Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The relationship between phonological delay and later metaphonological development Dewhurst, Jacqueline 1991

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

Item Metadata


831-UBC_1991_A6_7 D49.pdf [ 3.61MB ]
JSON: 831-1.0098483.json
JSON-LD: 831-1.0098483-ld.json
RDF/XML (Pretty): 831-1.0098483-rdf.xml
RDF/JSON: 831-1.0098483-rdf.json
Turtle: 831-1.0098483-turtle.txt
N-Triples: 831-1.0098483-rdf-ntriples.txt
Original Record: 831-1.0098483-source.json
Full Text

Full Text

THE RELATIONSHIP BETWEEN PHONOLOGICAL DELAY AND LATER METAPHONOLOGICAL DEVELOPMENT by JACQUELINE DEWHURST B.Sc, The University of Victoria, 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (School of Audiology and Speech Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1991 © Jacqueline Dewhurst, 1991 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. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) ABSTRACT This paper studies the effects of an early phonological delay on later metaphonological abilities, a hypothesised link generally neglected by other researchers. It also examines current definitions of metaphonology, and describes a rigorously controlled set of tasks devised to test phonological awareness. The study was limited to three subjects: two with a history of phonological delay, and one a typically developing control. The subjects were given rhyme and alliteration tasks that varied on early patterns of acquisition and the number of phonemes shared by the stimulus and the distractor items. The small number of subjects permitted a comprehensive analysis of results which indicated that residual, but non-overt, phonological processes had an effect on tasks requiring phonological manipulations. A further finding of the study was that several other factors, such as memory, attention, and the nature of the stimuli also affected performance. An analysis of test behaviours and strategies suggested that performance scores may be misleading when assessing ability. Response time, and problem solving strategies employed by subjects are also useful indicators of facility. In addition, subjects may not be performing the type of analysis that is assumed by researchers. Task analysis reveals that there may be more than one way to reach a solution, while subject behaviour suggests that children make use of the largest phonological units available to them to complete a task. ii TABLE OF CONTENTS ABSTRACT ii LIST OF TABLES v ACKNOWLEDGEMENTS vi CHAPTER 1. INTRODUCTION 1 Phonological Development 3 Tasks 4 Validity of Phonological Awareness Tests 8 Rime Awareness vs. Phonemic Awareness 9 Relative Difficulty of Tasks 10 . Age of Acquisition 11 Relationship between Linguistic and Metalinguistic Skill 14 The Role of Cognitive Development 16 Interaction between Metalinguistic And Literacy Skills 18 Conclusion 21 CHAPTER 2. METHOD 23 Subjects 23 Stimuli 26 iii Tasks 28 Procedure 33 CHAPTER 3. RESULTS: PART 1 - SUBJECT SCORES 35 Phonological Awareness Tasks 35 PART 2 - TEST BEHAVIOURS AND STRATEGIES 41 Rhyme Oddity Tasks 41 Rhyme Supply 44 Rhyme Judgment 44 Alliteration Oddity Tasks 45 Alliteration Supply 45 Alliteration Judgment 46 CHAPTER 4. DISCUSSION 47 Limitations of Study 47 Effects of Hypothesised Variables 52 Areas for Further Reseach 55 CONCLUSION 57 REFERENCES 59 APPENDIX A 62 APPENDIX B 65 APPENDIX C 70 iv LIST OF TABLES TABLE 2.1 25 2.2 25 3.1 36 3.2 37 3.3 37 3.4 39 3.5 40 APPENDIX 1 63 2 64 v ACKNOWLEDGEMENTS I would like to thank the many people who have been involved with this project. The children and their parents deserve special thanks for their tireless, good-humoured co-operation. Thanks go to John Gilbert, who not only expounded on the virtues of the straight and narrow path, but also pointed me in the right direction. Many thanks to Barbara Bernhardt who did so much more than supervise this thesis. Her generosity with her time, knowledge, and patience has been valued throughout this endeavor. I especially want to thank my family: my husband George and my sons James and Samuel, for their loving support and willing sacrifices over the last six years, and my parents and my brother and sisters, for their humour, encouragement and love. vi CHAPTER 1 INTRODUCTION This thesis examines the performance of phonologically delayed children on a set of metaphonological tasks. General metalinguistic ability can be considered the ability to reflect consciously on the form of language rather than, or in the absence of, content Metaphonological skills are a subset of metalinguistic skills which enable an individual to demonstrate awareness of the various sound units of speech at the syllabic and phonemic level. This ability to access the phonological units of the language should be demonstrable through performance on various tasks without regard to context or orthography. (This second condition creates obvious confounding difficulties when tasks are given to children or adults who have some literacy experience. These confounds will be discussed later in the paper.) In its more developed form, the ability to reflect on the different sounds of a language should be accompanied by awareness of how the sounds are combined and how sound structure accounts for similarities and differences between words (Grunwell, 1986). One way in which metaphonological skills have been distinguished from general linguistic ability is that they may be less necessary for speaking and listening (skills universally mastered in the absence of some social or physical disorder) than for development of literacy skills, which are not universally mastered, even with instruction (Mann, 1986). While the ability to do some metaphonological tasks appears to be a requirement for literacy, ability for other types of tasks may not develop unless the speaker is exposed to particular language experience such as reading 1 and writing (Mann, 1986). The literature on metaphonological skills is typically a discussion of the interaction between metaphonological skills and emerging literacy. Studies have generally been directed towards detecting and describing any predictive quality that the ability to complete metaphonological tasks may have regarding success in learning to read and spell. Results of these studies often fail to agree on the predictive nature of metaphonological ability, due in part to the variability in the tasks and the age of the populations studied, and in part to the lack of control over certain variables (see below). In addition, the usual cross-sectional and longitudinal designs have resulted in the averaging out of data, making it difficult to determine individual correlations between early language ability and metaphonological skill. This want of individual data has further exacerbated the lack of discussion in the literature as to why some children should fail to develop metaphonological skills at the same rate and with the same success as their peers. While subjects in most reported studies receive some degree of screening for general language ability (or extensive language testing in correlative studies), such is not the case for their phonological development. Potential subjects or may even be excluded from a study if they show evidence of an overt phonological disorder. Given that metaphonology requires the conscious manipulation of phonological segments, the absence of phonological data is a deficit of concern. In one of the rare studies that addresses this issue (Magnusson and Naucler, 1990), a correlation between early phonological delay and later metaphonological delay was demonstrated. The possibility that phonological deficits either overtly or covertly affect metaphonological development suggests that phonological representation may still be incomplete even after intervention has resulted in the production of adult forms. Brady, Shankweiler and 2 Mann (1983) have suggested that an impairment in short-term memory for phonetic units, which may result in reading deficits, may have its origins in an original failure "to adequately internalize properties relating to the phonetic pattern." They have hypothesised that a connected problem may be a subde impairment of perception. Phonological Development Phonological development begins early in life, and by the age of five, a child is normally producing forms of the phonemes in his language which are close to those of adults. During those five years the child has learned which sounds are contrastive in the language, and the constraints that govern their position and sequencing. The beginning speaker initially produces a number of incorrect forms which result from a variety of syllable structure and substitution processes. These processes act to simplify the adult form (Stoel-Gammon and Dunn, 1985). It is supposed that these processes gradually disappear, or are suppressed, resulting in the emergence of a mature sound system. There is evidence, however, that these phonological simplification processes re-emerge as a child learns to spell (Hoffmann and Norris, 1989). Spelling is assumed to be a language based task, rather than a visual one, based on evidence from various studies (e.g., Read, 1971; Gentry, 1982). At its early development, it is largely based on phonologically derived strategies rather than the orthographic rules of spelling which are only gradually learned as the child's abilities develop. Hoffman and Norris (1988) found that a number of phonetic and syllabic simplification processes were evident in children's spelling when their spelling errors were analyzed phonetically. These included syllable reduction (of the phonologically weak syllable), cluster 3 reduction, final consonant deletion, epenthesis and metathesis for syllabic processes, as well as stopping, deaffrication, voicing errors, palatalisation, stridency deletion, and assimilations. The question here is: if processes are suppressed during early language learning why do they apparentiy reappear in early spelling? If normal developmental processes continue to be present, albeit covertly, then there is reason to suppose that processes which result in delayed or disordered phonology will also continue to manifest themselves in some manner, resulting in metaphonological delays and difficulty with learning to spell. This position assumes that metaphonological and phonological development are structurally linked. Grunwell (1986), for example, considers metaphonological development to be a continuation of phonological development. Metaphonological ability cannot be assumed, therefore, to develop normally in the presence of disordered phonology, whether it is expressed or covert. Tasks Most metaphonological tasks designed for children at the pre-school or early school age level have suffered from a lack of control over certain variables and testing procedures. For example, word frequency and recognition, the native dialect of the speaker, task specificity, and randomised repetition of the stimuli items have commonly been disregarded in studies. Word frequency and recognition are important considerations as there is a possibility that some tasks may tap literacy rather than metaphonological ability. Metaphonological tasks should not require resorting to meaning, orthography or context for the solution, nor should these be distractions. Using a speaker with the same dialect as the subject is important for the same reasons. Divorced from context, the phonemic quality of each item should be as identical to the subject's own as 4 possible. The tasks used to determine metaphonological ability are not only highly variable but, in order to accommodate large numbers of subjects, may be under-analyzed and lack specificity in terms of the targeted skill. Thus, a group of children may be tested on one rhyming task with the results forming the basis for general conclusions about their ability in this area. For this reason, data on the normal age of acquisition for many metaphonological skills is both contradictory and inadequate. Randomised repetitions of stimuli allow an examiner to know more certainly whether an error occurred because of inability to do the task or for some other reason. The following section investigates the types of tasks that have been implemented in previous research and examines tasks from the point of view of the degree of difficulty and the age of acquisition. The types of tasks used to investigate children's metaphonological abilities are varied, and subject to design problems, but they are divisible into a few broad categories. The first major separation is between those tasks which require ability to segment the speech stream at the phonemic level and those which require a knowledge of larger phonological units, such as the syllable or the rime. Tasks which presumably tap these respective skills do not correlate well with each other (Stanovich, Cunningham and Cramer, 1984), and many studies suggest that this division is associated with the type of knowledge a child must possess in order to be successful at either skill (e.g., Morais, 1986). Tasks which require only syllable awareness may be completed by pre-literate children while tasks requiring phonemic awareness may also require exposure to a phonemic-based alphabet (Mann, 1986). This issue will be discussed more fully below. 5 Phonemic segmentation tasks take various forms, and are usually explicit. That is, there is direct manipulation of the phoneme to arrive at the solution (Smith and Tager-Flusberg, 1982). These tasks include: (1) phoneme deletion and addition, from either the beginning or end of a word, (e.g., fate becomes ate, or rose becomes row): (2) phoneme transposition (e.g., fast light becomes last fight): (3) phoneme counting, when the subject is asked how many sounds occur in a stimulus item. This task is sometimes done with the support of a peg board (e.g., Lundberg, Olofsson and Wall, 1980), with the child placing a peg in the board for every sound detected; (4) phoneme substitution, (e.g., rose becomes toes): and (5) phoneme blending, the child arriving at a solution by combining sounds. This often takes the form of a game. The examiner may have a picture face down on the table. The sounds are the child's "clues," which enable him to guess the contents of the picture. The second type of task, which requires knowledge of larger units, may include syllable tapping (which may be confounded by the motor component of this task), or counting, or may involve some type of rhyme or alliteration paradigm. Rhyming and alliteration tasks have been called implicit (Bowey and Patel, 1988), as direct manipulation of segments is not required, but knowledge of segments is implied in ability to complete the task. It was thought at one time that sensitivity to rhyme on the part of pre-school children indicated an ability to analyze the constituent sounds of words (Bryant, MacLean and Bradley, 1990), that is, an ability to break the speech stream into phonemic segments, but further analysis demonstrates that children can do many of these tasks by paying attention to onset/rime segmentation. Whether all permutations of this task can be solved with this level of analyis is still not determined.(See below for further development of this topic.) 6 A second division between tasks separates those which require making a judgment and those which require production of a solution either from a given array or from the child's own general language knowledge. These tasks vary in difficulty and the type of analysis the child is required to perform. For example, a very simple rhyme judgment task that may be given to children three or four years old is to ask them if a certain word rhymes with the name of a puppet held by the experimenter, e.g., does bed rhyme with Jed? (Smith et al., 1982). In this type of task one stimulus item remains constant, so that the child is only required to analyze the rime segment of the second stimulus. The difficulty of this task can be increased by varying both stimuli and by making non-rhyming pairs minimally (e.g., does hem rhyme with head), rather than maximally (e.g., does dog rhyme with head), different. The level of analysis required for a rhyme oddity judgment task can be increased by asking the child to judge either which two of an array of stimuli items rhyme (or are the same), or alternatively, which member of the array does not rhyme (or is different) (Bryant et al., 1990). An array might include three or four items. At this level memory becomes a factor so that some studies include picture support in order to control for this possible confound (e.g Bryant et al., 1990; Bowey et al., 1988; Katz, Healy and Shankweiler, 1983). However, Bryant and Bradley (1983) chose four items (without picture support) for their older subjects, on the grounds that using only three items made the task too easy. A further variation on the rhyme judgment task is to provide two or three possible rhymes for a stimulus item and ask the child to pick the correct one from this array (Knafle, 1984; Kamhi, Catts, Mauer, Apel and Gentry, 1988). Again, the amount of difference between the correct response and the distractor item can be manipulated. However, if the differences are too 7 great there is the possibility that the child may focus on some other cue besides the target one. Rhyme production tasks generally require a child to supply a rhyme for a given stimulus word. As the task is designed to test phonological awareness rather than lexical retrieval, researchers usually score both nonsense words and real lexical items as correct, provided they meet the criteria for making the rhyme (e.g., Lundberg et al., 1980). However, it is possible that the ability to provide a lexical item as opposed to a nonsense word is indicative of a developmental process in rhyming ability. It is also possible that the favoured onset phoneme in the creation of nonsense words may have some significance within the child's own phonological development. Since onset/rime segmentation skills emerge earlier than phoneme segmentation skills (Bowey et al., 1991), it is more likely that the child's sensitivity to rime will allow the child to categorise rhyming words, rather than sensitivity to final phonemes. Tasks that tap rime segmentation skills should, therefore, be better predictors of reading skill than phonemic segmentation tasks. Validity of Phonemic Awareness Tests Given the number of tasks, and the variation within the tasks, it is important that their reliability and validity be measured in some way, in order to achieve a consistent operational definition. Typically, the method used to achieve this involves giving a number of phonemic tasks to a single group, e.g., Lundberg (1980), Stanovich, Cranmer and Cramer (1984), Yopp (1987). Stanovich et al. (1984) found a high correlation between non-rhyming tasks and a measure of reading ability done one year later, but not between the rhyming tasks and the reading 8 task. There were only weak, or non-existent correlations between the rhyming tasks and the non-rhyming tasks. However, the most difficult rhyming task of the three given consisted of choosing a rhyming word for a stimulus item from a group of three items. This task varies from a straight rhyme judgment task only in its increased memory load (unconnected to metaphonological ability) and in the subject naming the item rather than simply answer "yes" or "no". Yopp (1987) undertook a study of a variety of tasks used to test phonemic awareness, in order to determine the reliability, construct validity, and predictive validity for reading acquisition. She found that most phonemic awareness tests correlated significantly, with rhyming tasks having the lowest correlation. Only three of the tests studied had a reliability co-efficient of .85 or above. These included a phoneme blending test, a phoneme deletion test and a phoneme segmentation test. A factor analysis of the tests indicated, as in the Stanovich study, that rhyming tasks tapped a different underlying ability than other phonemic awareness tasks, for which one factor accounted for 58.7% of the variance. The fact that rhyme tasks apparentiy differ from other phonemic awareness tasks by the size of the unit undergoing analysis, may partially account for this finding. Rime Awareness vs. Phonemic Awareness There is evidence that phonemes have some sort of constituent organisation within the syllable rather than simply being strung together (Treiman, 1983). In one major syllable theory, the two constituents are referred to as the onset, which is comprised of the initial consonant or consonant cluster, and the rime, which consists of the syllable nucleus (usually a vowel) and any following consonants. 9 (>iset/rime segmentation, as demonstrated in rhyming tasks, appears to be one metaphonological skill that is capable of producing ceiling effects in non-literate populations (children and adults). Children apparently learn to segment syllables into onset/rime constituents without any linguistic training beyond general langauge development. There is evidence that children are sensitive to onsets and rimes as units of speech production. Treimann (1983) found, in a series of "word games," that children were more likely to make substitutions that adhered to a rime/onset syllable segmentation pattern than ones which broke up word-initial clusters or word-final VC segments. The same developmental acquisition does not appear to occur for phonemic segmentation. Although some pre-readers are able to have some success at these types of tasks (less than 10% correct responses), phonemic segmentation is a task that only commands high scores after exposure to reading and spelling (Bowey et al., 1991). In summary, although researchers have categorised a number of tasks as metaphonological, it appears that not all of them (or even most of them) emerge as part of general phonological development. Relative Difficulty of Tasks While acquisition appears to depend on age and experience, the order in which skills are acquired is connected to their relative difficulty. As mentioned above, tasks which require phoneme manipulation appear to be more difficult than tasks requiring syllabic segmentation. Magnusson and Naucler (1990) rated phonemic awareness tasks in the following order of difficulty: rhyming, identification of phonemes, segmentation of syllables and segmentation of phonemes. The authors found that this order of difficulty held for normal and disordered children. Similar results were found in the comparative study undertaken by Stanovich et al. (1984) who found that kindergarten children had most difficulty "stripping" an initial consonant (phoneme 10 segmentation). For example, given the stimulus word "win", a child has to determine that "in" is the word that remains when the initial sound is removed. Stanovich and his colleagues, in the same study, found evidence that rhyming tasks may be at ceiling at kindergarten level. It should be noted, however, that only one or two tasks for each category was tested, and so it may not be possible to draw general conclusions about children's ability in each area. Other researchers such as Bryant and Bradley (1983, 1989), and Bowey and Patel (1988), did not find evidence of a ceiling. However, Bryant et al. (1989) do observe that alliteration and rhyming tasks are the only phonological awareness tasks that can be reliably administered to pre-readers. Age of Acquisition Determining a normal age of acquisition for any metaphonological skill is confounded by the variety of tasks and their implementation, as well as by the general conclusions drawn from non-generalised tasks. Two factors seem to be important in deciding whether a child might be expected to perform successfully on a given task. One is the specific requirements of the task, and the second is the type of language input the child has received, in particular the child's literacy training. However, even accounting for these two factors, results of different studies vary greatiy. Grunwell (1986) contends that children below 4;0 to 4;6 give litde evidence of having metaphonological skills, but she does not define what tasks might be used to determine this. She discusses phonemic segmentation, but this is a task which other studies have shown to be among the most difficult of metaphonological skills. When tasks more adapted to a child's age level are employed, results indicate a lower age of acquisition. For example, Smith et al. (1982) found that 11 children with a mean age of 3;6 were able to correctly judge whether a word rhymed with Jed 67% of the time, improving to 79% correct judgment by 4;7. However, not all the children were able to do this task beyond chance. Only 28% of the 3 year-old and 33% of the 4 year-old group could do this, suggesting that this type of task is one children either can or cannot do; i.e., it would produce a bimodal curve where the transition between ability and disability is short. This conclusion is not always supported by other research. In contrast somewhat to these findings on rhyming ability another study found that at 5; 6 children could still only perform at chance level on a two-word rhyme judgment task (Wagner and Torgeson, 1987). This finding, however, is itself contradicted by other studies by Bryant and Bradley (1990), who found that at an average age of 4;7, 50% of the children studied could perform significantly better than chance (7/10) on a simple rhyme judgment task. The ability of younger children Opre-literate) to complete rhyming tasks is demonstrated by the findings of Knafel (1972), who gave a more difficult forced choice rhyming task to kindergarten children. Knafel's stimuli and responses consisted of one syllable words of the shape CVCC. She alternately changed the third or last phoneme to discover whether phoneme position affected ability to choose the correct rhyming word. She found that the children were able to make correct responses concerning rhyme with CVXC words 75% of the time and CVCX words 55% of the time, a result significantly better than chance for the CVXC condition. The correct responses increased to 82% and 62% for first grade, 86% and 76% for second grade, and 93% and 78% for third grade. These results suggest a more normal distribution of ability than that suggested by Smith et al. (1982). 12 To assess the metaphonological skills of younger children, Bradley and Bryant (1983) gave a rhyme judgment task similar to that employed by Knafel, but with picture support and CVC stimuli, to three groups of children at the pre-school, kindergarten and first grade level The authors found that, although age was a factor, the pre-school group was still able to perform at a level significandy above chance. The authors manipulated the number and position of phonemes differing between the stimulus and response items and found that there was a significant effect on performance (it decreased) when there were two similar phonemes in the response and stimulus items, and when an identical phoneme occurred word initially. There was no effect resulting from similarity of the medial vowel. The hypothesis that children use the strategy of segmenting the rime and onset is corroborated by the finding that children had most difficulty with response items that shared the initial and medial (vowel) phonemes with the stimulus item. It also agrees with Treiman's (1983) findings that children have more difficulty manipulating syllabic constituents in a CV/C pattern. So, although there is not complete agreement about the age a child may be expected to acquire the ability to complete rhyming tasks, the majority of studies demonstrate that children have the ability to perform rhyming tasks at kindergarten age, before formal literacy training begins. In contrast to this, the ability to segment words phonemically appears to be a skill that develops simultaneously with beginning literacy at age 6 (Wagner and Torgesen, 1987). While many authors have found similar results, some researchers have found that even 3-year-olds can perform a phonemic segmentation of words, if the task is adapted to their level of development (Fox and Routh, 1975). 13 It is clear that age of acquisition is task dependent. There seems to be a developmental process which has a fairly wide age span for the acquisition of individual skills, with even individual skills, such as rhyme supply, appearing to have sub-stages. Relationship between Linguistic and Metalinguistic Skills Studies on the ability of pre-literate children and children in the early stages bf literacy to solve metaphonological tasks have generally failed to explain why some children should have more difficulty than others on these tasks. In those studies which address the origin of phonological awareness, some relation with linguistic ability is postulated. Three general hypotheses have been generated to describe this interaction. Autonomy hypothesis This states that the acquisition of metalinguistic skills is distinct from the early acquisition of basic comprehension and production processes of language, while later linguistic processes (such as reading and writing) are facilitated by developing metalinguistic skills. Under this hypothesis linguistic and metalinguistic skills do not develop simultaneously. The suggestion here is that metalinguistic skills develop later in childhood, between the ages of six to eight (Smith et al., 1982). Their development is related to other developments in information processing ability that occur at the same time (Morais, 1986). Interaction hypothesis This hypothesis asserts that metalinguistic and general linguistic skills each develop under the influence of each other. This two way interaction supports the possibility of certain metalinguistic skills, such as self-correction in 2- and 3-year olds (Smith et al., 1982), being acquired early in the pre-school years. Skills of this type may originate with the development of 14 error detection mechanisms (Bradley and Bryant, 1983). Bryant, Bradley and MacLean (1989) hypothesise an interaction between early knowledge of nursery rhymes and developing phonological skills. The knowledge of nursery rhymes is considered to increase the child's sensitivity to phonological segments, which increases the ability of the child to perform the phonological manipulations necessary for reading. Magnusson and Naucler (1990) provided evidence that a certain type of phonological disorder was more likely to result in difficulties with metaphonological skills. Children with segmental difficulties appear to develop normal metaphonological skills, while those who have sequential (syntagmatic) problems are more likely to have a metaphonological (and reading) difficulty. Grunwell (1986) sees metaphonolgical development as a continuation of phonological development; the new phonological skill is a natural outcome of previous maturation. Third variable hypothesis This hypothesis contends that both metalinguistic ability and literacy develop under the constraints of some third variable, such as general language ability or IQ. Bowey and Patel (1988) found that there was a higher correlation between syntactic awareness and word decoding skill than between syntactic awareness and reading comprehension, a finding that is not predicted by the view that syntactic awareness should contribute more to text integration processes. The authors hypothesised that both syntactic awareness and word decoding are related to a third variable which they thought might be either underlying general metalinguistic ability, or general language development, operating separately from IQ. 15 If metalinguistic ability and reading ability are both related to underlying general language development then there should be no variation in reading achievement accountable to metalinguistic ability after general language ability is partialled out. Bowey et al. (1988) felt that their statistical analysis showed that this was indeed the case. If this is the case, there should also see very strong correlations between tests of general language development and metalinguistic development, a correlation which Magnusson and Naucler (1990) found did not hold consistently. Rather, they found variability among both the language disordered and the normal populations in regard to metaphonological ability. They concluded that linguistic ability may not always be a predictor of metalinguistic ability. The Role of Cognitive Development in Metalinguistic Ability A question that is addressed in the literature either directly or indirectly is the degree of general cognitive ability involved in performing metaphonological tasks. While researchers agree that cognitive development is an essential factor, they are generally of the opinion that metaphonological ability is more directly related to some aspect of language development, via one of the mechanisms proposed above, (e.g., Stanovich et al., 1984; Bryant et al., 1990; Bowey et al., 1988). However, there is also a body of opinion that holds that metalinguistic skills (and language skills in general) require the same abilities as other cognitive tasks (e.g., Bialystok, and Ryan, 1985). Bialystok and Ryan argue that metalinguistic skills require an analysis of knowledge and control over that knowledge, and that this analysis and control are cognitive processes. Because metalinguistic skills apply in situations of decreased context, where there is a shift from content to form, they require a higher level of analysis and a higher level of control over linguistic 16 knowledge. They contend that analyzed knowledge and control form two dimensions of cognitive processing. The child has to access the knowledge, integrate it, and arrive at a solution, e.g., a new word with the correct properties, generally within a time constraint This requires cognitive control. Control is separate from analyzed knowledge - having one doesn't assume the other is available, or that it need to be equally accessed to complete a task. However, inequality of development along either dimension restricts expression of more advanced development along the other dimension. Metalinguistic skills, then, are not separate from other language skills but require that a child reaches a certain point along both dimensions to complete a certain task. Using this hypothesis, a child creating a rhyme, for example, would first need to analyze a word to arrive at its constituent parts, rime and onset, and then use (or control) that analyzed knowledge to arrive at a solution, e.g., a second word that shares the rime or the onset. Tunmer, Herriman and Nesdale (1988) agree that metalinguistic skills develop only when a certain level of cognitive ability has been reached. They argue that even a child who has only a little, or even no metalinguistic ability, can learn to read without difficulty, provided the child's cognitive ability has developed to the point where metalinguistic skills can be acquired. Lundberg, Olofsson and Wall (1980) also suggest that, in order for metalinguistic skills to develop, general intelligence should reach a pre-requisite level and the child should have the cognitive ability to analyze non-verbal tasks. Other researchers who are exploring the correlations between metaphonological ability and early reading success are aware that it is necessary to determine how much variance in reading ability is accounted for by different skills, such as IQ and language ability, and conditions such as social economic status, but reject a completely cognitive explanation. MacLean, Bryant and 17 Bradley (1987) found that early performance in rhyming tasks predicted success in reading even after the effects of IQ had been partialled out, suggesting that a cognitive account is not sufficient in looking at children's performance on metalinguistic tasks. Bowey and Patel (1988) found that children who have difficulty with phonemic segmentation are still able to perform analogous, non-linguistic segmentation tasks. They did find, however, a correlation between IQ and metalinguistic ability. Magnusson and Naucler (1990) were also unable to find a one to one correspondence between non-verbal cognitive level and metalinguistic skills. Interaction Between Metalinguistic and Literacy Skills While this study is more concerned with the possible origins and/or pre-cursors of metaphonological delay, it is necessary to examine the interaction between literacy and metaphonological development because of the body of research that suggests that metaphonological skills emerge at about the same time as a child is exposed to formal literacy training. The interaction or link between these skills may not be the same for all tasks, which suggests that not all tasks tap the same skill. Wagner and Torgesen (1987) oudined four possibilities described by Ehri (1979). Prerequisite A particular phonological ability may be a pre-requisite to literacy. This is a strong claim with its suggestion that if, for example, a child fails to leam to rhyme, he or she will be unable to read and spell. Facilitator A second, less stringent alternative, is a facilitative interaction whereby learning to read is aided by a particular metaphonological ability. This suggests that a child may learn to read and 18 spell but not as easily or as well as a child who has mastered the earlier skill. One factor may be English orthography which is derived from phonology. Phonological awareness allows the beginning reader to make a correspondence between the written symbol and the sound which would be difficult in the absence of phonological awareness. Also, in order to segment letter strings and blend sounds to produce words one must be aware that phonemes exist. Explicit phonological knowledge may then be enhanced from increasing literary skills. Bradley and Bryant (1983) studied four groups of children over a four year period where the amount of training varied with each group. They believe that their longimdinal data which covers shows a strong causal link between phonological awareness and reading. Consequence A third alternative is that metaphonological ability is a consequence of literacy, and so will not emerge until a child has developed some reading skills. Because phoneme segmentation skills emerge at about the same time as literacy skills, this is a very difficult hypothesis to prove. Morais (1986), in her study of Portuguese adults, found that those subjects who were unable to read were also unable to do phoneme segmentation tasks, which suggests that the ability to segment phonemically the speech stream is only acquired after exposure to literacy. Bowey and Francis (1991) contend that the child's reading ability does not have to be extensive to facilitate phoneme segmentation skills, but rather that the proficiency to read ten words may be sufficient to effect a shift from visual to phonological processing. Bradley and Bryant (1978) question this third hypothesis, as they say it does not explain why older children who are delayed readers but who have had as much exposure to reading as good readers still perform more poorly on tests of phonological awareness. 19 Correlate Most studies, in the absence of more conclusive data, prefer a fourth alternative which is that the development of some metaphonological skills are correlates of literacy with the amount and direction of causality undeterrnined. The division between syllabic and phonemic awareness is again evidenced. Syllabic segmentation, as demonstrated in rhyming tasks, is likely to be the subject of the first two hypotheses while phonemic segmentation occupies the arguments supporting the last two. Magnusson and Naucler (1990), for instance, found that speech disordered children who did unexpectedly well at reading and spelling had also generally scored higher than the mean for normal children in metaphonological tasks while those children without speech and language disorders who failed to do well at reading and spelling also had difficulty with metaphonological tasks. However, the correlations they found were between rhyming and reading and spelling but not between phoneme segmentation and identification and reading and spelling. They found this to be the case for all subjects, not only those who performed unexpectedly well or poorly. Alternatively, other researchers (e.g., Lundberg, Olofsson and Wall, 1980; Stanovich et al., 1984) have found the ability to perform phonological manipulations, such as word reversals and identification of phoneme position in words at the pre-reading level, (age 5 and 6), to be highly predictive of later reading success. Lundberg et al. also adrninistered a rhyming task and found this less predictive (though not non-predictive) than the phonemic tasks. However, the rhyming task used was a rhyme supply task and nonsense words were accepted as correct. A separation of children who used nonsense words from those who used real lexical items may have revealed slightly different results, as may other types of rhyme tasks. Also, age appears to 20 be a factor in the predictability of rhyme and reading. Bryant and Bradley (1990) found that when a child is about 6 years old the correlation between rhyme and reading is no longer significant. This may be because as a ceiling is reached in rhyming tasks, even children with poor reading skills are able to do them adequately. The data collected by Magnusson and Naucler (1990), and by Lundberg and his colleagues (1980) in Sweden is interesting, partly because children begin formal literacy training about a year later than North American children. As they also appear to be a little later in developing phonological segmentation skills, there is some reason to suppose that it is exposure to literacy rather than development that facilitates at least some metaphonological skills. Goals of the Study While the results produced by the research is variable, some points emerge which suggest that metaphonological delay may result from an earlier phonological deficit It appears that children acquire the ability to complete some metaphonological tasks prior to learning to read or spell, which indicates that metaphonological ability has its roots in skills which emerge prior to literacy. This skill may be general language ability, as suggested by Bowey and Patel (1988), or it may be more specific. Magnusson and Naucler (1990) have found a significant correlation between certain class of phonological disorder and success on rhyming tasks, while Grunwell and others maintain that metaphonological development proceeds from phonological development. The finding that syllabic simplification processes previously thought to be suppressed apparendy re-emerge as a child learns to spell suggests that disordered phonology may continue to affect later aspects of development even after adult forms are achieved. 21 The correlations which have been made between metaphonological ability and cognitive development, general language ability and socio-economic status, emerge from studies mainly undertaken on large samples of the normal population. Children who fell beyond normal limits in pre-study language and IQ tests were often eliminated from the study. For this reason, correlating deficits may have gone undetected. The suggestion of a correlation between early phonological delay and metaphonological skills and the lack of individual data from vigorously controlled tasks led to the following aims for this study: (1) To investigate the possible effects of a phonological disorder on metaphonological development, (2) To develop a more rigorously controlled set of metaphonological tasks for assessing children's ability to consciously control aspects of their sound system than demonstrated elsewhere, (3) To provide more detailed individual subject results on the set of tasks than is available elsewhere. 22 CHAPTER 2 METHOD In order to overcome some of the methodological problems related to large sample studies that were discussed in the introduction, this study was restricted to three subjects: two with a history of phonological delay, and one a typically developing control. Rather than measure performance on a battery of different metaphonological tasks, ability in only two skill areas was assessed, rhyme and alliteration. By assessing only two skill areas across three subjects, an in-depth analysis could be made of both the task types and individual subject performance. Rhyme and alliteration tasks were chosen for two reasons. (1) because of the evidence, as described in the introduction, that they give the most reliable results in the pre-literate population, and (2) because of the hypothesised predictive link (e.g., Bryant and Bradley, 1983) between rhyme ability and early reading success. In addition, in a pilot study it was noted that rhyming and alliteration oddity tasks still presented difficulty for one of the phonologically delayed subjects at age 6;4, in spite of claims by some researchers that a ceiling effect is operating by this age level. Subjects As noted above, three children served as subjects for this study, two of whom had documented phonological disorders, and one of whom was a typically developing control. All the subjects were in first grade (three different schools). Prior to the study, the subjects were given 23 tests of general language ability and development. These included the Peabody Picture Vocabulary Test-Revised (Form M), the Expressive One Word Picture Vocabulary Test, and the Test Of Language Development (Primary-Revised). In addition, all the subjects were given an IQ test using the WISC-R. (See tables 2.1 and 2.2 for specific results.) Subject 1 was 6;6 at time of testing. His father is a teacher and his mother a medical receptionist. He has one older sibling of whom there has been no reported speech or language disorder. The subject received speech therapy from 3;4 to 3; 10. (See table 1, appendix A, for a summary of his early phonological system.) The subject also had some minor delays in language development, notably in the areas of auxiliary and modal verbs, spatial concepts and deictic terms. He received two blocks of language therapy, the first from 4;2 to 4;6 and the second from 5;5 to 6;3. Language samples obtained in the last fifteen months indicate a word retrieval difficulty. His grade 1 teacher reported that at 6;7 his reading skills were within normal limits but that his writing skills were delayed. His early sound system, immediately prior to therapy, is summarized in table 2. With the exception of initial /r/, for which he occasionally substitutes [w], the subject no longer has overt phonological difficulties. As can be seen in tables 2.1 and 2.2, the results of the language tests are scattered and indicate some remaining weaknesses for SL, but even the lowest scores on the TOLD-2 are within normal limits. The noticeable difference in his scores on the PPVT and EOWPVT is probably a consequence of a word retrieval difficulty. Subject 2 was 6;7 at the time of testing. His parents are separated and he lives with his mother who has completed high school and now operates a family day care. He has one older sibling, who has a learning disability. S2 received speech therapy from age 3;5 to 3;11. 24 Table 2.1. - Summary of vocabulary and IQ scores (given in standard scores) Subject EOWPVT PPVT WISC-R Verb. Perf. FuU Sc. SI 97 116 46 58 104 S2 99 108 51 70 121 S3 145+ 131 73 68 141 Table 2.2. scores) —Results of Test Of Langauge Development (Primary-Revised) (given in standard Subject Picture Oral Vocab. Vocab. TOLD-R (Standard scores)' Grammat. Underst. * Sent. Imit. Grammat. Compreh. Word Disc. si 16 12 9 9 14 9 S2 5 10 13 6 9 9 S3 12 15 11 14 15 14 *Mean of 10 +/- 3 He received language therapy from age 4;3 to 4;7 and again from 5;6 to 5;8. (See table 2, appendix A, for a summary of his early phonological system.) He has currendy acquired all phonemes except hi. The voiceless interdental /Sf is inconsistent. He has difficulty with words 25 of more that two syllables and phonological processes include syllable reduction and transpositions. He is presently experiencing some difficulty in learning to read and write. Language scores for S2, completed as part of this study (see tables 2.1 and 2.2), show weakness in sentence imitation. The word discrimination sub-test score on the TOLD-R was only just within normal limits. This test was initially administered by a speaker of British English and produced results that were little better than chance, (13/20). The test was re-administered by a native Vancouver speaker, and then again by the original speaker, and the results were inconsistent. For purposes of test scoring those items on which he produced the correct result two or more times were considered correct. Subject 3 was 6; 11 at the time of testing. His parents are university graduates. He has never required speech or language therapy, and is experiencing no difficulties with the grade 1 literacy curriculum. S3 was well within normal limits in all areas of testing, but, as can be seen from the results, both .SI and S2 had scores that were higher than S3's on at least one sub-task. The scores on the WISC-R were high for S3 (130 and 128), but the Performance Scale score was matched by S2. SJ. had a lower Performance scale score, though still in the high end of average. The verbal score for both Sl_ and S2 were notably lower than their Performance scale scores. Stimuli The verbal stimuli used in this study were selected to control for: (1) word frequency and the subjects' vocabulary, (2) the subjects' available sound patterns at the pre-treatment stage, and (3) phonological form. 26 Word Frequency and Subjects' Vocabulary Using words that are familiar to a subject is necessary to prevent a child avoiding the correct solution when confronted with an unknown item. Metaphonological tasks should be independent of meaning, and therefore a subject should not be distracted by an unfamiliar word. A second alternative is to use nonsense words, but this would preclude the using of pictures. Also, although nonsense words are structurally consistent with the phonology of the language, there is little theory to account for their absence. To control for word frequency, stimuli were chosen from Kucera and Francis (1982), and as a further control, subjects had to show that they were familiar with the items chosen. This was done by having them match one of array of pictures (later used in the oddity tasks) to a word given by the examiner. Any words that the subjects failed to identify immediately were removed from the task. (See appendix B for stimuli lists.) General Phonological Form The opinion of most researchers at the present time (e.g., Bowey and Patel, 1990) is that rhyme tasks involve the segmentation of the onset and rime, and not of the individual phonemes. In order to better understand the size of the unit to which the subjects were paying attention, two conditions for vowel quality were applied. In the first condition, the distractor item contained a vowel similar to that in the paired items for the rhyme tasks, e.g., dog rock lock, while in the alliteration tasks the distractor item contained a vowel similar to that in one of the paired items, e.g., meet man knee. Here the vowel in the distractor item, knee, is similar to that in meet, one of the paired items. In the second condition the vowel in the distractor item varied from the vowels in the paired items. 27 The items chosen for the rhyme and alliteration oddity tasks were all of one syllable in length. The majority of the items were simple CVC shapes but some items contained clusters. Where an item contained a cluster, all three items in the array contained clusters in order to avoid the possibility of a subject mistakenly using the cluster as an indicator. Pre-treatment Sound Patterns In order to control for the available sound patterns of SI and S2 at the pre-treatment stage, the experimenter E examined the phonological data collected by Bernhardt (1990). Because SI and S2 had acquired different phonemes at the pre-treatment level, different sets of stimuli for each one had to be derived for tasks relating to those stimuli. S3, who had no reported phonological deficits, was administered the same stimuli as SL An unfamiliar, native Vancouver, female adult (not E) recorded the stimuli, in order to control for any vowel quality that might be dissimilar to that of the subjects. Tasks Several different sub-tasks for rhyme and alliteration were designed and presented to the subjects. The sub-tasks were devised to test the effect of several variables. The first variable, the early phonological systems of the subjects, was incorporated into the tasks as described above. Tasks which reflected phones that were both available and unavailable were given in order to assess a general effect for phonological delay. A second variable was the degree of difficulty intrinsic to the tasks. As was demonstrated in the introduction, there are contradictory opinions concerning the age of acquisition of the ability to complete rhyming tasks. Most studies only administer one type of rhyming task, an omission that has led to some of the contradictions. In order to reveal any possible hierarchy of 28 ability within a particular skill area three types of tasks for both rhyme and alliteration were aclministered. In order to complete the rhyme (or alliteration) oddity tasks, the subject has to retain all three items in working memory in order to process them. In order to alleviate the load on memory, picture support, in the form of black and white line drawings, was provided for four of the five tasks. The fifth task was given without picture support, in order to test the role of memory in this task. Twelve tasks were administered to the subjects: seven of these were rhyming tasks while five were alliteration tasks. (See appendix B.) All the rhyme and alliteration oddity tasks were randomised and administered twice. Although there were three possible answers to the oddity tasks, two administrations of the same task were considered sufficient. There were several reasons for this. One was that previous experience with SI suggested a strong learning effect; secondly, the tasks, although testing different variables, were similar enough to become quite tedious to the subject, a factor which could possibly confound the results. It was also thought that two sets of results for each task would give the required information concerning ability, or alternatively, if two sets of data were very contradictive, a third set was not likely to be conclusive. Rhyme Oddity Tasks In the rhyme oddity task the child was required to identify which of an array of three items did not rhyme with the others. This type of task has been administered in other studies with an array of four items (e.g., Bryant and Bradley, 1983). The possibilities of the array were AAB (where B is the distractor item), ABA and BAA. In the randomisation procedure each array had 29 a similar chance of being selected. There were ten items in each task (i.e., 30 stimulus words). However, as the stimulus lists were not divisible by three one array was presented more than the others in each list. There were five sub-tasks which reflected the subject's phonology at the pre-treatment stage. These are labelled Tasks 1 to 5 and examples of each task are given below. Tasks 1 to 4 were presented with pictures. Task 1 Task 1 used only those word final phones unavailable to the subject at the pre-treatment level for the matched item, while the distractor item contained the word final phone most commonly substituted by the subject. For example, in the array bed head gate, /d/ is the unavailable phone, while [t] is the substitution. Task 1 had a dissimilar vowel in the distractor item. Task 2 Task 2 used the same type of word-final phones in each item as task 1, but the vowel quality in the distractor item was similar to that in the matched items; e.g., fish rich wish. This was considered to be more difficult as a rhyming task because only the final phone differed in the distractor item, rather than the whole rime. Task 3 Task 3 used word final phones available to the subject prior to treatment and had a dissimilar vowel in the distractor item. The example hen pen sea indicates that S_l could use word-final /n/ or no word-final consonant at the pre-treatment stage. The vowels in the distractor items differed from the vowels in the paired items. 30 Task 4 Task 4 used the same type of word-final phones as in Task 3, but had a similar vowel in the distractor item, e.g., train pay rain.  Task 5 Task 5 used the same format as Task 2, though with different lexical items, e.g. beg neck  check. No pictures were presented. This task was used to assess the role of visual support in task completion. Rhyme Judgment Task A rhyme judgment task using two stimuli per item was used to control for any inhibition in the child's ability to judge rhymes, due to the presence of a distractor item. This was considered an easier task than the rhyme oddity task because (1) it only required a comparison of two items, and (2) the child only had to give a yes or _no answer. Rhyme Supply Task Due to observations from the pilot study that it may be easier for a child to produce rhymes than to make a rhyme oddity judgment, a rhyme supply task was included in the study. The subject was required to produce an item which rhymed with a stimulus item. Stimulus items were derived from two sources: Kucera and Francis (1982), and Si's language samples. Alliteration Oddity Tasks (9-11) Three alliteration oddity tasks were presented. Alliteration for all alliteration tasks was considered to refer to the initial phone in a word rather than the onset. Where clusters were used in the oddity stimuli, the paired item also had an onset cluster, e.g., scarf scare car, but this was 31 not necessarily the case in the alliteration judgment task, e.g., free five. The subject's task was to identify the item that did not have the same initial phoneme. Picture support was given for all the alliteration oddity tasks. As in the rhyming oddity tasks the stimuli were prepared from Bemhardt's (1990) early phonological data. Task 9 Task 9 used word initial phones available to the subject at the pre-treatment stage. The vowel in the distractor item was similar to one of the vowels in the paired items. The paired items matched only on their initial phone. For example, in the array time two fight, Ix/ and IfJ were both available, while the vowel in the distractor item, fight, was the same as the vowel in time-Task 10 Task 10 used the same initial sounds as in Task 9, but the distractor item differed in CI, V, C2. The paired items were only matched only on CI, e.g., fat ten team. Task 11 Task 11 used word-initial phones unavailable to the subject prior to treatment, contrasted with the phone most commonly substituted. The vowels had the same distribution as in Task 9. For example, the array feed fit thick represents the substitution of [f] for /of. Alliteration Judgment This task required the subject to say whether or not two verbal stimuli had similar initial phones. The stimuli for this task were based on the early phonological data of both SJ. and S2 (i.e. there were two separate lists). Initial sounds were paired with an identical sound, and also 32 with the sound most commonly substituted by the subject at the pre-treatment stage. The items were randomised. No picture support was given for this task. Alliteration Supply This task required the subject to produce a word that had the same first sound as the stimulus. Procedure Subjects 1_ and 3 were tested individually over a period of five weeks, with one session per week. S2, was tested over three weeks, with two sessions per week in the last two weeks. Generally four tasks were administered in a session, although this number varied depending on the child's state of fatigue or distractibility. Alliteration and rhyming tasks were not administered during the same session, except for S2 in the last session. First and second randomisation of the same task were also given on separate days. Instructions Before each task, the subjects were given instructions about the task. For the rhyme oddity tasks the instructions were: You will hear three words. Two will rhyme and one will not. It will have a different ending. I want you to tell me which one is different, the one that doesn't rhyme.1 To ascertain that the child understood the concept E said the nursery rhyme "Jack and Jill went up the hill" and asked the subjects which words rhymed. In the alliteration oddity tasks the child was told: *Exact wording may have varied slightly, but the intent was the same. The goal was comprehension of task requirements. 33 You will hear three words. Two of them start with the same sound, and one of them starts with a different sound. So if I said to you cake and cup, do they begin with the same sound? (Wait for child's response.) That's right, they do. What about cake and dog? (Wait for child's response.) That's right. They are different. For the rhyme and alliteration judgment tasks the child was asked to say respectively whether or not two words rhymed (or ended the same) or whether they began with the same sound. In the supply tasks the child was asked to provide a word that either rhymed with the stimulus item or began with the same sound. Two training items were given for each task in order to ascertain that the subjects understood the instructions and were able to do the task. As a further control, subjects were required to tell the test administrator what they were supposed to do. For the rhyme and alliteration oddity tasks the pictures were placed in front of the subject, slightly after the recorded item was delivered on the tape recorder. This was to prevent any noise from the placement of the picture interfering with the child's comprehension of the item. It was thought to be preferable to present the pictures after the item had been heard so that the subject did not label the picture incorrecdy prior to hearing the stimulus. This was a possibility as some of the items were abstract terms. The pictures were only to alleviate the memory load rather than be a substitute for the spoken item. The child was given as much time as he needed to complete the task. When the child had made his selection the administrator gathered up the pictures and recorded the result. The second set of items was then given. Where no pictures were involved the response was recorded and the next item played. At various points between the administration of items the child was either reinstructed, or asked to tell the administrator the rules of the task, to ascertain that the child had not forgotten them. 34 CHAPTER 3 RESULTS: PART 1 - SUBJECT SCORES Phonological awareness tasks The results of each task are given in Tables 3.1 - 3.5. For purposes of presentation, the tasks have been tabled in three groups: (1) rhyme oddity tasks, (2) alliteration oddity tasks, and (3) supply and judgment tasks. The control subject (S3) was able to perform with 100% accuracy on ten of the twelve tasks (eighteen out of twenty trials), with a total of two errors overall. Because of his high performance score, combined with the facility with which he completed the tasks (see below for further discussion), it was considered unnecessary to administer a second trial of most tasks. An exception was made for task 5, on which he made one error. Subject 1 achieved scores of 70%, or better on most tasks. (Where there were three possibilities for each item, chance was at 3.33/10, and 7/10 was significantly better than chance.) His responses did not reach significance on one trial of the rhyme oddity task that was given without picture support (Task 5a). Subject 2 attained a significance-level score on only 10 of the 20 possible trials, and scored at chance level on four trials. The latter included the first trial of the rhyme oddity task (without picture support), the rhyme judgment task, and both trials of the alliteration oddity task based on phonological segments unavailable to the subject prior to treatment. Statistical analysis, 35 using the Wilcoxon's signed-rank test, showed the differences between all three subjects to be significant, p < .001, for both the rhyme oddity tasks and the alliteration oddity tasks. The subjects ranked in the following order for both sets of tasks: S3, SL, S2. Table 3.1. - Results of rhyme oddity tasks 1, 2 and 5. TASK si S2 S3 Task la 9/10 8/9 10/10 Task lb 9/10 6/9 ~ Task 2a 9/10 9/10 10/10 Task 2b 9/10 7/10 Task 3 a 9/10 6/10 10/10 Task 3b 10/10 8/10 ~ Task 4a 9/10 5/10 10/10 Task 4b 8/10 7/10 — Task 5a 7/10 3/10 9/10 Task 5b 6/10 6/10 10/10 Tasks are designed to test word final contrasts unavailable to the subjects at the pre-treatment level. Available sounds are represented in tasks 3 and 4. Task 5 is presented without picture support. A comparison of tasks accompanied and unaccompanied by picture support, but with the same contrasts, (tasks 2 and 5) shows a difference in performance for SL, S2, and S3. J51 achieved a score that was lower by 20% than that achieved with picture support, and S2 36 demonstrated an overall decline in performance of 35%. S3 failed one item in the first trial of task 5. Table 3.2 .-- Results of alliteration oddity tasks °* Alliteration Tasks Score SI S2 S3 9a 7/10 (.7) 8/10 (.8) 10/10 (1.0) 9b 10/10 (1.0) 6/10 (.6) 10a 7/10 (.7) 7/10 (.7) 10/10 (1.0) 10b 9/10 (.9) 6/10 (.6) 11a 10/14 (.71) 4/11 (.36) 10/10 (1.0) l ib 12/14 (.86) 4/11 (.36) °~ Tasks 9 and 10 contain word initial phonemes available at the pre-treatment level; task 11 contains unavailable phonemes. Table 3.3. - Results of supply and judgment tasks Task SI S2 S3 Rhyme Supply 19/20 20/20 20/20 (.95) (1.0) (1.0) Rhyme Judgment 19/20 11/20 20/20 (.95) (.55) (1.0) Alliteration Supply 18/20 15/20 20/20 (.90) (.75) (1.0) Alliteration Judgment 19/20 14/20 19/20 (.95) (.7) (.95) 37 There was no difference between tasks that varied with the vowel quality in the distractor item, in either the rhyme or alliteration condition. In a comparison of phonological contrasts available and unavailable at the pre-treatment level, there was no difference for SI. on either the rhyme tasks or the alliteration tasks. S2 performed similarly in both the "available" and "unavailable" conditions on the rhyme oddity tasks, with 67% correct for each condition. However, in the alliteration oddity tasks there was a notable difference, with 67.5% correct in the "available" contrast condition (tasks 9 and 10), and 36% correct in the "unavailable" condition (task 11). The results were quite consistent over both trials of task 11, with 7 of the 11 items receiving the same answer in both trials (See table 6). Contrast between word-initial /r/ and /w/ is still problematic for S2, and may represent an effect of his current phonology as well as, or instead of, early phonology. The errors occurring on voicing contrasts may also result from passing the stimuli through his own system. These contrasts account for 5/14 errors made on this task. Based on previous research (see Chapter 1.), the predicted order of difficulty for the tasks presented in this study would be, from easiest to most difficult, rhyme judgment, rhyme supply,  alliteration judgment, alliteration supply, rhyme oddity (with pictures), rhyme oddity (without pictures), and alliteration oddity. SI and S2 both found the rhyme oddity task without picture support the most difficult, but did find the alliteration oddity task amongst the most difficult. In spite of the overall difference in scores, the rank order of tasks for both S i and S2 was similar. 38 Table 3.4. -- Results of Task 11, for comparing for available and unavailable word-initial phones Available Unavailable b/h 3/4 r/w 0/2 w/t 3/4 f/h 0/2 m/t 2/4 t/d 1/2 k/d 1/4 s/h 0/2 P/t 3/4 h/f 2/2 g/w 2/4 p/b 0/2 h/n 3/4 1/2 w/b 4/4 2/2 k/h 3/4 ©/h 0/2 w/m 3/4 t/t$ 1/2 1/0 1/2 39 Table 3.5. ~ Rank order of scores, ranked from highest (1) to lowest (7) Task SI S2 Rhyme oddity, 4 4 (picture support) Rhyme oddity 6 7 (w/o pictures) 6 Rhyme judgment 2 Rhyme supply 1 1 Alliteration oddity Alliteration supply Alliteration judgment 40 RESULTS: PART 2 - TEST BEHAVIOURS AND STRATEGIES The small sample used in this study allowed detailed observations of the different strategies and test behaviours displayed by the subjects, and of how some of those strategies were modified as the subjects became more familiar with the task requirements. Because of the relevance of these behaviours to the performance of the subjects, a detailed description of them is given. Rhyme Oddity Tasks On the first day of testing, SI completed four rhyme oddity tasks with picture support. At that time, he combined strategies of audible self-vocalisation and comparison to arrive at the solution. He matched two items at a time until he was satisfied he had found a matching pair and then presented the non-matching item as the solution. For example, in task 4a, item 2, he self-vocalised the following: day case: day face; case face and then gave day as the answer. This strategy was quite successful, producing an accurate answer 90% of the time. By the third session, however, he had abandoned at least the overt self-vocalisation aspect of this strategy for rhyme oddity tasks. When confronted with the rhyme oddity task with no picture support (task 5a), the above strategy was no longer useful to him, possibly because he was unable to retain the items in short term memory long enough to perform the necessary manipulations. He appeared to adopt the strategy of responding with the item he could remember. Six of the first seven responses 41 consisted of the final item in the array. He presumably had enough information from previous trials to suspect that this strategy was unlikely to produce consistendy correct results, and thus apparently altered his strategy so that he retained only the first item for the next two arrays. The last item, which he guessed correcdy, was the medial item. Due to the randomisation procedure, these strategies resulted in an apparently significant score of 7/10. However, this score cannot be considered reliable because of the pattern of response described. On the second trial of this task (task 5b), given on a different day, he adopted another strategy. He first used self-vocalisations consisting of the final sounds in each item, and then gave his response as a number: e.g., "the second one." When asked for the word, he was unable to produce it, or he gave a response that was apparently a blend of two possible answers. For example, in the array mass rag sag he gave the response "rass." He maintained this strategy, the apparent purpose of which was to decrease memory load, for the first seven items. This resulted in five correct responses. He then reverted to the strategy of responding with the final item, as he had done on trial 1, for a final score of 6/10, which approaches significance. While this was lower than 7/10 scored on the first trial, the result seems more reliable. Subject 2 also used audible self-vocalisations for the rhyme oddity tasks, although they were apparently less well organised than STs. S2's repetition of the stimulus items also resulted in distortion of the items where his own current phonology deviated from the recorded form. Because his responses lacked consistency, the examiner reminded S2 of the task periodically, or asked him to identify or describe the task. Although he was able to do this, some responses indicated that he was not attending to the task. For example, after making an incorrect response 42 on item 2 in task 4 (pay gate date), he was asked which two rhymed. He answered correctly, gate and date, but persisted in saying that gate was the odd word. Another example occurred in the third testing session, when one might have expected task familiarity to be established. He followed up a correct response by saying: "I knew that was the right one because it's stuck on differently." (The picture was off-centre). In task 5, presented without pictures, he retained his strategy of vocalising the items. This appeared to indicate that he was either not able to correctly remember all the items, or was "buying" processing time. For example, where E gave the stimulus pain rain hay he repeated the items as pain ray hay, and gave pain as the response. Similarly, he recalled maze lace ways as maze lace wace, and then gave maze as the (wrong) response. For both trials of this task he exhibited a preference (5/10) for the final item. Subject 3. the control subject, did not audibly vocalise any of the items in the rhyme oddity tasks, and had no difficulty in scoring 100%. His behaviour is worth mentioning in that he seemed to require much less attention to the task. He often talked while the stimulus items were being given, in spite of reminders to listen carefully, but this did not interfere with his performance. He also had a tendency to stamp his feet and tap the table. His correct responses were generally prefaced with information about the rhyme: "x and z rhyme, y is the odd one." In task 5 (rhyme oddity without picture support) he showed a decrease in extraneous activity, suggesting an increase in attention, but did not show any difficulty recalling the items. His failure on one item in the first trial may have been due to the lexical items themselves (roof  truth youth), which were of lower frequency than most of the stimuli. His response was correct on the second trial. 43 Although the response time was not measured in these tasks, later analysis of sessions revealed that S3 was able to complete tasks at a faster rate than either SI or S2, in spite of the fact that S i approached him in accuracy on many of the tasks. Rhyme Supply As mentioned in chapter 2, this task was to provide an item that rhymed with the stimulus item. Nonsense words were accepted for purposes of this test. On sixteen out of twenty items S i responded by replacing the initial sound with a bilabial consonant, usually [b]. On the remaining four items he used a coronal stop consonant ([t] or [d]). This resulted in a high percentage of responses consisting of lexical items, but the consistency of the bilabial response indicates that finding a lexical item was not a priority. There were other indications of this. For example, after giving his response to the word "king" as "bing," he said: "I know, the Lady Byng Trophy," indicating that a lexical connection had been made only after the response was given. 52 adopted a similar strategy, but replaced initial sounds with [d]. He did this in seventeen of the twenty items. He adopted a bilabial stop for the remaining three items. 53 responded to all the stimuli with an unambiguous lexical item, e.g., liquor as a rhyme for wicker. Rhyme Judgment SI and S3 had no observable strategies for this task and their performance indicated that they were efficient at making a judgment between two items. S2 utilised vocalisation for the majority of the items, providing evidence of hearing the words correctly. His score of 11/20, however, was only minimally above chance. 44 Alliteration oddity SI and S2 both used audible vocalisations to assist them in this task. S2, however, tended to vocalise the initial sound + [a], e.g., /k/ went to [k3>]. As in the rhyming tasks, the current phonology of S2 sometimes resulted in an incorrect form of the sound. For example, he made no contrast between /r/ and /w/, producing [w] in all cases. He also devoiced several stops. S3 did not employ any observable strategies, but there was a decrease in his extraneous behaviours during these tasks. Alliteration supply SI appeared to use knowledge of the orthography to reach a solution. After each answer E asked him what sound he heard at the beginning of the words (stimulus and response). He responded with a letter name rather than a phone. Also, the two responses he gave incorrectly are indicative of an orthographic strategy. For example, he said sugar has the same first sound as salt. When asked, he identified the initial sound as "s." Similarly, he produced eagle for the stimulus indian, and identified the first sound as "e." S2, who has less developed literacy skills, used phones rather than letters to reach a solution. He vocalised the phones without prompting, as part of his overall strategy. He repeated the initial phone, or his own form of it, several times for most items. However, his responses also suggested some recourse to orthography. Several items resulted in the following observations (the alliteration supply task was administered during session 1 for S2): (1) He produced the correct phoneme [dz] for just, but, after repetition it was reduced to [d]. This in turn led him to the letter "d," and his final pronouncement that the word "began with "e"." He gave the word egg as a response. 45 (2) He identified /k/ as the initial phone in catch, but gave the response house. His mother said that he often confused "k" and "h" and was under the impression that a letter "kaitch" existed. (3) He identified "s" as the initial letter of school, but his response was given as "fun, sun, snake." (4) Voicing confusions in his own productions led to /g/ becoming [k], and /t/ becoming [d] in the responses. Unlike the results for the rhyme supply task, proper lexical items were supplied as responses, with one exception for S2, who supplied the word lostv as a response the stimulus word lost. All the subjects apparently responded to the first phone, rather than the onset, so that words with clusters were given in response to stimuli containing a single consonant onset, and vice versa. Alliteration Judgment This task was the only alliteration task during which S3 gave an incorrect response, saying that hope and have did not begin with the same sound. When questioned about this he said: "hope begins with "ho" and have begins with "ha." S i and S2 also gave an incorrect response to this item. As on the alliteration supply task, the subjects were apparently attending to the first phone, rather than the onset. Although there was only one item for each subject that contrasted a single onset with a cluster, all the subjects agreed that these words began with the same sound. 46 CHAPTER 4 DISCUSSION The goals of this study were to investigate the possible effects of a phon ological disorder on metaphonological development. A controlled set of rhyme and alliteration tasks was developed to test two subjects with a previous phonological disorder. Limitations of the Study Before reaching a conclusion about the effect of the variables under consideration, it is necessary to enter into some discussion of other factors, besides test behaviours, that may impinge on the results. Stimuli The stimuli for this study were derived from a word frequency list compiled by Kucera and Francis (1982), primarily from the first 2000 words. This word list, however, is compiled from adult sources, with the result that many words which might be considered familiar to children, such as common animal names, are omitted, or are very low on the list. Conversely, items that are beyond the normal vocabulary of a 6-year-old child may have a high frequency in adult material. Although all the items were presented to the subjects prior to testing (see above), this situation demonstrates the need for a reliable word frequency list reflecting vocabulary development and exposure over the early school years. 47 Response Strategies The response strategies and testing behaviours described above direcdy affect the reliability of the results. The strategy of selecting the final item, as S i did in task 5, produced a significant, but obviously invalid, score. Similarly, the disclosure by S2 that he was paying attention to physical properties of the stimuli, in spite of frequent repetitions of the task procedure, also jeopardises test reliability. However, it is not sufficient to say that, because of this, any conclusions concerning metaphonological ability derived from performance scores are questionable. There are two possibilities: (1) the subjects were experiencing difficulty with the phonological awareness tasks which led them to adopt faulty strategies, or (2) the subjects were able to perform the necessary phonological manipulations and the adoption of poor strategies arose from a deficit in some other contributing ability such as memory, attention and/or, as hypothesised, from the interference of residual, immature phonological processes. (See below for further discussion of this topic.) Homogeneity of Tasks A key area of concern is the range of skills tapped by these tasks. Several of the tasks required the child to select the odd word from an array. Evidence from the subjects' observable test behaviour suggests that they arrive at this solution by first determining which words have the same pattern. This has implications for the type of analysis the subject is applying to the task. In the introduction the different levels of analysis, e.g., phonemic or segmental, were discussed. Phonemic analysis requires an ability to segment the speech stream into phonemic/phonetic units, a skill that is linked to emerging literacy. A segmental analysis requires separation of rime and onset only. For some oddity selection tasks there is a degree of ambiguity. 48 For example, in the array book cook gas the rimes differ maximally (vowel and consonant). Thus the child can arrive at the correct solution by segmentation at the rime/onset division. In the array, lake wake race, the rimes differ minimally (final consonant only). Thus a child could be performing a subsyllabic, segmental analysis. If, however, the approach of the child is to select the rhyming items first, as suggested by the task behaviour of SI and S3, the odd item may be ascertained by default, and a syllabic analysis is sufficient. The fact that no subjects showed an effect correlating with similar vowels in the distractor items, suggests that the levels of analysis were comparable for both types of array. In addition, the fact that S3, who showed complete mastery of this task, apparently matched items before selecting the odd one, implies that this is the strategy of choice, and that rhyme "oddity" tasks may be misnamed. Yopp (1987) found that rhyme tasks were easier to perform than other metaphonological tasks, and a factor analysis of various tasks led her to think that they may not tap the same skills as tasks requiring phonemic segmentation. However, the limited number of rhyming tasks used in her study may be insufficient to make general conclusions about rhyming tasks. If rhyming tasks and phonemic segmentation tasks do in fact represent different skill areas, then the question arises as to whether both skills demonstrate metaphonological knowledge. Yopp analyzed each task according to the cognitive processes) she believed were involved in the task. The most pertinent difference in the cognitive processes outlined in her analysis is that rhyme does not involve phonemic segmentation, only "sound," or "word," comparison. This implies that such a comparison is possible in the absence of segmentation, an implication which is in contention with the belief of Bowey and Patel (1988) that segmentation is implicit in rhyme oddity tasks. The evidence from the present study suggests that children solve rhyme oddity tasks by making 49 comparative judgments, but at a supra-segmental (i.e., syllabic) level. The fact that children do not generally provide a similar word in response to a rhyme supply task, e.g., do not reply "bat" to the stimulus bat, suggests they have some knowledge that rhyme requires some kind of segmentation and substitution. The employment of comparative processes suggests they will use the larger units (i.e., onset/rime) afforded by knowledge of rhyme to analyze the speech stream, if possible. The conclusion that children are unable to perform a phonemic analysis is not warranted, however. SI's strategy of vocalising the word final phones to reduce memory load in task 5 (see chapter 3), suggests that he had the ability to segment a word in this manner, as presumably does S3. Both SI and S2 experienced more difficulty with the alliteration oddity task, which required a comparison of first phones. Again, this task should be solvable by segmenting the rime from the onset, and then comparing rimes, comparing onsets. Some evidence from this study, however, suggests that children may not be as sensitive to onset as they are to rime. Two items in the alliteration supply task contained onset clusters, but all of the subjects responded with a word beginning with the initial single phone. For example, the word straw elicited the response silent in SL, sun in S2, and soot in S3. Kiparsky (1979), however, suggests that /s/ may be extrametrical in some clusters (sp, sk, st), and therefore "outside" the onset, which would provide an alternate explanation for single consonant responses. Better evidence may be that they sometimes responded to words with a single consonant onset with words beginning with clusters. For example, S2 gave the word frog in response to find, and S3 gave the name Greg in response to get. 50 Yopp (1987) included a "word to word" matching task in her study, that was similar to the alliteration judgment task administered in the present study. Her cognitive process analysis of the task included the ability to "identify a sound in a given position" and to "isolate a given sound." Again, these skills were considered to operate in the absence of sound segmentation. It may be, however, that in order to "isolate" an initial sound, the child must have (1) some knowledge that the speech stream is segmentable, and (2) awareness of the appropriate place to execute the segmentation. Evidence of this knowledge can be found in S3's response to the stimulus hope have (see chapter 3). His failure to identify these words as having the same onset may have arisen from faulty identification of the initial phone, but his response that "hope begins with "ho" and have begins with "ha"," is unmistakable evidence of a conscious, phonological segmentation process. The finding that several metaphonological tasks correlate highly with each other is useful for further research as it gives some indication that researchers are using valid measurements of phonological awareness. However, to conclude, as Yopp suggests, that those tasks which do not correlate well, such as rhyming tasks, may utilise different skill areas, is not satisfactory. If phonological awareness is not necessary for rhyming tasks, then what skills are required? It is not sufficient to say that rhyme involves the comparison of two sounds, or words, while failing to define the parameters of comparison. The question becomes: does rhyming involve a larger unit of analysis, (VC or VCC) and whose difference in unit size is sufficient to account for the lack of correlation with other tasks, or, does rhyming depend on some other cognitive skill, as yet undefined. Because metaphonological skill is defined by the ability to perform on certain tasks thought to require metaphonological awareness, it may be more useful to abandon the term 51 "metaphonology." Instead referral might be made to the ability to perform certain tasks that, while being individually defined and analyzed, have some commonality. For instance, all these tasks require phonological manipulation, and all have been hypothesised as having some predictive relationship with literacy. Further abstraction as connoted by "metaphonology" is premature and possibly invalid. Memory While the relationship between phonological disorders, metaphonological delay, and short-term, or working memory deficits is an area that requires considerable research, memory appeared to be a factor in this study. The subjects who had a disorder both experienced difficulty recalling the stimulus items, while the typically developing subject showed only a slight increase in attentive behaviour. This lends some support to the hypotheses of Brady, Shankweiler and Mann (1983) that a deficit in memory for phonetic units may result from an earlier failure to internalise sufficiently characteristics of the phonetic pattern. Bryant and Bradley (1983) considered that processing of three words was too easy for 5-year-old subjects, and consequentiy gave them lists of four words without picture support. They do not provide correlations between memory (measured over 30 memory trials) and task performance, but they do note a correlation between initial memory scores and final reading scores of 0.40 for four-year-old subjects and 0.22 for five-year-olds. However, their memory trials only required direct recall of items, without processing of phonological information. Our results suggest that memory load is a factor in performance of these tasks. Effects of Hypothesised Variables As noted in the chapter 3, there was evidence for a general effect for a phonological ( 52 disorder, indicated by the overall difference in scores between SI and S2 and those of S3. The results suggest some effect of a child's early sound system on rhyme and alliteration tasks, and also an effect of current, immature phonology. These effects are demonstrated most clearly in the results of S2, particularly in the alliteration oddity tasks, 11a and lib. Although his present speech is not marked by an absence of voicing contrasts, word initially or finally, word-initial voicing contrasts caused difficulties for him in these tasks. Similarly, the substitution of [h] for other fricatives, present in his early speech but not his current speech, is clearly reflected in his responses. The confusion between /r/ and /w/, still a factor in his current speech, is also evident. Some of the errors are attributable to S2's response strategy of self-vocalisation. Stimuli were passed through his own phonological system, and his responses were then consistent with his own productions. He apparently was unaware of, or disregarded, any differences between the speaker's production of the stimuli and his own. However, this does not explain the confusion between /h/, and /s/ or /f/, which suggest that apparently suppressed phonological processes are still operative at some level in the system. Residual effects are also discernable in the results of SL I have discussed several factors that may have influenced the results, beyond those hypothesised. These include test behaviour, memory, attention, and the stimulus list itself. However, it is not clear that the extraneous factors completely explain all the errors committed by Sl_, although they appear quite random. He scored an average of 9/10 on the rhyme oddity tasks, scoring 10/10 in only one instance. S3 demonstrates that complete facility with this task means that virtually no errors will be made, and that errors that are made will be readily explainable. I have noted above that the most salient difference between these two subjects is found in response time and strategy, but some 53 explanation for the slight, though consistent, difference in scores is necessary. The errors made by SJ_ appear random in nature, rarely being repeated. It is possible, however, to detect some patterns. Nine errors (11% of stimuli) were made on the rhyme oddity tasks (with picture support). Four (44%) of the errors involved V/VC rime contrasts, e.g.. no top show (V, VC, V). (See appendix C, list 1.) As this type of contrast constituted only 20% of the stimuli, random errors should only result in 1.8 errors for these stimulus items, as opposed to 4. On the alliteration oddity tasks, a different pattern is discernible. One uncontrolled variable for stimuli construction was the word-final phone. To find sufficient, high frequency words, with this added constraint, was too difficult a task. This resulted in some of the items in an array having a word-final phone that matched the word-initial phone, e.g., bone knife knock. (See appendix C, list 2.) In some cases (two, with one repetition of each), this was deliberately contrived, in order to assess any residual effects of the word-initial labial assimilation process present in Si's early speech, e.g., pipe pay tape. Seven of the thirteen errors (53%) made in these tasks were committed on this type of array, which again constituted only about 20% of the possible arrays. Random responses would only account for 2.6 errors on this type of array, suggesting that the assimilation process may not be entirely suppressed, or that he was confused by the fact all words included the same phoneme. In any case, he wnot attending to word position. This reflects syllable shape errors in his early phonology (assimilation and metathesis), or memory recency effects for order in tasks. A further point to consider is the three different approaches to the rhyme supply task that the subjects demonstrated. As noted above, S3 responded with real lexical items; S i responded with real lexical items, but employed [b] for the word initial phoneme in 80% of the answers; 54 S2 substituted a [d] for the word initial phoneme in 85% of his responses, all of which were nonsense words. One question that arises from this is whether or not JS1 produced real words through some process of selection, or whether the real words were coincidental. There is some evidence, e.g., the Lady Bvng response mentioned above, that he used the phoneme substitution technique and then, by some internal process, checked the word against his lexicon. If this is indeed the case, then the three subjects may represent three stages in the acquisition of rhyme supply skills, with S2 demonstrating an early level, S3, mastery, and SI some intermediate level. Lundberg, Olofsson and Wall (1980) mentioned that some of the children in their study used nonsense words in their responses to this type of task, but they do not say whether the subjects used a phoneme substitution strategy, whether they used a combination of nonsense and real words, nor do they give any description of the subjects' phonological development. Stanovich, Cranmer and Cunningham (1984) gave tests of both rhyme supply and initial phoneme substitution to their kindergarten-age subjects, and found the scores, and their correlation with a reading measure, almost identical. (Mean score for rhyme supply: 8.57; for initial phoneme substitution: 8.63.) Again, however, there is no information on the presence or distribution of nonsense words in either task. Thus, their claim that the subjects had reached ceiling level on these tasks (based on the fact that 16 of their 49 subjects scored 10 out of 10 on the phoneme substitution task) may not be justified when the mean scores of all their subjects are compared to the performance of S3. Areas for Further Research The results of this, and other studies, have provided some insights into the development of the ability to perform tasks that require phonological manipulation. However, many areas for 55 further research remain. It appears from this study that small differences in results may indicate significant underlying problems. Other methods of measuring the differences between subjects, such as response times, may. be useful. Given the significant effect of memory on the subjects' ability to perform the tasks in the present study, the relationship between short-term memory for phonetic units, early phonological delay, ability to perform phonological manipulations, and later reading ability should be further examined. More studies that examine the performance of phonologically delayed children, as opposed to those with typical development, need to be done to explore the link between an early disorder and later development. However, there is a need for longitudinal studies with larger subject samples, from pre-school age to late primary, which will clarify the developmental process underlying metaphonological skills. There is a need for further task analyses. Little has been done in this area, which has led to some possible errors in researchers' assumptions concerning the type of analysis subjects are performing. A further breakdown of all tasks is required, in order to more closely define the developmental process. Several variations on one task were performed in this study, but others could have been included without redundancy. For example, a more controlled assignment of word-initial clusters might provide more insights into how children perceive onsets; requiring subjects to produce real, as opposed to nonsense words in the supply task might supply information as to whether this ability was present, but ignored in favour of an easier strategy. A more thorough analysis of the tasks should be followed by the development of norms, in order to improve the reliability and validity of tasks presently being administered. 56 CONCLUSION In this paper I have studied the effects of early phonological delay on later metaphonological development. In spite of the small sample, the results of this study seem to indicate a relationship between an early phonological disorder and a later delay in ability to complete tasks that test phonological awareness. The performance of S3 clearly demonstrated a ceiling effect in this skill area, and provided a useful illustration of optimal achievement. SL, who no longer exhibits signs of a phonological disorder, was able to perform at a level that indicated ability to do the task, but without demonstrating the same facility as SL Careful analysis of his responses suggested they were influenced by early, and currently non-overt, phonological processes. S2, whose speech still exhibits immature forms, was able to achieve some success, but obviously still experiences difficulty in this area. His responses also indicate some residual effect of early phonological delay, as well as showing an effect from his current sound system. Returning to the definition of metaphonology, several points emerged as being salient from definitions by other researchers. These included awareness and access to one's phonology (Mattingly 1972); the view that such awareness should be explicit (Mann 1986); the ability to reflect on and manipulate the sounds of the language (Tunmer at 1988), and that phonological manipulation occurs in the absence of context and meaning. Bialyostok and Ryan (1985) presented a cognitive view of metaphonological development, which perceives such awareness 57 emerging from the development of the cognitive processes of control and access. Metaphonological awareness is measured by ability to perform certain tasks that purportedly tap the above-mentioned skills. Yopp (1987) and Stanovich et al. (1984) have compared various tasks and found high correlations between several of them. The lack of adequate task analysis makes it difficult to operationally define "metaphonological ability." It may be preferable, especially when exatnining the links between literacy and phonological awareness, to merely state whether or not the child was able to complete successfully a certain task. As a caveat to this, success should be carefully defined. The results for both of the phonologically delayed subjects demonstrates that performing at a statistically significant level does not necessarily indicate the absence of problems; nor does failure on a task necessarily indicate an inability to consciously manipulate phonemes, but may result from a residual (or current) phonological deficit. While evidence exists for task comparability, data from this study suggests that tasks may not tap the skill area that is intended. In addition, several factors, such as memory, attention and phonological impairment, apparendy influence outcome. Results of this study, therefore, suggest that the concept of metaphonological ability is poorly defined and inadequately tested. 58 REFERENCES Bernhardt, B. (1990). Non-linear Phonological Intervention. Unpublished doctoral dissertation, University of British Columbia. Bialyostok, Ellen, & Ryan, Ellen (1985). Towards a definition of metalinguistic skill. Merrill- Palmer Quarterly, 31, 229-251. Bowey, Judith A. & Patel, Rinu K. (1988). Metalinguistic ability and early reading achievement. Applied Psvcholinguistics, 9, 367-383. Bowey, Judith A. (1990). On Rhyme, language, and children's reading. Applied  Psvcholinguistics, 11, 439-448. Bowey, Judith A. & Francis, J. (1991). Phonological analysis as a function of age and exposure to reading instruction. Applied Psycholonguistics, 12, 91-121. Brady, S., Shankweiler, D. & Mann, V. Speech perception and memory coding in Relation to Reading Ability. Journal of Experimental Child Psychology. 35, 345-367 (1983). Bradley, L. & Bryant, P. (1978). Difficulties in auditory organisation as a possible cause of reading backwardness. Nature, 271, 746-747. Bryant, P.E., Bradley, L., MacLean, M. & Grassland, J. (1989). Nursery rhymes, phonological skills and reading. Journal of Child Langauge, 16, 407-428. Bryant, P.E., MacLean, M. & Bradley, L. (1990). Comments "On Rhyme, language and children's reading". Applied Psvcholinguistics, 11,449-450. Bradley, I. & Bryant, P.E. (1983) Categorising sounds and learning to read-a causal connection. Nature, 301 : 419-421. Bryant, Peter, MacLean, Morag & Bradley, Lynette (1990). Rhyme language and children's reading. Applied Psvcholinguistics, 11, 237-252. Fox, B., & Routh, D.K. (1976). Phonemic analysi and synthesis as word-attack skills. Journal of  Educational Psychology, 68, 70-74. 59 Gentry, J.R. (1982). An analysis of developmental spelling in GYNS AT WRK. The Reading  Teacher, 36, 192-200. Grunwell, Pamela (1986). Aspects of phonological development in later childhood. In K. Durkin (Ed.). Language development in the school years. London: Croom Helm; Cambridge, Ma., Brookline Books. Hoffman, Paul r. & Norris, Janet A.(1989). On the nature of phonological development: Evidence from normal children's spelling errors. Journal of Speech and Hearing Research. 32,787-794. Kamhi, Alan G., Carts, Hugh W., Mauer, Daria, Apel, Kenn & Gentry, Betholyn F. Phonological and spatial processing abilities in language and reading impaired children. Journal of  Speech and Hearing Disorders, 53. 316-327. Katz, Robert B., Healy, Alice F. & Shankweiler, Donald (1983) Phonetic coding and order memory in relation to reading proficiency: A comparison of short-term memory for temporal and spatial order information. Applied Psvcholinguistics. 4, 229-250. Kiparsky, P. (1979). Metrical structure assignment is cyclic. Linguistic Inquiry. 10, 421-441. Knafle, June D. (1972). Children's discrimination of rhyme. Journal of Speech and Hearing  Research. 17, 367-372. Kucera, H. & Francis, W. Nelson (1982). Frequency analysis of English usage: lexicon and  grammar. Boston: Houghton Mifflin. Lenel, Julia C. & Cantor, Joan H. (1981) Rhyme recognition and phonemic perception in young children. Journal of Psycholinguistic research. 10, 57-67. Lundberg, I., Olofsson, A. & Wall, S. (1980). Reading and spelling success predicted from phonemic awareness skills in kindergarten. Scandinavian Journal of Psychology. 21, 159-173. McReynolds, Leija V. & Thompson, Cynthia K. (1986). Flexibility of single-subject experimental designs. Part 1. Review of the basics of single-subject designs. Journal of Speech and  Hearing Disorders. 51, 194-203. Mann, Virginia A.(1986). Phonological awareness: The role of reading experience. Cognition. 10, 65-92. Olofsson, A. & Lundberg, 1.(1983). Can phonemic awareness be trained in kindergarten? Scandinavian Journal of Psychology, 24, 35-44. 60 Rohl, M. & Tunmer, W.E. (1988). Phonemic segmentation and selling acquisition. Applied  Psvcholinguistics 9, 335-350. Read, C. (1971). Preschool children's knowledge of English Phonology. Harvard educational  review. 41, 1-34. Smith, Carol L. & Tager-Flusberg, Helen (1982). Metalinguistic awareness and language development. Journal of Experimental Child Psychology. 34, 449-468. Stanovich, Keith E., Cunningham, Anne E. & Cramer, Barbara B. (1984) Assessing phonological awareness in kindergarten children: issues of task comparability. Journal of  Experimental Child Psychology. 38, 175-190. Stoel-Gammon, C. & Dunn, C. Normal and Disordered Phonology in Children. University Park Press. Baltimore (1985). Treiman, Rebecca (1985). Onsets and Rimes as units of Spoken Syllables: Evidence from children. Journal of Experimental Child Psychology. 39, 161-181. Tunmer, William E., Herriman, Michael L. & Nesdale, Andrew R. (1988). Metalinguistic abilities and beginning reading. Reading Research Quarterly 23, ppl34-158. Wagner, R.K. & Torgesen, J.K. (1987).The nature of phonological processing and its causal role in the acquisition of reading skills. Psychological Bulletin. 101, 192-212. Wells, Kathleen (1987). Scientific Issues in the Conduct of Case Studies. Journal of Child  Psychology and Psychiatry. 28, 783-790. Yopp, Hallie Kay (1987). The validity and reliability of phonemic awarenes tests. Reading  Research Quarterly. 23, 159-177. 61 APPENDIX A EARLY PHONOLOGICAL PATTERNS 62 Table 1.—SI: Summary of singleton consonant acquisition at age 3;4 Sound class Word position matches with adult target Word position nonmatches Glides --/w, j / /w/: WI, (SIWW) 1' 2 /j/: WI /j/: SIWW Nasals WI /m/: WF, (SIWW) M: (WF) Iml: SFWW /n/: (SIWW), (SFWW)3 ((A)/)) Stops [-voice] /p,t/: (WI) /p/: (SIWW), (WF2/4) /t/: (SIWW), ((WF)) /k/: (WT), WF All: SFWW Stops [+voice] /b/: WI /d,g/: (WI) /b/: (SIWW) /d/: (SIWW) /g/: SIWW All: WF, SFWW Fric/Affricates /f, v/: WI /s/: WF /f, s/: (SIWW) All others Liquids None /V: (WT):2/13, /r/: ((WF)) (1/16) All others M None ((SIWW)) Note. Reprinted by permission of the author, from Barbara Bernhardt (1990) ^^word-initial, SIWW=syllable-initial-within-word, SFWW= syllable-final-within-word, WF=word-final ^Single parentheses indicate partial establishment - 40% - 80% in the matches column, 10% - 40% in the nonmatches column. Double parentheses denote marginal establishment - 1 - 10% . 63 Table 2.--SJ2: Summary of singleton consonant acquisition at age 3;5 Sound class Word position matches with adult target Word position nonmatches Vowels Singleton vowels Diphthongs Glides - /w, j / /w/: WI.(SIWW) 1' 2 /j/: (WI) /j/: SIWW Nasals WI /m/: SIWW /n/ : (SIWW) /m,n,rj/: SFWW /m/: (WF) /n,n/: WF Stops [-voice] /p/: WF, (SFWW) AJc/: (WI) /k/ : (SIWW, SFWW, WF) /p/: (WI), (SIWW) /t/: ((SIWW, WF)) Stops [+voice] /b,g/: WI /d/: (WI) /g/: SIWW /b/: (SIWW) /d/: (SIWW) All: WF, SFWW Fric/Affricates /* / : (WI) /z/: (SIWW) /s/: (WF) A//: (WI) /J/: (WF) All others Liquids None ((SIWW N and WI A/)) / h / (WI) Note. Reprinted by permission of the author, from Barbara Bernhardt (1990) lWI=word-initial, SIWW=syllable-initial-within-word, SFWW= syllable-Final-within-word, WF=word-final 2Single parentheses indicate partial establishment ~ 40% - 80% in the matches column, 10% - 40% in the nonmatches column. Double parentheses denote marginal establishment -- 1 -10% . 64 APPENDIX B STIMULI 65 APPENDIX B Subjects 1 and 3  Rhyme Oddity Tasks Task 1 Task 2 Task 3 watch bush push fish rich wish key bun one bed head gate big miss dig ice mice cow base leg egg knight ride guide she suit root rise shoe eyes rose snow nose arm go farm pool boy toy saw fall ball feet in win wing king man young run sun hen pen sea book cook gas lake race wake no top show knees face race hose toes close kiss miss rain tooth wife knife half laugh path sit hit type bag walk talk dog lock rock date room late Task 4 Task 5 sign eye line side right wide day case face beg neck check white tie light school cool who game same pay wish dish itch low bone cone ring skin sing rain train pay shake take pace three tree sweep guys lies nice S ten less boy toys noise heat eat deep mass rag sag team sheet seat truth roof youth shape space stick pick fine nine old cold rest chest pull fun pull full dress red break lake old hole Rhyme Judgment hand band wheel eat tea seat house out green queen store board hand bat nine ride ski tea card yard 66 Rhyme Supply knock word soap king tacks wicker shop square bags right fall star cow bike trip guess tie cat Alliteration Oddity Task 9 Task 10 Task 11 man knee meet mouth ear moon feed fit thick time two fight fat ten team scarf scare car gun pick gas goal guess pair share soup shoe day maid door dear pack post teach sheet shake eye night note bone knife knock rain wash weight food boy book bear five park take case cut pen park neck pay pool fun job judge dog face eight foot four few buy kid tick key key cool meat catch ark cow team seed tooth ice tool tie mine tear tear boot tub bed pipe pay tape pick big bad half art help lake load eight Alliteration Judgment year yard small snail free five near most small meal last white word rope cheat safe be peach like last think fine reason rice eel eat with wood very able very voice join dive hope have year lime think thumb Alliteration Supply banana water get find tell school rest four high straw salt bear catch just indians market last lost week take 67 Subject 2 Rhyme Oddity Task 1 Task 2 Task 3 bun one key ice cow mice race face knees half two laugh boy pool tool war sew door roof base hoof sheet seat tea tie head bed ehe he bee sign eye line case day face close hose toes wife knife buy call saw ball law four core rough puff bus hide pie ride white tie light play sky fly book mouse cook boat goat V type room pipe rock cape tape rake dip bake boot low snow pray weigh cross stop suit cop mat weak cheek sky draw claw Task 4 lake wake race pay gate date deep beak leap pack gas sack kick rip lick slow crow float mice tie he tip sit mitt cat back jack raw eye saw Task 5 pain rain hay use few moose leaf reef sea maze lace ways toe bowl coal jaw pour sore nice life rice who hoot shoot loud bowed how fee why tee Rhyme Judgment and Rhyme Supply As for SI and S3. 68 Alliteration Oddity Tasks Task 9 Task 10 Task 11 horse board big one white type teach man meet day case catch pool tooth push gun give win net nut hen wind bird work cold kid hill wash mice white house bear boy walk work time map take mud door car cut page park toy go watch game night hill note ball weight one key hurt king wheel wake miss run road won fight high five touch dust team cent head sign four foot hall pick bill buy judge dog job share shape help thief thin hit heat chief charge arm line laugh knife knee meet Alliteration Judgment land fire trade drive spin spider fire free phone far port pitch bath bad pile baby radar water potato pile dome devil south hope scare car trade tree soap shore rent rail far hungry wait round hope hockey kite hum Alliteration Supply As for SI and S2. 69 APPENDIX C SELECTED LIST OF ERRORS FOR SI 70 List 1 saw fall ball watch bush push cook gas book young run sun no show top team sheet seat light tie white cone low bone APPENDIX C List 2 knock bone knife key tick kid  boot tub bed pipe pay tape man knee meet fat ten team knock bone knife day maid door 71 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items