Phonological Representations in Deaf Children: Rethinking the ...

Empirical Articles

Phonological Representations in Deaf Children: Rethinking the ‘‘Functional Equivalence’’ Hypothesis Lynn McQuarrie University of Alberta Rauno Parrila University of Alberta

The sources of knowledge that individuals use to make similarity judgments between words are thought to tap underlying phonological representations. We examined the effects of perceptual similarity between stimuli on deaf childrens’ ability to make judgments about the phonological similarity between words at 3 levels of linguistic structure (syllable, rhyme, and phoneme). Manipulation of stimulus contrasts (acoustic, visual/orthographic, tactile/motoric) allowed a finer-grained estimate of the sources of knowledge that deaf individuals use to make similarity judgments between words. The results showed that the ability to make syllable-, rhyme-, and phoneme-level judgments was not tied to ‘‘phonological’’ facilitation when these conditions are contrasted. These findings are inconsistent with long-held assumptions of ‘‘functional’’ equivalence between ‘‘heard’’ and ‘‘seen’’ speech in the development of phonological representations in deaf learners. We argue that previous studies reporting evidence for phonological effects in similarity judgments have failed to sufficiently control for alternative sources of sensory information, namely, visual and tactile/motoric.

The scientific and educational communities have long speculated that visible speech (speechreading) can develop ‘‘functional’’ phonological representations of speech in individuals for whom auditory speech experience is absent from birth. Although it is acknowledged that precluded auditory access to spoken language limits deaf children’s abilities to learn spoken language phonology through hearing, it is widely speculated that other sensory processes (mainly vision) Grateful appreciation is extended to the participating children, parents, and school staff who ultimately made this study possible. No conflicts of interest were reported. Correspondence should be sent to Lynn McQuarrie, Department of Educational Psychology, 6-102 Education North, University of Alberta, Edmonton, AB, Canada T6G2G5 (e-mail: [email protected]).

can, to varying degrees, compensate or substitute for inaccessible acoustic evidence in developing an internal representation of spoken language (Alegria, 1998; Campbell, 1987, 1997; Dodd, 1987; Hanson, 1982, 1991; Leybaert, 1993, 1998) and, in turn, a ‘‘functional phonology’’ that will then support reading acquisition (for a review, see Perfetti & Sandak, 2000). In this paper, we will call this position the functional equivalence hypothesis. In what follows, we will first briefly review the functional equivalence hypothesis, then examine existing evidence for it, and finally, present the current study aimed directly at examining the nature of phonological representations in deaf children. Throughout this paper, deaf children are those with a congenital or early-acquired severe to profound hearing loss (e.g., greater than 75 dB in the better ear) that precludes auditory perception of conversational speech. For these children, irrespective of communication method, access to the ‘‘continuous phoneme stream’’ of a spoken language is mediated through visual perception.

The Functional Equivalence Hypothesis The functional equivalence hypothesis claims that deaf children’s phonological development is qualitatively similar, albeit quantitatively delayed, in comparison to hearing children (see Paul, 2001, for a review). The central claim of the functional equivalence hypothesis posits that visible speech information (seen articulatory gesture) extracted from the speech signal by the deaf learner is interpreted as a phonologically

Ó The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

doi:10.1093/deafed/enn025 Advance Access publication on July 16, 2008

138

Journal of Deaf Studies and Deaf Education 14:2 Spring 2009

plausible signal by the brain (Campbell, 1987; Dodd, 1976; Dodd & Hermelin, 1977). This claim is based on theories of speech perception that propose that articulatory gestures (vocal tract movements) are the primitives or objects of speech perception (e.g., Fowler, 1986; Liberman & Mattingly, 1985). These theories posit that phonetic information derived from both auditory and visual inputs map into the same motor representation of vocal tract gestures. Thus, it is hypothesized that a common abstract phonological code underlies the phonological representations established in long-term memory irrespective of the modality (auditory or visual) through which they are activated. Indeed, Campbell (1990) proposes that whereas auditory and visual speech information may differ with respect to the phonetic information they afford, they do not differ in the phonological representation they activate. Visual speech then is seen simply as a degraded or informationally poorer phonetic form of the auditory-visual speech available to hearing children (for a review, see Leybaert, 1993). On this basis, it has been further suggested that with the help of the visual information acquired through speechreading (Campbell, 1987; Dodd, 1976; Dodd & Hermelin, 1977) and the articulatory feel of words that comes through intensive speech training (Marschark & Harris, 1996), deaf children can develop phonological representations of words. Finger spelling (Campbell, Burden, & Wright, 1992), learning to write (Hanson, 1989), and extended experience with words in print (Hanson & Fowler, 1987) are proposed as additional sources of information that can help deaf individuals to develop awareness of the phonological structure of words. Although it is generally agreed that no one source of information alone is sufficient, it is argued that in combination these sources of information contribute to developing the phonological representations that underpin the coding of words in the mental lexicon for deaf individuals (see review in Perfetti & Sandak, 2000). Difficulties in reading are then seen as a consequence of delays in learning and difficulties in accessing abstract representations for speech sounds (e.g., Alegria, 2004; Hayes & Arnold, 1992; Paul, 1998) and not as a result of underlying differences in the nature of the representations themselves.

Current Evidence for the Functional Equivalence Hypothesis Experimental evidence on the representation of spoken language phonological structure by congenitally deaf individuals is surprisingly scarce given that the functional equivalence hypothesis has been the central assumption guiding educational methods for deaf children throughout the past century. Consequently, and in sharp contrast to current understanding of how phonological representations are progressively elaborated by children with intact hearing (see discussion in Werker & Yeung, 2005), our understanding of both the development and the level of specification of deaf children’s underlying representations is severely limited. Such a gap in our knowledge would appear to constitute the weakest link in determining the extent to which speech perception ‘‘in the absence of audition’’ may result in similarities and/or differences in the way that speech sounds are represented or processed between deaf and hearing individuals. Traditionally, children’s phonological representations have been examined through administration of measures testing their phonological awareness skills (Swan & Goswami, 1997). Previous studies of phonological awareness in deaf children have produced inconsistent results with some studies reporting ‘‘phonological effects’’ whereas others have found no such evidence (for a review, see Perfetti & Sandak, 2000). For the most part, studies reporting negative findings come from investigations of phonological awareness in young deaf readers (e.g., Izzo, 2002; Miller, 1997), whereas studies reporting positive findings come from investigations of older or skilled deaf readers (e.g., Hanson & Fowler, 1987; Hanson & McGarr, 1989). These observations are often interpreted as supporting developmental delay rather than deficit assumptions and a connection between higher levels of reading achievement and access to spoken language phonology (see review in Paul, 2001). A careful consideration of the past studies, however, offers several alternative explanations of why the behavioral evidence of deaf individuals’ phonological awareness appears mixed. First, although the studies have used a wide variety of different phonological awareness tasks, they have focused on a limited range

Phonological Representations in Deaf Children 139

of phonological structures (mainly rhyme). Critically, studies investigating phonological awareness have almost always (for an exception, see Sterne & Goswami, 2000) measured only one level of phonological structure and thus cannot provide evidence regarding the extent to which the developmental continuum in deaf individuals resembles that observed in hearing individuals—moving from awareness of larger units (syllables and rimes) to smaller units (phonemes; e.g., review in Swan & Goswami, 1997). Importantly, only a few studies (Izzo, 2002; Miller, 1997) have examined phonemic awareness (fine-grained contrast sensitivity) that appears to be a critical factor in both lexical acquisition (e.g., Werker & Yeung, 2005) and reading acquisition (e.g., Fowler & Swainson, 2004; Goswami, 2000) for typically hearing children. Thus, existing data offer little insight into the extent to which the well-documented difficulties that deaf children encounter in learning to speak (see review in Marschark, 2001) and learning to read (see review in Paul, 1998) may actually arise from deficits in the accuracy and the segmental organization of the underlying representations of words in their mental lexicons. Second, positive evidence of phonological effects come largely from studies that required the participants to make a two-choice discrimination judgment between words presented in print or on picture tasks requiring a two-choice discrimination between word pairs (e.g., Campbell & Wright, 1988; Harris & Beech, 1998; Sterne & Goswami, 2000). The conclusion that deaf individuals’ successful completion of these tasks reflects phonological awareness is problematic. As Sterne and Goswami (2000) observed, deaf children are likely to use orthographic information when making phonological judgments whenever possible. Thus, when test items are presented in print and the visual word form is readily available (e.g., night–fight), there may be little need for a deaf child to access phonological representations. In this sense, print presentation makes ‘‘positive’’ findings less generalizable to phonology. Of note, the conventional practice of reporting a composite score on phonological tasks that includes performance on both orthographically similar (i.e., king–ring) and orthographically dissimilar (i.e., tree– key) test items is arguably misleading and has likely led to an overestimation of deaf individuals phonolog-

ical awareness. It is also possible that a correct judgment on a two-choice discrimination task using items that are globally orthographically different (e.g., hair and bear) is made simply by detecting local orthographic similarity at the end of words. Finally, stimuli used in two-choice discrimination tasks seldom allow ruling out alternate sensory accounts of deaf individuals’ successful performance. Similarity in sound often leads to similarity in articulatory movement and in tactile/proprioceptive stimulation. Thus, on a twochoice matching task requiring children to choose the item (e.g., rocks, shoe) that rhymes with the cue (e.g., box), one cannot be certain whether the participants’ correct response was arrived at because the chosen word (rocks) was more similar to the target in tactile and/or visual articulation pattern than was the only additional foil (shoe). Consequently, what appears to be a phonological effect may be due to uncontrolled differences in tactile or visual similarity between cue and the target. Thus, an argument for phonological effects on a two-choice task is weakened by the fact that this effect ignores the possibility that different sources of information may have been applied to the signal that produced what ‘‘looks like’’ a correct response. While both deaf and hearing individuals may judge ‘‘box and rocks’’ or ‘‘hair and bear’’ to rhyme, they may have reached this decision through very different means. Importantly then, performance above chance level on such tasks does not necessarily imply use of a phonological strategy. Finally, lack of clearly defined parameters in relation to degree of loss and age of onset of hearing loss has likely contributed to the varied and contradictory results of phonological studies with deaf participants. Children who are born deaf experience a different multisensory mix of perceptual inputs than do children who are late deafened after having developed an auditory memory of spoken language or children who have enough residual hearing to access the speech frequencies (from birth—not several years later). Critically, the pattern of phonetic similarity between words (perceptual acquisition of phonological contrasts) is generated by different sets of conditions between these learners, and thus, it is reasonable to suppose that the percept that develops might also be different in significant ways.

140


The methodological concerns pertaining to past studies raise questions about the long-standing proposal of equivalence between heard and seen speech inputs in providing deaf learners with access to spoken language phonology. Based on the empirical evidence available, the central question of whether deaf individuals have awareness of phonological structure remains unanswered. Consequently, and in relation to reading acquisition, conventional arguments that deaf children can similarly benefit from phonological reading strategies as hearing children are equally without an established empirical foundation. There is now a compelling body of evidence that indicates that hearing children who have difficulty representing spoken words in memory, and in representing written words in memory, tend to have underlying phonological representations that are underspecified and lack segmental organization (see reviews in Metsala & Walley, 1998 and Mody, 2003). A recent hypothesis related to phonological processing deficits in reading disability suggests that difficulties that manifest as spoken and/or written language deficits in children may be more centrally tied to the quality of initial encoding of the underlying lexical representations of speech than has been traditionally assumed. In this sense, quality pertains to differences in the amount of phonological information used to represent items in the lexicon (see review in Goswami, 2000; Morais, 2003; Swan & Goswami, 1997). According to this hypothesis ‘‘if the underlying representations of words are imprecisely specified, then their lexical structures will not be segmentally organized and available for inspection at any phonological level’’ (Swan & Goswami, 1997, p. 22). This argument suggests that poor performance on phonological awareness tasks may not reflect a lack of phonological analysis skills per se, but rather that retrieval strategies that are premised on fine-grained representational structure are not available because the structure has not been developed to support such analysis (Goswami, 2000; Morais, 2003; Swan & Goswami, 1997). In light of the growing body of evidence underscoring the significant role that the phonological specification of words exerts on both lexical acquisition and reading acquisition in typically hearing individuals, the question of whether deaf individuals also show sensitivity to phonological structure takes on some

importance. Specifically, equivalence arguments proposing that speech perception in the absence of audition can support the development of a functional phonology for deaf learners must be evaluated against the extent to which visible speech is similarly able to support deaf individuals in the development of a robustly specified lexical system of representation that will then support efficient written word reading strategies. The goal of this study was to investigate the sources of knowledge that deaf children employ in making a phonological judgment. We attempted to address the issue of representational structure and quality of coding by investigating (a) awareness of phonological structure at three levels of linguistic complexity (syllable, rhyme, and phoneme) and the extent to which matching judgments across the phonological awareness tasks were influenced by the perceptual status (phonological, visual-tactile (articulatory), visualorthographic) of distracter items and (b) whether age and/or reading ability would differentially affect the performance of children and adolescents with severe to profound, prelingual hearing losses on the three phonological awareness tasks. If the functional equivalence hypothesis is correct and deaf children are using a phonological strategy to make form-based similarity judgments, we should observe equally accurate performance across conditions and controlling for the tactile and visual (orthographic) status of the distracter items should have no noticeable effect on performance. Thus, we should observe similar elaboration of phonological structure to that observed for hearing children (e.g., Blachman, 1994; Caravolas, 1993, see review in Goswami, 2002) where awareness of larger phonological units (syllable and rhyme awareness) precedes awareness of smaller units (phoneme awareness) with age and/or reading experience having a reciprocal effect on refinement of phonological representations. If deaf children show differential sensitivity in their performance patterns across levels of phonological structure while still demonstrating competence in reading, the conventional argument that word reading processes are ‘‘qualitatively similar’’ for deaf and hearing learners is weakened and alternative explanations for the reading progress of deaf individuals would be warranted.


Method

Measures

Participants

Phonological awareness—perceptual similarity judgment (PA–PSJ). The test battery in the phonological awareness—perceptual similarity judgment (PA-PSJ) consisted of three parallel receptive-based tasks (picture matching-to-samples) at each level of phonological structure (syllable, rhyme, and phoneme). For each task, a test item consisted of a four-picture quadruplet made up of a cue, a phonological target response that matched the cue in phonological structure (syllable, rhyme, or phoneme), and two additional distracters. The tasks were designed to assess phonological representations in a graded manner by testing matching judgments under four comparison conditions in which the orthographic, phonological, and visual-tactile similarity of cue, target, and distracter items was manipulated. At each level of phonological structure, word sets were included where children were first asked to discriminate between words that varied along a single perceptual dimension (orthographic, visual tactile, or phonological). On the final word set in each task, the critical contrast set in terms of a functional equivalence hypothesis involved making a similarity judgment between a cue, a phonological target, a visual-tactile distracter, and an orthographic distracter. At each level of phonological structure, the position of the target phonological item was randomized across the three serial response choice positions. The picture stimuli were presented by computer. The cue picture was outlined by a box with a red frame and was centered on the computer screen. The three matched sized picture response choices were bounded by yellow frames and were aligned in a row directly underneath the cue picture. The four pictures appeared at the same time on the computer screen. Participants were required to point to the picture on the computer screen that matched the cue picture in target phonological structure (syllable length, rhyme, or phoneme) from the three alternative pictures. The examiner recorded the participant’s choice by pressing a number pad connected to the computer that allowed for accuracy and response choice data to be recorded. For each task, the dependent variable was the number of correct phonological matches made by the participant. Comparing accuracy and response choices should allow a finer-grained

Sixty students enrolled in two specialized schools for the Deaf in Western Canada were given parental permission to participate in the study. All of the students had a prelingual severe to profound hearing loss, no additional exceptionality, and used a natural sign language as their primary mode of communication in their educational placements. Of the 60 students, four students were excluded because they scored two standard deviations (SDs) below the mean on a measure of nonverbal IQ. Background screening identified two participants whose degree of hearing loss did not meet sampling requirements (.75 dB better ear average), and they were excluded from further analyses. The data on two additional students whose age of diagnosis was unreported were removed from the study because the prelingual criteria for onset could not be confirmed. The final sample consisted of 52 students (19 girls and 33 boys). The students ranged in age from 6 years 6 months to 18 years 10 months. The mean age of participants was 13 years 1 month (SD 5 3.41). Thirty-eight students (73%) were born profoundly deaf (.90 dB), and seven students (13.5%) were born with a severe loss in the 75- to 89-dB range. Seven students (13.5%) were prelingually deafened with profound hearing losses diagnosed prior to 18 months of age. Seven students had families headed by deaf parents, 45 students had families headed by hearing parents with 18 of these students having some family history of deafness. From the most recent audiological records provided by the schools, each student’s pure-tone average (PTA) for the frequencies of 500, 1K, and 2K Hz for the better ear was computed. The PTA threshold for participants in this study ranged from 75dB HL to 120 1 dB HL with a median hearing loss of 103.1 dB HL. All participants had normal or corrected to normal vision. The students were in a dual language (American Sign Language [ASL]–English) environment in their educational setting. All participants used American Sign Language as their preferred conversational language.

142


estimate of the extent to which the distinctiveness of visually perceived speech could support phonological strategies.

Syllable task. This task assessed awareness of phonological structure at the syllable level by seeing whether deaf children can discriminate phonological differences in word lengths (syllable number). The syllable judgment task consisted of 25 test items arranged in five sets of five items. The five sets increased in difficulty and were grouped as follows: 1. Congruent sets (Set 1 and Set 2): In congruent sets, the cue picture and the phonological target were closely matched in both phonological length (number of syllables) and orthographic length (number of letters). Thus, if the cue was a three-syllable word with a long spelling pattern (e.g., crocodile), the target response choice was a three-syllable word that also had a long spelling pattern (e.g., kangaroo). Similarly if the cue was a one-syllable word with a short orthographic length (e.g., box), the target response choice was also a one-syllable word with a short orthographic length (e.g., van). For both sets, the two distracters in a quadruplet were orthographically and syllabically distinct from the target. Set 1 and Set 2 were differentiated by the size of syllable discrepancy between the target response and the additional distracters. Set 1 had a larger syllable discrepancy between the target response and the two distracters (e.g., helicopter–caterpillar, gum, socks). Set 2 had a one-syllable discrepancy between the target response and the two distracters (e.g., moon–blue, teacher, mountain). These sets were designated as congruent (P 5 O) sets because word length judged on a phonological (number of syllables) basis agreed with word length judged on an orthographic (letter counting) basis. 2. Incongruent sets (Set 3 and Set 4): In incongruent sets, all words in a quadruplet shared the same number of letters to control for possible orthographic bias (counting letters) in length judgments. Thus, the cue and phonological target agreed in syllable length and in spelling length. The two additional distracters also agreed in spelling length but were distinct from the target in syllable length (e.g., an eight-letter, threesyllable cue: hospital; an eight-letter, three-syllable

phonological target: elephant, and two eight-letter, two-syllable distracters: thirteen, football). A twoand one-syllable discrepancy between target response and the two alternative response choices again differentiated Set 3 and Set 4, respectively. These sets were designated as incongruent sets (O 5 O) . It is important to note that on these sets, whereas orthographic influence (letter counting) in supporting a correct judgment is reduced, course-grained (superficial) knowledge of differences in articulatory path or mouth movement patterns is not. 3. Mixed set (Set 5): In the mixed set, the cue picture and the phonological target agreed in syllable length but differed in spelling length (e.g., cue: sandwich—phonological target: taxi, distracter items: comb, helicopter). One distracter item was similar to the cue in spelling length but not in syllable length (e.g., helicopter from the example above); the other distracter was orthographically and syllabically distinct from the cue (e.g., comb from the example above). This set was designated as mixed (P 6¼ O) because judging whether two names are the same length on a phonological basis did not agree with matching judgments made on an orthographic basis. Similarly, judging whether two names are the same length based on knowledge of mouth movement patterns would no longer be supported by course-grained knowledge. Successful performance on this set should provide the clearest case for awareness of phonological structure at the syllable level. Median word frequencies based on Kucera and Francis (1967) were matched for all words in a quadruplet. Prior to beginning the test items, 10 practice items were presented on the computer. Feedback was provided on the practice items. Students who could not complete the practice trial did not continue with the task. No feedback was provided on test item trials. As determined by student performance, Cronbach’s alpha reliability coefficient on the syllable matching task was .75. Additional examples of test items used in the syllable task are included in Appendix A.

Rhyme task. This task consisted of 40 test items. Each test item consisted of four pictures made up of a cue, a phonological rhyme target, and two additional distracters. An equal number of orthographically


similar (e.g., O1, ‘‘king–ring’’) and dissimilar (e.g., O2, ‘‘cry–pie’’) cue and rhyme targets were used in each set. The 40 test items were arranged in four sets with 10 items in each set. The sets were grouped according to distracter type. To perform the task, participants viewed a cue picture and three response choice pictures. Their task was to choose the response choice picture whose name rhymed with the cue picture. The composition of the four sets was as follows: 1. Set 1 (no pattern) consisted of five O1 and five O2 cue and rhyme targets. The two additional distracter items shared no orthographic, phonological, or tactile similarity with the cue and rhyme target. These distracters were designated as no pattern distracters (O–P–T–). Set 1 example: (O1) king–ring, blue, cheese; (O2) cry–pie, bed, dog. 2. Set 2 (tactile pattern) consisted of five O1 and five O2 cue and rhyme targets. One distracter was similar to the cue word in articulatory/tactile contact but was orthographically and acoustically distinct from the cue (e.g., kite–gun). These distracters were designated as tactile pattern distracters (O–P–T1). The second distracter was a no pattern distracter (O–P–T–). Set 2 example: (O1) night–fight, tent, blue; (O2) plate– wait, bleed, dog. 3. Set 3 (visual pattern) consisted of five O1 and five O– cue and rhyme targets. One distracter item was orthographically similar (spelled alike) but acoustically and tactilely distinct (sounded and felt different) from the cue (e.g., pour–sour). These distracters were designated as visual pattern distracters (O1P–T–). The second distracter was a no pattern distracter (O–P–T–). Set 3 example: (O1) four–pour, sour, hand; (O–) shoe–blue, toe, ring. 4. Set 4 (visual and tactile pattern) contained five O1 and five O– cue and rhyme targets. One distracter was a visual pattern distracter (O1P–T–), and the other was a tactile pattern distracter (O–P–T1). Set 4 example: (O1) kite–white, knit, gun; (O–) sour– flower, soup, zero. Successful performance on the O– set should provide the clearest case for awareness of the phonological structure of rhyme as an accurate response cannot be made based on the orthographic or tactile representation of the word.

Prior to beginning the test items, six practice items (three O1 rhymes and three O2 rhymes) were presented on the computer. Feedback was provided on the practice items. No feedback was provided on test item trials. Cronbach’s alpha reliability coefficient was .88 for the rhyme task. Additional examples of test items used in the rhyme task are included in Appendix B.

Phoneme task. This task investigated whether deaf children have awareness of phonological structure at the phoneme-level probing awareness of phonemes in word—initial, final, and medial positions. The task consisted of 48 test items. The 48 test items were arranged in sets of 16 items grouped according to phoneme position: initial, final, and medial. An equal number of graphically similar (e.g., O1, ‘‘milk– moon’’) and dissimilar (e.g., O–, ‘‘car–key’’) items were presented at each phoneme position. Consistent with the distracter type descriptions outlined for the rhyme task, an equal number of distracters varying in phonological, orthographic, or tactile similarity to the cue item were included at each phoneme position. Computer practice trials preceded test items at each of the three phoneme positions. Feedback was provided on practice items. No feedback was provided on test items. Cronbach’s alpha reliability coefficient was .87 for the phoneme-matching task. An example of the test items grouped according to condition is outlined in Appendix C.

Reading comprehension. Scores from the paragraphmeaning subtest of the Stanford Achievement Test— Special Edition for Hearing-Impaired Students (Harcourt Brace, 1996) were collected through school records when available. In addition, sentence comprehension skills were assessed using the 82-item reading comprehension subtest of the Peabody Individual Achievement Test—Revised (PIAT-R) (Markwardt, 1989). Standardized administration of the test was followed with the caveat that instructions were delivered in ASL rather than in spoken English. In this test, participants were required to read a sentence silently and point to one of four pictures that best illustrated the sentence that was just read. The examiner recorded the student’s response on an answer form.

144


Procedure Between February and April of the school year, all participants were administered a test battery of English and ASL experimental tasks as part of a larger study. The total test battery took approximately 120 minutes to administer spread over three or four sessions. The spoken language phonological matching tasks reported in this study were completed in one session lasting approximately 30–40 min. Administration of the reading comprehension measure (PIAT-R) required an additional testing session of a maximum of 30 min. All participants were tested individually in a private room free of distractions in their respective schools during school hours. All testing was administered directly (not through an interpreter) by a fluent signer, and all instructions were given in ASL and print. Following Sterne and Goswami’s (2000) administration procedures, explicit instruction and a training session preceded phonological awareness testing where understanding of the concept of syllable, rhyme, and phoneme was assured. No pictures from the experimental stimuli were used in the training session, and participants received feedback on the correctness of their response. Prior to beginning the computerized trials, familiarity with the correct English name of all experimental picture stimuli was assured through pretesting using a task-specific picture dictionary created for that purpose. Understanding of the task requirements was further reinforced through computer practice trials with feedback that preceded each phonological awareness experimental task. No feedback was provided on test items. The nameable picture stimuli for the syllable, rhyme, and phoneme tasks were presented on a Dell Latitude C840 laptop computer using DirectRT v2004 precision timing software (Jarvis, 2004). Students sat facing the laptop’s 15-inch SXGA viewable screen with the examiner sitting to the right of the student. The experimental procedures for the task were outlined. Any questions the participant had were answered at this time. Results The data for each task displayed relatively normal distributions, and all analyses were performed with

raw scores. Mean accuracy scores for the syllable, rhyme, and phoneme tasks are given in Table 1. Inspection of the table shows that composite (overall task) mean accuracy scores were significantly above chance for all three tasks, t(51) 5 10.20, p , .001; t(51) 5 8.44, p , .001; t(51) 5 5.75, p , .001; respectively. Similarly, across all three tasks the data is consistent with the prediction that students would perform better overall on the graphically similar (O1) word types where a successful match can be made on the basis of orthographic knowledge. As well, across all three levels of linguistic structure, performance scores were above chance on several graphically dissimilar (O2) word types but never on the critical contrast sets: syllable Set 5 (P 6¼ O): t(51) 5 24.06, p 5 .000; rhyme Set 4 (O2): t(51) 5 22.78, p 5 .008; and phoneme Set 4 (O2): t(51) 5 24.87, p 5 .000. Analysis of the Effect of Perceptual Similarity on Phonological Judgment To determine if performance on the three tasks varied as a function of stimulus similarity (phonological, tactile, and visual/orthographic) of target words to distracter items, performance accuracy was compared across conditions using repeated measures analysis of variance (ANOVA). Separate analyses were computed for each task. An analysis of the response choice pattern of the students was also conducted to provide insight into the strategies deaf children were using to make syllable, rhyme, and phoneme judgments. Syllable awareness. At the syllable level a one-way repeated measures ANOVA was calculated with condition (with five levels: Sets 1–5) as the within-subject factor. The sphericity assumption was not met so the Greenhouse–Geisser correction was applied. Results showed a significant main effect of condition, F(3.07, 156.68) 5 74.04, p , .001, g2 5 .59. Post hoc comparisons among the five sets (performed at .05 significance level using the Bonferroni adjustment for multiple comparisons) indicated that mean performance accuracy for the two congruent (P 5 O, Sets 1 and 2) sets was significantly better than mean performance accuracy on the incongruent (O 5 O, Sets 3 and 4) sets. The mean difference in performance on Set 1 and Set

Phonological Representations in Deaf Children 145 Table 1 Mean (SD) performance scores on congruent (O1) and incongruent (O2) trials on the syllable, rhyme, and phoneme judgment tasks

Task Syllable (maximum five/set) Set 1 (P 5 O) Set 2 (P 5 O) Set 3 (O 5 O) Set 4 (O 5 O) Set 5 (P 6¼ O) Task total raw score Composite (O1 and O2) score 13.79 (3.91)* Rhyme (maximum five/set) Set 1 (no pattern) Set 2 (tactile pattern) Set 3 (visual pattern) Set 4 (visual and tactile) Task total raw score Composite (O1 and O2) score 22.13 (7.63)* Phoneme (maximum six/set) Set 1 (no pattern) Set 2 (tactile pattern) Set 3 (visual pattern) Set 4 (visual and tactile) Task total raw score Composite (O1 and O2) score 22.17 (7.93)*

Congruent (O1)

Incongruent (O2)

M (SD)

M (SD)

4.06 (1.16)* 3.90 (1.32)*

7.96 (2.31)*

2.73 2.10 1.00 5.83

(1.29)* (1.26) (1.16)1 (2.58)

4.37 3.48 3.37 2.67 13.88

(1.27)* (1.38)* (1.50)* (1.26)* (4.39)*

3.23 2.42 1.96 1.19 8.25

(1.46)* (1.94)* (1.48) (1.19)1 (3.96)

4.77 3.94 3.50 2.33 14.54

(1.73)* (1.70)* (1.45)* (1.31) (5.30)*

3.15 1.85 1.50 1.13 7.63

(1.60)* (1.23) (1.26)1 (1.25)1 (3.61)

Note. P 5 phonological; O 5 orthographic. Congruent and O1 sets are those where a correct judgment can be made using either a phonological strategy or an orthographic strategy; incongruent and O2 sets are those where orthographic knowledge (letter matching) cannot be relied on to make a correct judgment. Asterisks (*) show performance levels above chance at p , .01 significance level. A plus sign (1) indicates performance below chance at p , .01. Mean raw scores reported.

2 (P 5 O) was not significant indicating that the size of syllable discrepancy between target and distracter items (one vs. two syllables) was not a significant factor in participant performance on these word sets. Similarly, there was no significant difference between mean performance accuracy on Set 3 and Set 4 (O 5 O) again indicating that a larger syllable discrepancy between target and distracter items had no significant effect on performance. Mean accuracy scores were significantly above chance levels across the first three conditions of the task and above chance on the fourth condition where an accurate length judgment could be made using global (course grained) perceptual strategies. On the final and most revealing level of the task, (mixed, P 6¼ O), mean performance was significantly worse than mean performance on the other four sets with the deaf students demonstrating below chance-level performance, Set 5: t(51) 5 24.06, p 5 .000. Analysis of the response choice patterns for

the mixed (P 6¼ O) set showed that 37 participants (71%) chose an orthographic match on two or more items (M 5 2.52, SD 5 1.45) indicating use of a letter counting (orthographic) strategy to match word lengths. Overall, the participants were better able to make a successful length judgment when a word’s syllable length was congruent with its orthographic length (P 5 O). They had significantly more difficulty in making a successful length judgment when orthographic length cues were no longer available to support length discrimination (O 5 O) and were reduced to below chance performance when orthographic lengths cues were misleading (P 6¼ O). Rhyme awareness. At the rhyme level, a repeated measures ANOVA was calculated with distracter type (four levels: no pattern, visual pattern, tactile pattern, visual and tactile pattern) and orthographic similarity (two levels: O1 and O2) as the within-subject factors.

146


The sphericity assumption was met. The analysis yielded a significant main effect of distracter type, F(3, 153) 5 65.68, p , .001, g2 5 .56, and a significant main effect of orthographic similarity, F(1, 51) 5 122.57, p , .001, g2 5 .71. The interaction between distracter type and orthographic similarity was not significant, F(3, 153) 5 1.15, p 5 .332, g2 5 .02, indicating that the impact on accuracy due to orthographic similarity was fairly uniform across the different distracter types. Post hoc comparisons among the four conditions (performed at .05 significance level using the Bonferroni adjustment for multiple comparisons) indicated participants were significantly more successful in making an accurate rhyme judgment on Set 1 words where the distracter items shared no similarity to the cue and target than they were on all other sets. Performance means for Set 2 (tactile pattern) and Set 3 (visual pattern) were significantly different than performance means on each other set; however, there was no significant difference in participants’ mean performance between Set 2 (tactile pattern) and Set 3 (visual pattern). Mean performance on Set 4 (visual and tactile pattern), where one distracter item shared a tactile pattern and another shared a visual-orthographic pattern but both were phonologically distinct from the cue and target, was significantly worse than mean performance on the other three sets. As with the syllable task, the response choice analysis indicated that the deaf students were predominantly using orthographic strategies (M 5 2.19, SD 5 1.43; median 2.0; mode 3.0) to support rhyme-matching judgment. Overall, performance accuracy was significantly better on sets where distracter items shared no phonological overlap or spelling pattern similarity with the target rhyme in comparison to performance accuracy on any other word sets. The addition of a tactile pattern distracter (Set 2) or a visual pattern distracter (Set 3) caused enough of a decrement in performance to suggest that these distracters were a significant source of interference. Finally, on the critical contrast condition, Set 4, where the application of phonological knowledge is clearly essential to override any tactile or visual distracter interference, the decrement in performance was marked with 85% of the participants scoring at or below chance.

Phoneme awareness. A repeated measures ANOVA with distracter type (four levels: no pattern, visual pattern, tactile pattern, and visual and tactile pattern) and orthographic similarity (two levels: O1 and O2) as within-subjects factors was calculated to examine the effect of perceptual similarity of distracter types on performance accuracy at the phoneme level. The sphericity assumption was met. The analyses revealed a significant main effect of distracter type, F(3, 153) 5 71.14, p , .001, g2 5 .58, a significant main effect of orthographic similarity, F(1, 51) 5 126.84, p , .001, g2 5 .71, and a significant interaction between distracter type and orthographic similarity, F(3, 153) 4.69, p 5 .004, g2 5 .08. The distracter type by orthographic similarity interaction indicates that performance accuracy on orthographically similar and dissimilar words was differentially affected by the different distracter types. Post hoc comparisons among the eight subsets (performed at .05 significance level using the Bonferroni adjustment for multiple comparisons) indicated that the mean performance on words within the Set 1 (O1) no pattern subset was significantly better than mean performance on words within any other subset. Performance on the Set 3 (O1) visual pattern word types was significantly different than on most other subsets with the exception that there was no significant difference in mean performance on words in the Set 3 (O1) visual pattern and the Set 2 (O1) tactile pattern subset nor between the Set 3 (O1) visual pattern and the Set 1 (O2) no pattern subset. Differences in mean performance on words in the Set 4 (O1) visual and tactile pattern subset were significant in comparison to the other subsets with the exception of the Set 2 (O2) tactile pattern subset. Mean performance on words in the Set 3 (O2) visual pattern subset was not significantly different than mean performance on words in either the Set 2 (O2) tactile pattern subset or the Set 4 (O2) visual and tactile pattern subset. Finally, mean performance accuracy on words within the Set 4 (O2) visual and tactile pattern subset was significantly worse than mean performance on words in each of the other subsets with the exclusion of the Set 3 (O2) visual pattern subset. In summary, at the phoneme level, the results indicate that when task demands were low (O1,


orthographically similar word types) and distracter items did not require fine-grained (phonological) comparisons (e.g., no pattern), the participants were better able to make a successful matching judgment. However, the inclusion of distracter items that were orthographically distinct but overlapped the cue in tactile contact (O1, tactile pattern) or were orthographically misleading (O1, visual pattern) interfered with accuracy and resulted in comparable performance between these O1 subsets and the easiest level of the O2 subsets (O2, no pattern). Overall, there was a significant decrement in performance when graphic similarity could not be relied on to support phoneme judgment, (O2 word types), and performance fell to chance level on O2 subsets requiring fine-grained phonological sensitivity to successfully discriminate form-based similarity between distracter items.

Effects of Age and Reading Ability on Performance To examine the effects of age and reading ability on the elaboration/refinement of phonological representations, performance levels for the critical contrast set at the levels of syllable (Set 5, P 6¼ O), rhyme (Set 4, O2), and phoneme (Set 4, O2) were statistically compared to expected chance-level performance using one-sample t-tests. Age and reading ability were only moderately correlated, r 5 .473, p , .01, thus separate analyses were conducted in order to better discriminate the effect of age from the effect of reading ability. First, to determine if phonological development was differentially related to age, the paricipants were divided into younger (8–13 years) and older (14–19 years) age categories consisting of 26 participants each. At the syllable level, mean performance accuracy on Set 5 (P 6¼ O) was significantly below chance for both the older group, t(25) 5 23.28, p 5 .003; M 5 0.92, SD 5 1.13 and the younger group, t(25) 5 22.44, p 5 .022; M 5 1.08, SD 5 1.19. At the rhyme level, performance accuracy on the graphically dissimilar (O2) words in Set 4 was not significantly different from chance for the older group, t(25) 5 21.46, p 5 .156; M 5 1.31, SD 5 1.19 and was significantly below chance, t(25) 5 22.44, p 5 .022, for the younger

students (M 5 1.08, SD 5 1.20). Performance at the phoneme level Set 4 (O2) was significantly below chance for both the older, t(25) 5 22.33, p 5 .028 (M 5 1.35, SD 5 1.38), and younger, t(25) 5 -4.93, p , .001 (M 5 0.92; SD 5 1.09) groups. Chance or below chance patterns of performance across all the critical contrast sets at each level of phonological structure indicate that increasing age was not associated with the elaboration of phonological representations for the participants in this study. Next, to determine if phonological development was related to reading development, the participants were separated into two groups determined by reading age (RA), those reading above and below age nine level because nine years of age corresponds to the third grade median level of reading achievement of the deaf school-aged population. There were 31 participants in the RA ,9 group and 21 participants in the RA .9 group. Again, the students’ performance on the final set of each task was analyzed separately using one sample t-tests. At the syllable level, the analysis showed below chance-level performance for both groups with the RA .9 readers performing with a mean accuracy of 1.05 (SD 5 1.44), t(20) 5 22.29, p 5 .033, and the RA ,9 readers performing with a mean accuracy of 0.97 (SD 5 1.14), t(30) 5 23.33, p 5 .002. At the rhyme level, the RA .9 readers performed at chance level, t(20) 5 2.097, p 5 .924 (M 5 1.62, SD 5 1.47). The RA ,9 readers performance was significantly below chance, t(30) 5 24.78, p 5 .000 (M 5 0.90, SD 5 .870). A similar pattern of performance was noted at the phoneme level, with the RA .9 readers performing at chance level, t(20) 5 21.46, p 5 .161 (M 5 1.52, SD 5 1.44). The RA ,9 readers performance was significantly below chance t(30) 5 25.85, p 5 .000 (M 5 0.87, SD 5 1.06). Chance or below chance performance on the critical contrast sets at each level of phonological structure for the good reader group indicates that increased reading ability did not result in functional phonological representations for these deaf students.

Discussion This study examined the performance of children and adolescents with a prelingual severe to profound

148


hearing loss on experimental tasks designed to tap awareness of phonological structure at the level of the syllable, rhyme, and phoneme. In addition to our specific question concerning the extent of phonological elaboration in the underlying representations of deaf children, our interest touched on the specificity of these representations given that the modalities that transmit spoken language are altered. First, it is important to reiterate that nameable pictures were used as the stimuli for each experimental task because the use of picture stimuli provides a cleaner measure of phonological awareness than print presentation allows. As Schiano and Watkins (1981) demonstrate, nameable pictures are remembered in phonological rather than visual codes. However, because phonological information is not present in picture stimuli, a participant has to internally generate the phonological representation of the intended item label before the name of the picture can be held in short-term memory. The phonological tasks used in this study were not intended to tap the speed of word identification, but instead the process by which the identified item label activated by a picture stimulus was judged as being similar to a cue in the phonological unit tested. Thus, the process being studied is the process of encoding the picture stimuli, and the labels activated by the stimuli and then deciding which item label is a better fit with the cue within the phonological units of syllable, rhyme, or phoneme. The use of perceptually similar distracters was intended to clarify the extent to which the resolution of phonological units may be differentially affected by an acoustic factor, a tactile factor, or an orthographic factor. The first finding of interest across the three experimental tasks was that a composite score reflecting overall task performance on both orthographically similar (O1) and dissimilar (O2) words was significantly above chance for the deaf participants studied. This finding is consistent with the reports of positive evidence of phonological awareness reported in previous studies. However, in this study when the data was reanalyzed according to O1 and O2 categories, there was a significant effect of orthographic similarity with performance accuracy seriously reduced across all O2 categories. For example, on the final level of the syllable task, orthographic length between cue and

target was mismatched (same number of syllables but different number of letters e.g., ‘‘sandwich–taxi’’) to determine if phonological specifications for item labels were sufficiently distinct to overcome this kind of orthographic confusion. As well, the final O2 level of the rhyme and phoneme tasks included both visual and tactile pattern distracters to determine whether phonological specifications for item labels were sufficiently defined to overcome these kinds of orthographic and/or tactile confusions. At this level of task difficulty across all three tasks, the application of phonological knowledge was clearly essential in order to make a successful syllable, rhyme, or phoneme match. Significantly, participant performance fell to below chance in every instance. Such performance patterns indicate that, unlike hearing children of a much younger age for whom these phonological structures are well established (see review in Goswami, 2002), the syllable, rhyme, and phoneme are not well-represented phonological structures for the deaf children and adolescents studied. Of note, the younger deaf children studied (6–13 years) did not show less sensitivity to phonological structure relative to the older children (14–19 years). Although the older students were slightly more accurate in their ability to make matching judgments across all three tasks, the difference in performance accuracy between groups was not significant suggesting that representation of phonological structure is not a ‘‘developing skill’’ for these children with severe to profound, prelingual hearing loss. Similarly, the reading data indicated that reading abilities ranged from poor to skilled despite similar insensitivity to phonological structure across all participants. This would argue that for these learners, awareness of phonological structure was not refined or educated by increased exposure to or experience with print. Arguably, however, slightly better performance of the more skilled readers on the orthographically similar (O1) words may indicate that orthographic word-analytic abilities may be developed through increased exposure to print. These findings then are in direct contrast to typical patterns reported in the literature with hearing individuals and are inconsistent with traditional equivalence proposals in the deafness literature that awareness of


phonological structure is a developing skill that improves with age and/or reading ability. Lending support to the argument above, a finding of interest across all three tasks was that there was a significant ‘‘stimulus similarity’’ interference effect induced by the perceptual overlap of distracter items to the cue stimulus. Difficulty in differentiating between phonetically similar but phonologically contrastive words (e.g., kite and gun or plate and bleed) was evidenced across all three tasks suggesting that the underlying representation for the students in this study may not be informationally rich enough to support accurate discriminations. For example, reliable and fairly large differences in performance accuracy were obtained on the rhyme and phoneme task by stimuli that differ from the cue word in phonological, tactile, and orthographic representations. On the first level of each task in which the distracter items were neutral with respect to shared perceptual similarity with the cue and target, accuracy was better than on all other remaining sets. At this level of the task, however, successful performance could be made using either a phonological or an orthographic strategy. Across all other sets in each task, the error rate difference was bigger when the distracter items were orthographically related to the cue, but there was still a significant interference effect with tactile pattern distracters— especially when there was no orthographic (spelling) similarity between the cue and the target phonological response (e.g., O2 cue: kite, response choices: night, gun, moon). Response choice analyses revealing that tactile patterning for the deaf students studied rendered the phonetic realization of a word such as ‘‘kite’’ as a companion rhyme to a word like ‘‘gun’’ suggests that the internally generated code (the code activated) may not include information about phonemic (i.e., meaning bearing) contrasts between one word and another. More significant perhaps, on O1 word sets (e.g., O1 cue: boat, response choices: coat, jump, pool) where a successful match could have been made on an orthographic basis by matching spelling patterns, the internally generated code often failed to distinguish between words that have different orthographic representations resulting in responses that, for example ‘‘pool’’ was the companion rhyme to the cue ‘‘boat’’. Importantly, these types of errors are thought

to denote the state of the internal representation. Clearly, if a visual/tactile code were able to compensate or substitute as a phonological representation for deaf learners such confusions should not be evident. Typically hearing children rarely, if ever make mistakes of this type. Visual analysis of seen speech may be sufficient for establishing representations that have some global speech-like properties and thus be sufficient for recognizing spoken words. Critically however, the ability to perceive global characteristics of the speech signal (i.e., detect and encode course-grained speech-like properties) is not the same as the ability to detect and represent fine-grained phonetic details. It is the latter ability (fine-grained contrast sensitivity) that appears critical in supporting word learning in both spoken (e.g., Fowler, 1991; Jusczyk, Hohne, & Mandel, 1995; Metsala & Walley, 1998; Pater, Stager, & Werker, 2004; Werker & Yeung, 2005) and written language acquisition (e.g., Brady, 1997; Carroll & Snowling, 2004; Chiappe, Chiappe, & Siegel, 2001; Elbro, 1996; Goswami, 2000; Hayes,Tippana, Nicol, Sams, & Kraus, 2003; Mody, 2003; Morais, 2003; Serniclaes, Sprenger-Charolles, Carre´, & De´monet, 2001; Swan & Goswami, 1997; Ziegler & Goswami, 2005). Phonological awareness tasks requiring a simple matching judgment measure what Stanovich (2000) refers to as the basic or surface level of phonological awareness. At a surface level, the participants in this study were not able to reliably exploit the phonological, meaning bearing contrasts that normally enhance discrimination across syllable, rhyme, and phoneme boundaries. The findings of our investigation then are in line with this growing body of data that suggest that both the type and the amount of information that is perceptually available in the speech signal may be critical factors in enabling children to distinguish patterns of phonetic similarity between words and thereby establish segmental representational structure. It is reasonable to conclude that for deaf learners, difficulty in establishing spoken language segmental structure would have similar predictable effects as those noted for hearing children on acquisition, access, and retrieval of lexical representations and on the acquisition of other grammatical properties of spoken language (see review in Lewkowicz & Kraebel, 2004).

150


It is important to reiterate that although the use of phonology predicts above chance-level performance on tasks of phonological awareness, performance above chance does not in itself indicate the use of a phonological strategy. Past studies involving phonological awareness testing of deaf participants have produced inconsistent results. In contrast, the results from this study using tasks requiring phonological disambiguation between sensory cues were straightforward. When deaf children made decisions requiring syllable, rhyme, or phoneme discrimination, matching judgments were not tied to phonological facilitation as they are with hearing learners. On these tasks, although it is clear that some sort of pattern code was computed that activated a lexical item, there is little evidence to suggest that there was anything fundamentally phonological in the code that was generated. Although our findings suggest that both visual (seen speech) and tactile (felt speech) information may be integral parts of the input signal for deaf learners, they similarly suggest that deaf individuals may have difficulty in representing and/or integrating multisensory information. As such, they underscore the important role played by auditory information in the refining and shaping of phonological representations (Campbell & MacSweeney, 2004; D’Hondt & Leybaert, 2003; Nittrouer & Burton, 2005) and the limitations of visual/tactile information alone in establishing meaningful linkages between sensory cues. Although it may be argued that visual/tactile cues are a part of the phonological evidence in the input signal, it is clear that for these deaf learners it is not the ‘‘right kind’’ of sensory evidence that supports awareness of similarities and contrasts in the sounds of words. With the understanding that it is contrast that is fundamental to phonology, it is proposed that use of the term phonological to account for deaf learners’ performance patterns on spoken language phonological tasks may be misleading—at least to the extent that conventional use of the term carries with it an underlying assumption of ‘‘meaningful’’ integration of sensory cues. This study examined the extent to which deaf learners have awareness of phonological structure across three levels of linguistic complexity. The results of this study represent a replication of previous find-

ings of insensitivity to phonological structure at the rhyme and phoneme level in children with severe to profound prelingual hearing losses and extend those findings to a wider age range and to all three levels of linguistic structure. These findings also offer an account for the inconsistent findings reported in the literature with respect to deaf learners phonological awareness skills. That is, it is possible to account for the ‘‘phonological awareness’’ reported in previous studies without necessarily positing a link to phonology. In light of current understandings of the multisensory nature of speech perception, further study is necessary to clearly explicate the properties or the nature of the code that is represented and generated by deaf individuals before any conclusions of phonological effects can be validated. Demonstration of insensitivity to phonological structure in young deaf children and older adolescents and in good and poor readers strongly suggests that difficulties with spoken language phonological awareness are not outgrown, do not represent a developmental lag and are persistent and pervasive throughout at least adolescence despite intensive and long-term interventions. Thus, the conjecture that deaf children’s phonological development is qualitatively similar though quantitatively delayed in comparison to hearing children is not supported. For the students in this study, the visual (lip-read) and tactile evidence in the input signal did not appear to be sufficiently informative to provide these students with the means by which to shape and refine phonological information. The current investigation provides support for the notion that reduced input specificity of seen (speech read) as compared to heard speech is not likely to support deaf children in developing a spoken language phonological system with accurately specified or segmentally organized internal representational structure. As such, our results do not support long-held assumptions of functional equivalence between perceptual modalities (auditory and visual) in the development or elaboration of phonological representations in deaf learners. By extension, approaches to reading instruction that are focused on making sound-based phonology ‘‘visible’’ to deaf students must be evaluated against their potential to develop a robust and reliable internal system of representation of spoken language phonological


structure for deaf learners. If lacking such potential, difficulties in securing written words in memory, as well as with accuracy and speed of word recognition would be predicted. It must be noted, however, that the reading abilities for the participants in this study ranged from poor to very skilled despite similar insensitivity to spoken language phonological structure across all participants. Of particular interest, the children in this study communicate via natural sign language and thus have complete access to the sublexical (sign phonemes) structure of a natural language. This provides an alternate means of coding words and establishing fully specified phonological representations of words as they are learned. Given the intimate link observed between phonological development and lexical development in spoken language users, the extent to which sign language phonological development might similarly provide the linguistic and cognitive underpinnings for successful use of written language for sign language users is an open question that deserves further experimental scrutiny. The struggle with reading is a well-documented fact in the lives of most deaf children. Oakhill, Cain, and Bryant (2003) suggest that ‘‘skilled comprehenders build better-integrated and informationally richer text representations’’ (p. 444). Similarly, in relation to word recognition, it appears that skilled word readers build better-integrated and informa-

tionally richer word representations. As Phillips (2002) asserts, the ability to read well depends on how readers employ the evidence available to them. Perhaps, the more general debate about the use of spoken language or the use of sign language in the education of deaf children is not as important as turning our research attention to a more detailed investigation of what critical evidence must be available in the input signal to support deaf learners in the process of lexical elaboration and, in turn, the construction of rich and robust word representations. Such research may further our understanding of reading and its development for deaf learners and may help to clarify what works, for whom, and under what conditions. Importantly, the findings of this study indicate that there may be skills other than spoken language phonological abilities that play critical roles in the reading achievement of signing deaf individuals. For deaf children who use a natural sign language, it appears that the available evidence in the input signal does not need to be ‘‘spoken.’’ Like all children, however, the available evidence may need to be phonological.

Funding Social Sciences and Humanities Research Council of Canada Doctoral Fellowship; David Peikoff Chair of Deafness Studies (University of Alberta) to L.M.

Appendix A: Syllable judgment task examples Congruent sets (P 5 O): cue and target match in both phonological length (no. of syllables) and orthographic letters) Set 1 Cue Target Distracter Distracter Set 2 Cue Target Distracter Two-syllable bug van eraser cereal One-syllable moon blue mountain discrepancy helicopter caterpillar socks gum discrepancy yellow window rain Incongruent sets (O 5 O): orthographic length (no. of letters) is matched across all words in a quadruplet Distracter Target One-syllable Cue Distracter Set 4 Target Cue Set 3 stairs Distracter discrepancy potato eleven football elephant Two-syllable hospital table horse green thirteen piano mouse discrepancy swing video Mixed set (P 6¼ O): phonological length does not agree with orthographic length Set 5 Cue Target Distracter Distracter sandwich taxi helicopter comb flag tongue lion pencil school sun money family

length (no. of Distracter teacher grass Distracter friend river

152


Appendix B:

Rhyme judgment task examples

Set 1 (no pattern): both distracters have no orthographic, phonological, or tactile similarity to the target (O–P–T–) Type Cue Target No pattern distracter No pattern distracter Type Cue Target No pattern distracter No pattern distracter O1 king ring cheese blue O2 cry pie bed dog O1 car star tree boy O2 blue two rain hand Set 2 (tactile pattern): one distracter item shares a similar tactile/motoric pattern to the target (O–P–T1) Type Cue Target Tactile distracter No pattern distracter Type Cue Target Tactile distracter No pattern distracter O1 boat coat pool jump O2 plate wait bleed dog O1 night fight tent blue O2 kite night gun moon Set 3 (visual pattern): one distracter item shares a similar visual/orthographic pattern to the target (O 1 P 2 T 2) Type Cue Target Visual distracter No pattern distracter Type Cue Target Visual distracter No pattern distracter O1 phone bone one milk O2 boot fruit foot cave O1 four pour sour hand O2 shoe blue toe ring Set 4 (visual and tactile pattern): one distracter item shares a similar visual pattern to the target (O 1 P 2 T 2). One distracter item shares a similar tactile pattern to the target (O 2 P 2 T 1) Type Cue Target Visual distracter Tactile distracter Type Cue Target Visual distracter Tactile distracter O1 kite white knit gun O2 sour flower soup zero O1 tail jail tall night O2 chair square chain shy O1 bead leaf head pen O2 pour sour sour boy Note. Type: O1: orthographically similar rhyme pairs (share a spelling pattern); O2: orthographically dissimilar rhyme pairs (do not share a spelling pattern). Distracters: OPT 5 orthographic, phonological, tactile/motoric; 1 indicates overlap with the target; 2 indicates no overlap with the target.

Appendix C:

Phoneme judgment task examples

Set 1 (no pattern): both distracters have no orthographic, phonological or tactile similarity to Phoneme No pattern No pattern position Type Cue Target distracter distracter Type Cue Target Initial O1 milk moon bear key O2 car key Final O1 balloon clown girl tree O2 grass dance Medial O1 plane face bed dog O2 coat rope Set 2 (tactile pattern): one distracter item shares a similar tactile/motoric pattern to Phoneme Tactile No pattern position Type Cue Target distracter distracter Type Cue Initial O1 foot farm van jump O2 gym Final O1 walk milk rug car O2 knife Medial O1 boat soap four moon O2 rain

the target (O2P2T2) No pattern No pattern distracter distracter apple plane bed hand ball eye

the target (O 2 P 2 T 1) Tactile No pattern Target distracter distracter jump chip dog cough dive plane face white cow

Set 3 (visual pattern): one distracter item shares a similar visual/orthographic pattern to the target (O 1 P 2 T 2) Phoneme Visual No pattern Visual No pattern position Type Cue Target distracter distracter Type Cue Target distracter distracter Initial O1 bed ball red cat O2 gym jam gum pig Final O1 key candy sky snow O2 movie baby pie rain Medial O1 tail rain hair boat O2 soup juice mouse tree Set 4 (visual and tactile pattern): one distracter item shares a similar visual pattern to the target (O 1 P 2 T 2). One distracter item shares a similar tactile pattern to the target (O 2 P 2 T 1) Phoneme Visual Tactile Visual Tactile position Type Cue Target distracter distracter Type Cue Target distracter distracter Initial O1 knee knit kiss tea O2 walk one write rug Final O1 cat boat car gun O2 snow toe cow store Medial O1 bead leaf head pen O2 pour door sour boy


References Alegria, J. (1998). The origin and functions of phonological representations in deaf people. In C. Hulme & R. M. Joshi (Eds.), Reading and spelling: Development and disorders (pp. 263–286). Hillsdale, NJ: Lawrence Erlbaum Associates. Alegria, J. (2004). Deafness and reading. In T. Nunes & P. Bryant (Eds.), Handbook of children’s literacy (pp. 459–489). Dordrecht, The Netherlands: Kluwer Academic. Blachman, B. A. (1994). Early literacy acquisition: The role of phonological awareness. In C. Wallach & K. Butler (Eds.), Language learning disabilities in school age children and adolescents: Some underlying principles and applications (pp. 253–274). Columbus, OH: Merrill/Macmillan. Brady, S. (1997). Ability to encode phonological representations: An underlying difficulty of poor readers. In B. A. Blachman (Ed.), Foundations of reading acquisition and dyslexia: Implications for early intervention (pp. 21–47). Mahwah, NJ: Lawrence Erlbaum Associates. Campbell, R. (1987). The cerebral lateralization of lip-reading. In B. Dodd & R. Campbell (Eds.), Hearing by eye: The psychology of lip-reading (pp. 215–226). Hillsdale, NJ: Lawrence Erlbaum Associates. Campbell, R. (1990). Lipreading, neuropsychology and immediate memory. In G. Vallar & T. Shallice (Eds.), Neuropsychological impairments of short term memory (pp. 268–287). Cambridge, UK: Cambridge University Press. Campbell, R. (1997). Read the lips: Speculations about the nature and role of lipreading in cognitive development of deaf children. In M. Marschark, D. Lillo-Martin, R. Campbell, & V. S. Everhart (Eds.), Relations of language and thought: The view from sign language and deaf children (pp. 110–146). New York: Oxford University Press. Campbell, R., Burden, V., & Wright, H. (1992). Spelling and speaking in pre-lingual deafness: Unexpected role for isolated ‘‘alphabetic’’ spelling skills. In C. M. Sterling & C. Robson (Eds.), Psychology, spelling and education (pp. 185–199). Philadelphia, PA: Multilingual Matters Ltd. Campbell, R., & MacSweeney, M. (2004). Neuroimaging studies of crossmodal plasticity and language processing in deaf people. In G. A. Calvert, C. Spence, & B. A. Stein (Eds.), The Handbook of multisensory processing (pp. 773–784). Cambridge: MIT Press. Campbell, R., & Wright, H. (1988). Deafness, spelling and rhyme: How spelling supports written word and picture rhyming skills in deaf subjects. Quarterly Journal of Experimental Psychology, 40A, 771–788. Caravolas, M. (1993). Language-specific influences of phonology and orthography on emergent literacy. In J. Altarriba (Ed.), Cognition and Culture: A cross-cultural approach to psychology (pp. 177–205). Amsterdam, The Netherlands: Elsevier Science Publishers. Carroll, J., & Snowling, M. (2004). Language and phonological skills in children at high risk of reading difficulties. Journal of Child Psychology and Psychiatry, 445, 631–640. Chiappe, P., Chiappe, D. L., & Siegel, L. S. (2001). Speech perception, lexicality, and reading skill. Journal of Experimental Child Psychology, 80, 58–74.

D’Hondt, M., & Leybaert, J. (2003). Lateralization effects during semantic and rhyme judgment tasks in deaf and hearing subjects. Brain and Language, 87, 227–240. Dodd, B. (1976). The phonological systems of deaf children. Journal of Speech and Hearing Disorders, 41, 185–198. Dodd, B. (1987). Lip-reading, phonological coding and deafness. In B. Dodd & R. Campbell (Eds.), Hearing by eye: The psychology of lip-reading (pp. 177–189). Hillsdale, NJ: Lawrence Erlbaum Associates. Dodd, B., & Hermelin, B. (1977). Phonological coding by the prelinguistically deaf. Perception and Psychophysics, 21, 413–417. Elbro, C. (1996). Early linguistic abilities and reading development: A review and a hypothesis. Reading and Writing: An Interdisciplinary Journal, 8, 453–485. Fowler, A. (1991). How early phonological development might set the stage for phoneme awareness. In S. Brady & D. Shankweiler (Eds.), Phonological processes in literacy: A tribute to Isabelle Y. Liberman (pp. 97–117). Hillsdale, NJ: Lawrence Erlbaum Associates. Fowler, A., & Swainson, B. (2004). Relationships of naming skills to reading, memory, and receptive vocabulary: Evidence for imprecise phonological representations of words by poor readers. Annals of Dyslexia, 54, 247–280. Fowler, C. A. (1986). An event approach to the study of speech perception from a direct-realist perspective. Journal of Phonetics, 14, 3–28. Goswami, U. (2000). Phonological representations, reading development and dyslexia: Towards a cross-linguistic theoretical model. Dyslexia, 6, 133–151. Goswami, U. (2002). Early phonological development and the acquisition of literacy. In S. Neuman & D. Dickinson (Eds.), Handbook of early literacy research (pp. 111–125). New York: Guilford Press. Hanson, V. (1982). Short-term recall by deaf signers of ASL: Implications of encoding for order recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 572–583. Hanson, V. (1989). Phonology and reading: Evidence from profoundly deaf readers. In D. Shankweiler & I. Liberman (Eds.), Phonology and reading disability: Solving the reading puzzle (pp. 69–89). Ann Arbor, MI: University of Michigan Press. Hanson, V. (1991). Phonological processing without sounds. In S. Brady & D. Shankweiler (Eds.), Phonological processes in literacy: A tribute to Isabelle Y. Liberman (pp. 153–161). Hillsdale, NJ: Lawrence Erlbaum. Hanson, V., & Fowler, C. (1987). Phonological coding in word reading: Evidence from hearing and deaf readers. Memory and Cognition, 15, 199–207. Hanson, V., & McGarr, N. S. (1989). Rhyme generation in deaf adults. Journal of Speech and Hearing Research, 32, 2–11. Harcourt Brace. (1996). Stanford Achievement Test: Ninth Edition. New York: Harcourt Brace and Company. Harris, M., & Beech, J. (1998). Implicit phonological awareness and early reading development in prelingually deaf children. Journal of Deaf Studies and Deaf Education, 3, 205–216.

154


Hayes, E. A., Trippana, K., Nicol, T. G., Sams, M., & Kraus, N. (2003). Integration of heard and seen speech: A factor in learning disabilities in children. Neuroscience Letters, 351, 46–50. Hayes, P., & Arnold, P. (1992). Is hearing-impaired children’s reading delayed or different? Journal of Research in Reading, 15, 104–116. Izzo, A. (2002). Phonemic awareness and reading ability: An investigation with young readers who are Deaf. American Annals of the Deaf, 147, 18–28. Jarvis, B. (2004). DirectRT precision timing software. (version 2004.1.0.13 [computer software]). New York: Empirisoft. Jusczyk, P. W., Hohne, E., & Mandel, D. (1995). Picking up regularities in the sound structure of the native language. In W. Strange (Ed.), Speech perception and linguistic experience: Issues in cross-language research (pp. 91–119). Baltimore, MD: York Press. Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American English. Providence, RI: Brown University Press. Lewkowicz, D. J., & Kraebel, K. S. (2004). The value of multisensory redundancy in the development of intersensory perception. In G. Calvert, C. Spence, & B. Stein (Eds.), Handbook of multisensory processing (pp. 1–76). Cambridge: MIT Press. Leybaert, J. (1993). Reading in the deaf: The roles of phonological codes. In M. Marschark & M. Clark (Eds.), Psychological perspectives on deafness (pp. 269–309). Hillsdale, NJ: Lawrence Erlbaum. Leybaert, J. (1998). Phonological representations in deaf children: The importance of early linguistic experience. Scandinavian Journal of Psychology, 11, 1–29. Liberman, A. M., & Mattingly, I. G. (1985). The motor theory of speech perception revised. Cognition, 21, 1–36. Markwardt, F. C. (1989). Peabody Individual Achievement Test— Revised (PIAT-R). Circle Pines, MN: American Guidance Service. Marschark, M. (2001). Language development in children who are deaf: A research synthesis (Project FORUM Report). Alexandria, VA: National Association of State Directors of Special Education. Marschark, M., & Harris, M. (1996). Success and failure in learning to read: The special case (?) of deaf children. In C. Cornoldi & J. Oakhill (Eds.), Reading comprehension difficulties: Processes and intervention (pp. 279–300). Mahwah, NJ: Lawrence Erlbaum Associates. Metsala, J., & Walley, A. (1998). Spoken vocabulary growth and the segmental restructuring of lexical representations: Precursors to phonemic awareness and early reading ability. In J. Metsala & L. Ehri (Eds.), Word recognition in beginning literacy (pp. 89–120). Mahwah, NJ: Lawrence Erlbaum Associates. Miller, P. (1997). The effect of communication mode on the development of phonemic awareness in prelingually deaf students. Journal of Speech, Language, and Hearing Research, 40, 1151–1163.

Mody, M. (2003). Phonological basis in reading disability: A review and analysis of the evidence. Reading and Writing: An Interdisciplinary Journal, 16, 21–39. Morais, J. (2003). Levels of phonological representation in skilled reading and in learning to read. Reading and Writing: An Interdisciplinary Journal, 16, 123–151. Nittrouer, S., & Burton, L. T. (2005). The role of early language experience in the development of speech perception and phonological processing abilities: Evidence from 5-year-olds with histories of otitis media with effusion and low socioeconomic status. Journal of Communication Disorders, 38, 29–63. Oakhill, J. V., Cain, K., & Bryant, P. E. (2003). The dissociation of word reading and text comprehension: Evidence from component skills. Language and Cognitive Processes, 18, 443–468. Pater, J., Stager, C., & Werker, J. (2004). The perceptual acquisition of phonological contrasts. Language, 80, 384–402. Paul, P. (1998). Literacy and deafness: The development of reading, writing and literate thought. Needham Heights, MA: Allyn & Bacon. Paul, P. (2001). Language and deafness. San Diego, CA: Singular. Perfetti, C., & Sandak, R. (2000). Reading optimally builds on spoken language: Implications for deaf readers. Journal of Deaf Studies and Deaf Education, 5, 32–50. Phillips, L. M. (2002). Making new or making do: Epistemological, normative and pragmatic bases of literacy. In J. Brockmeier, M. Wang, & D. R. Olson (Eds.), Literacy, narrative and culture (pp. 283–300). Surrey, United Kingdom: Curzon Press. Schiano, D. J., & Watkins, M. J. (1981). Speech-like coding of pictures in short-term memory. Memory and Cognition, 9, 110–114. Serniclaes, W., Sprenger-Charolles, L., Carre´, R., & De´monet, J. (2001). Perceptual discrimination of speech sounds in developmental dyslexia. Journal of Speech, Language & Hearing Research, 44, 384–399. Stanovich, K. E. (2000). Progress in understanding reading: Scientific foundations and new frontiers. New York: Guilford Press. Sterne, A., & Goswami, U. (2000). Phonological awareness of syllables, rhymes, and phonemes in deaf children. Journal of Child Psychology and Psychiatry, 41, 609–625. Swan, D., & Goswami, U. (1997). Phonological awareness deficits in developmental dyslexia and the phonological representations hypothesis. Journal of Experimental Child Psychology, 66, 18–41. Werker, J., & Yeung, H. (2005). Infants speech perception bootstraps word learning. Trends in Cognitive Science, 9, 518–526. Ziegler, J., & Goswami, U. (2005). Reading acquisition, developmental dyslexia, and skilled reading across languages: A psycholinguistic grain size theory. Psychological Bulletin, 131, 3–29. Received February 17, 2008; revisions received May 22, 2008; accepted June 3, 2008.