you can. I'll type the words you remember into the computer. If you can't remember the word that went with the one you saw before, just guess. [administer task] Appendix E Pseudoword Reading Word List (Word Attack subtest from the Woodcock Reading Mastery Test-Revised, WRMT-R) Woodcock, R.W. (1987) ift ziz tob telequick polybendable bim ott chen shenning dinlan ut nudd hets quib eldop rayed weet plon laip wubfambif kak plen lundy fubwit wotfob maft twib hode pertome cigbet nen beb expram sloy conration ab rejune stabe subcrote biftel tash knap imbaf pipped bafmotbem wip's ain earn etbom nolhod Appendix F Orthographic Awareness Task Word List Siegel, Share, & Geva (1995) filv tolz powl filk tolb lowp togd wolg moke togn wolt moje hift gwup nitl hifl gnup nilt dlun fant miln lund tanf milg jofy cnif bnad fojy crif blad clid vism cdil visn Appendix G Task Instructions: Strategy Report 75 Researcher: I'm going to give you four words that I'd like you to try to spell, one word at a time. 'While you're spelling the word, I'd like you to tell me how you're spelling it, what you're thinking while you're trying to spell it. Prompts include: - "Tell me how you knew to spell it that way?" (for correctly and incorrectly spelled words) - "Tell me how did you know that was right?" (for correctly spelled words only) - "Tell me how did you spell that so quickly?" - "Tell me more about how you're spelling that?" Appendix H Means and Standard Deviations on Dependent Measures for Gender Scores Boys (n = 28) Girls (n = 22) Spelling Percentile M SD 32.39 25.64 46.90 28.78 Reading Percentile M SD 43.46 29.61 59.18 26.89 Pseudoword Reading M SD 34.42 10.98 36.68 11.08 Orthographic Awareness M SD 14.28 1.65 14.95 1:21 Error Analysis (phonological) M SD 85.28 7.26 87.81 6.29 Error Analysis (orthographic) M SD 65.32 5.04 66.22 7.06 Rhyme Judgment Accuracy (coal-pole) M 14.25 SD 2.87 14.22 3.63 Rhyme Judgment Accuracy (weed-need) M • 15.39 SD 1.39 15.95 1.08 Rhyme Judgment Accuracy (wash-cash) M 12.03 SD 4.51 13.22 4.34 Rhyme Judgment Accuracy (best-card) M 15.46 SD .79 15.81 .39 Rhyme Judgment Latency (coal-pole) M 2.52 SD .17 2.48 .20 Rhyme Judgment Latency (weed-need) M 2.20 SD .13 2.13 .15 Rhyme Judgment Latency (wash-cash) M 2.72 SD .17 2.81 .19 77 Appendix H continued Scores Boys Girls Rhyme Judgment Latency (best-card) M 15.46 15.81 SD .79 .39 Cued Recall (coal-pole) M 2.32 1.81 SD 1.78 1.40 Cued Recall (weed-need) M 3.89 4.31 SD 2.11 • 2.43 Cued Recall (wash-cash) M 1.10 1.77 SD 1.16 . 1.65 Cued Recall (best-card) M .18 .12 SD .61 .21 Phonological Strategy M 24.74 25.53 SD 23.70 14.40 Practice/Exposure Strategy M 32.40 27.81 SD 23.49 26.01 Visual Strategy M 11.53 15.96 SD 14.60 18.38 Direct Retrieval Strategy M 29.37 23.92 SD 19.30 25.99 Rule-Based Strategy M 1.87 5.02 SD ' 5.63 14.04 Phonological Strategy (misspellings) M 57.21 52.59 SD 21.96 21.73 Practice/Exposure Strategy (misspellings) M 9.73 7.31 SD 13.16 10.94 Visual Strategy (misspellings) M 27.65 31.97 SD 20.83 26.81 78 Appendix H continued Scores Boys Girls Direct Retrieval Strategy (misspellings) M 2.03 1.47 SD 10.77 4.91 Rule-Based Strategy (misspellings) M 5.51 5.37 SD 11.03 10.51 79 Appendix I Means and Standard Deviations on Dependent Measures for Grade Scores Grade 4 (n = 21) Grade 5 (n = 29) Spelling Percentile M SD 39.42 27.05 38.31 28.72 Reading Percentile M SD 51.38 32.40 49.65 27.30 Pseudoword Reading M SD 34.04 11.65 36.41 10.55 Orthographic Awareness M SD 14.71 1.65 14.48 1.59 Error Analysis (phonological) M SD 84.76 6.81 87.58 6.84 Error Analysis (orthographic) M SD 64.47 6.53 66.62 5.45 Rhyme Judgment Accuracy (coal-pole) M 14.33 SD 3.36 14.17 3.12 Rhyme Judgment Accuracy (weed-need) M 15.33 SD 1.59 15.86 .35 Rhyme Judgment Accuracy (wash-cash) M 13.57 SD 3.61 11.82 4.89 Rhyme Judgment Accuracy (best-card) M 15.52 SD .74 15.68 .60 Rhyme Judgment Latency (coal-pole) M 2.41 SD .49 2.57 1.15 Rhyme Judgment Latency (weed-need) M 2.17 SD .45 2.16 .86 80 Appendix I continued Scores Grade 4 Grade 5 Rhyme Judgment Latency (wash-cash) M 2.67 SD .58 2.82 1.06 Rhyme Judgment Latency (best-card) M 2.25 SD .42 2.23 .80 Cued Recall (coal-pole) M SD 2.10 1.37 2.10 1.81 Cued Recall (weed-need) M SD 4.80 2.40 3.55 2.01 Cued Recall (wash-cash) M SD 1.47 1.32 1.34 1.51 Cued Recall (best-card) M SD .47 .17 .21 .60 Phonological Strategy M SD 22.56 19.55 26.92 20.40 Practice /Exposure Strategy M SD 29.08 20.36 31.33 27.39 Visual Strategy M SD 14.43 18.49 12.79 14.91 Direct Retrieval Strategy M SD 29.63 26.07 25.05 19.61 Rule-Based Strategy M SD 4.09 9.05 2.65 11.10 Phonological Strategy (misspellings) M SD 54.32 19.19 55.79 23.76 Practice/Exposure Strategy (misspellings) M 10.67 SD 14.38 7.21 10.32 81 Appendix I continued Scores Grade 4 Grade 5 Visual Strategy (misspellings) M 28.35 30.43 SD 23.65 23.76 Direct Retrieval Strategy (misspellings) M 1.54 1.96 SD 5.02 10.58 Rule-Based Strategy (misspellings) M SD 5.96 12.27 5.08 9.60 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0054604/manifest

Comment

Related Items