SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 1 Spelling of Deaf Children Who Use Cochlear Implants

The spellings of 39 profoundly deaf users of cochlear implants, aged 6–12 years, were compared with those of 39 hearing peers. When controlled for age and reading ability, the error rates of the 2 groups were not significantly different. Both groups evinced phonological spelling strategies, performing better on words with more typical sound–spelling correspondences and often making misspellings that were phonologically plausible. However, the magnitude of these phonological effects was smaller for the deaf children than for hearing children of comparable reading and spelling ability. Deaf children with cochlear implants made the same low proportion of transposition errors as hearing children. The findings indicate that deaf children do not rely primarily on visual memorization strategies, as suggested by previous studies. However, deaf children with cochlear implants use phonological spelling strategies to a lesser degree than hearing peers. SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 3 Spelling of Deaf Children Who Use Cochlear Implants For people who have early hearing loss, achieving average literacy levels by adulthood is a difficult task. The median reading comprehension skills for deaf 17-year-olds in the U.S. is at a fourth-grade (9-year-old) level, a delay of about eight years (Holt, 1993; Traxler, 2000). Deaf children’s spelling is also poorer than that of hearing children (Allman, 2002; Burden & Campbell, 1994; Gates & Chase, 1926; Harris & Moreno, 2004; Sutcliffe, Dowker, & Campbell, 1999; Watson, 2002). Even in this age of Twitter and texting, errors in written spelling can lead to negative perceptions about a writer’s abilities (Figueredo & Varnhagen, 2005; Kreiner, Schnakenberg, Green, Costello, & McClin, 2002; Varnhagen, 2000), and poor literacy skills are linked to later economic and employment success (e.g., Kutner et al., 2007). Thus, it is important to study spelling skills, particularly those of deaf children, who may have limited educational and employment opportunities if their literacy development is significantly delayed. Spelling is fundamentally a transcription of spoken language, and current theories consider the acquisition of sound-to-spelling mappings to be critical for spelling success (Ehri, 1997). Given this, it is not surprising that people who do not have access to spoken language have difficulty linking words to spellings. Hearing spellers take advantage of sound-to-spelling mappings: They perform better on words with typical spellings—where each sound is represented by one of its most common spellings—than on words with atypical spellings (Leybaert & Alegria, 1995; Treiman, 1993; Waters, Bruck, & Seidenberg, 1985), and they often make phonologically plausible errors, using letters that are legal spellings for the same sound in other words (Leybaert & Alegria, 1995; Treiman, 1993; Waters et al., 1985). Because phonemegrapheme knowledge is a critical component for successful acquisition of spelling skills in SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 4 hearing children, it is important to investigate how deaf children’s spelling skills might benefit from knowledge of sound-spelling relationships that they may acquire through advanced listening devices, such as cochlear implants. With cochlear implantation, profoundly deaf children have the opportunity to hear spoken language. Although the implant does not deliver the exact equivalent of natural speech, it may make it easier for deaf children to learn mappings between sounds and spellings. Studies have shown that many children with cochlear implants achieve reading levels that are comparable to hearing children of the same age (e.g., Geers, 2003; Geers, Tobey, Moog, & Brenner, 2008). However, spelling has not been adequately investigated in this rapidly growing population. In order to determine whether cochlear implants have an effect on deaf children’s spelling, we compared oral deaf children with implants to hearing peers and asked the following questions: Do implanted children spell as well as hearing children? When they make an error, is it of the same nature as hearing children’s? Prior research on spelling in deaf children has focused on deaf children without cochlear implants, and those studies will be discussed later in this introduction. A few spelling studies included a few implanted children but did not analyze their data separately from those of children who used either hearing aids or no device. To our knowledge, only one study exclusively examined spelling of children with cochlear implants. Watson (2002) looked at eight children who received their implants prior to age five. There was no comparison group of hearing children, and the author did not report the primary communication method of the deaf children with implants. Watson divided errors into visual (e.g., maicg for magic, wrok for work) and phonological (e.g., casl for castle, dirdy for dirty) categories. She reported that all eight of SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 5 the implanted children made visual errors and that six used some phonological strategies. However, she did not report the proportion of visual and phonologically plausible errors, and she provided no additional analyses of the data other than listing examples of spelling errors from each child. In addition, Watson acknowledged that the error categories were quite vaguely defined. Accordingly, the results from this study must be interpreted with caution. In the current study, we wanted to examine several properties of spelling in children with cochlear implants. We looked to the literature of past studies of spelling in deaf children without cochlear implants. (In this review of past research, the unqualified term deaf children is used whenever the study did not report whether the participants used cochlear implants. In most such cases, the reasonable assumption is that the children did not use them.) These studies examined error rates, evidence for phonological strategies, and evidence for visual memorization strategies. As mentioned earlier, deaf children make more errors than hearing children of the same age. They are as accurate as hearing children of the same reading age (Burden & Campbell, 1994; Gates & Chase, 1926; Harris & Moreno, 2004; Kyle & Harris, 2006), but the strategies that they use to achieve this level of performance may differ from those used by hearing children The evidence for phonological strategies is somewhat conflicting. Most studies have found that deaf children spell words with more typical spellings more accurately than words with atypical spellings (Burden & Campbell, 1994; Kyle & Harris, 2006; Leybaert & Alegria, 1995; Leybaert, 2000; Leybaert & Lechat, 2001; Sutcliffe et al., 1999), but Dodd (1980) did not find this typicality effect. One might expect that deaf children would be less influenced by typicality than hearing children. Burden and Campbell (1994) found this to be the case, but Kyle and Harris (2006) did not. Another diagnostic of a phonological strategy is errors that are based on SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 6 how the word sounds. Phonologically plausible errors, such as pennsul for pencil, represent all the phonemes of a word but use some letter correspondences that are orthographically correct only in other words. These could be considered good errors, in that they are likely to be interpreted correctly by the reader. The spelling errors of deaf children are sometimes phonologically plausible, but they make phonologically plausible errors far less frequently than hearing children (Aaron, Keetay, Boyd, Palmatier, & Wacks, 1998; Dodd, 1980; Harris & Moreno, 2004; Hoemann, Andrews, Florian, Hoemann, & Jensema, 1976; Leybaert & Alegria, 1995; Leybaert, 2000; Leybaert & Lechat, 2001; Sutcliffe et al., 1999). Evidence for deaf children’s use of visual memorization strategies toward spelling is more robust. Deaf children show a frequency effect, in that they spell more frequent words more accurately than less frequent words (Burden & Campbell, 1994; Leybaert, 2000; Sutcliffe et al., 1999). Hearing children show a frequency effect also (Burden & Campbell, 1994; Lété, Peereman, & Fayol, 2008; Leybaert, 2000; Treiman, 1993), but there are mixed findings as to whether it is more or less strong than the frequency effect found in deaf children. Burden and Campbell (1994) found that deaf children were more sensitive to frequency than hearing children matched for reading ability, but Leybaert (2000) found that deaf children who received intensive cued-speech instruction were less affected by frequency than hearing children. 1 Deaf children also make letter transposition errors that are phonologically implausible, such as wrom for worm (Hoemann et al., 1976; Padden, 1993). Hearing children occasionally make these types of errors 1 Cued speech is a communication method whereby a person uses spoken language while simultaneously making hand gestures near the face to disambiguate certain phonemes that are difficult to lip read. SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 7 also, but several studies have reported that deaf children make transposition errors far more often than hearing peers (Aaron et al., 1998; Leybaert, 2000; Leybaert & Alegria, 1995). For example, Leybaert and Alegria reported that 7% of deaf children’s errors were transpositions as compared to 1% of hearing children’s. One concern with the above-cited studies is that most failed to report the type of hearing device used, if any (Aaron et al., 1998; Burden & Campbell, 1994; Dodd, 1980; Harris & Moreno, 2004; Hoemann et al., 1976; Johnson, Padak, & Barton, 1994; Leybaert & Alegria, 1995; Padden, 1993; Sutcliffe et al., 1999). As mentioned earlier, if any of the participants did use cochlear implants, their data were not analyzed separately (Kyle & Harris, 2006; Leybaert & Lechat, 2001). Also, several of the studies did not focus specifically on children who used oral language. Instead, many of the children used sign language, finger spelling, or cued speech (Harris & Moreno, 2004; Hoemann et al., 1976; Leybaert, 2000; Padden, 1993; Sutcliffe et al., 1999), all of which can affect how the children learn and relate to the alphabet and pronunciation. Another concern with previous studies is in the poor definitions of properties such as typicality and phonological plausibility. All previous studies of deaf children categorized words as being either typical or atypical (using the terms regular or irregular). However, words are rarely wholly typical or atypical. Phonological errors were defined as misspellings that could be pronounced as the target words, but none of the studies reported any reference material or guidelines that were used to determine how the misspelling could be pronounced. One study simply stated that the experimenter deemed an error plausible if it could be recognized by sounding it out (Aaron et al., 1998). There are problems with this approach, because phonemes SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 8 can be represented by a variety of graphemes and not all may be obvious possibilities to an adult reader. One study (Harris & Moreno, 2004) did report using a second rater to confirm whether misspellings were phonologically acceptable. Overall, however, the definition of sound-based errors in these studies was vague at best. A final concern is that all of the above-cited studies of spelling in deaf children reported statistical analyses only by subjects; none investigated whether the results were consistent across items. By-subjects analyses collapse results across items, which eliminates the possibility of analyzing variance due to item characteristics. One might think that the addition of a traditional by-items analysis (in which results are collapsed across subjects) would solve this problem, but this is not necessarily the case. As described in more detail below, we used an alternative statistical analysis—mixed modeling—which has been shown to be a more robust tool for investigating potential sources of variance than separate by-subjects and by-items analyses (Baayen, 2004; Jaeger, 2008; Locker, Hoffman, & Bovaird, 2007). The current study aims to address the methodological concerns cited above and to expand the quantitative study of spelling of deaf children to those who use oral language and cochlear implants. We compared deaf implanted children to hearing peers on three properties: error rate, use of phonological strategies, and use of visual memorization strategies. Children spelled words in a picture spelling task, and these spellings were analyzed for spelling accuracy, factors such as typicality that may contribute to accuracy, phonological plausibility of errors, and rate of transposition errors. In addition, we asked whether the age at which children receive their implant makes a difference in spelling accuracy and plausibility of errors. To our knowledge, no studies to date SPELLING OF DEAF CHILDREN WITH COCHLEAR IMPLANTS 9 have investigated age at implant effects on spelling. Investigation of this factor is necessary because results in the literature are mixed on whether age at implantation affects performance on other aspects of literacy (Connor & Zwolan, 2004; Geers, 2003).