Visual Speech in the Head: The Effect of Cued-Speech on Rhyming, Remembering, and Spelling

Jacqueline Leybaert Brigitte Charlier Universite Libre de Bruxelles Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021

Deaf children rely mainly on lipreading to understand spo- memory during a short time, and when recognizing ken language. The phonological representations they develop printed words. To complete the picture, reliance on from the lipread signal are underspecified, leading to poor phonological representations is also evidenced when performances in all mental activities relying on such repre- sentations. To overcome these difficulties, systems have been hearing individuals spell unfamiliar words. The term designed that deliver entirely visually specified information "inner speech" will be used in this article to refer to about the phonological contrasts of spoken language. The pa- the phonological codes involved in these different ac- per explores the consequences of exposure to one of such sys- tivities, without assuming that it refers to one single tems, namely cued-speech (CS) on the development of pho- nological representations. Deaf children exposed early to CS phenomenon (see Besner, 1987). Experience with spo- at home show a reliance on inner speech for rhyming, remem- ken language thus leads hearing individuals to develop bering, and spelling similar to that displayed by hearing chil- phonological representations recruited at various levels dren but different from that of deaf children not exposed of the cognitive architecture. early to CS. We argue that the degree of specificity of phono- logical information delivered to the deaf children is more im- With regard to deaf persons, an important question portant than the modality though which they perceive speech is to what extent the development of inner speech de- for the development of phonological abilities. pends in a necessary sense upon acoustic experi- ence. Or, to formulate the problem differently: Can Phonological representations play an important role in profoundly, prelingually deaf children acquire inner the mental life of hearing persons. In particular, a set speech? To our knowledge, Dodd was the first re- of cognitive skills called the three R's (rhyming, re- searcher arguing theoretically against the view that membering, and reading) involve inner speech (Camp- acoustic information is a necessary condition for the bell, 1991). Hearing subjects typically get access to development of inner speech. She pointed out that lip- speech representations when judging whether two reading may constitute a primary input for deaf chil- words rhyme, when maintaining verbal information in dren to gain information about the phonological struc- ture of spoken language (Dodd & Hermelin, 1977; see

OUT research his been supported by grants from the Fondation Van also Dodd, 1987). Goethem-Brichant in 1983, 1986, and 1989 and by a grant of the Fonda- Dodd's view was elaborated in the context of a tion Houtnun in 1993. We thank Jesus Alegria who helped at various stages of our research and two anonymous reviewers for their comments change of theoretical perspective about lipreading. Up and correction of the English. Correspondence should be sent to Jacque- to the mid—70s, lipreading was considered as improv- line Leybaert, Laboratoire de Psychologic Experimentale, Universite Li- bre de Bruxdles, CP. 191, SO, Avenue Franklin Roosevelt, B-1050 Brus- ing speech perception in the case of a degraded acous- sels, Belgium (e-mail: [email protected].). tic signal, either because of an unfavorable signal-to- Copyright © 1996 Oxford University Press. CCC 1081-4159 noise ratio or because of impairment of the sensory Visual Speech in the Head 235 acoustic channel (Binnie, Montgomery, & Jackson, tions. While hearing subjects derive their representa- 1974; Erber, 1969; Sumby & Pollack, 1954). Empirical tions from audio-visual input, deaf children depend studies performed since the mid—70s have shown that more on the visual channel to sustain the development visual speech information is processed unavoidably of their phonological system. The major obstacle for by hearing listeners, even in the case of a nondegraded the development of phonological representations in acoustic signal. Seeing a face saying "ga" while hearing deaf children does not lie in the modality through the syllable "ba" leads to a compound percept, like which they perceive speech, but in the nature of the "da" (McGurk & MacDonald, 1976). The influence of perceived information. Lipreading, although providing visual speech on the perception of acoustic speech in- information about some phonological contrasts, does formation has also been demonstrated in serial recall not allow the perception of all of them. Lip movements Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 experiments (Campbell & Dodd, 1980; Spoehr & give some clues about place of articulation, but no clues Corin, 1978). These pioneer studies were later largely at all about features like nasality and voicing (Erber, confirmed (e.g., the collection of articles edited by 1979; Walden, Prosek, Montgomery, Scherr, & Jones, Dodd & Campbell, 1987). The main message of this 1977). For example, visual information allows the dis- bulk of studies is that the interaction between visual crimination of articulated in the front from and acoustic information occurs at a low, precategorical those articulated in the back of the mouth, but does not level of the processing of speech input. The audio- permit differentiation of consonants produced at the visual interactions further suggest that auditory and same articulation point (e.g., /b/, /p/, and /m/ or the visual speech information share some common stage of /D/ and /5/. Through lipreading, then, deaf phonological processing (Summerfield, 1987). Devel- children only have access to phonetically underspeci1 opmental data are also worth considering in this dis- fied information. The representations they derive from cussion. The integration of auditory and visual speech this input are also necessarily underspecified. information seems already established during the first The lack of well-formed phonological representa- weeks of life. Five- to six-month-old infants prefer tions adversely affects all higher-order functions based to look to faces pronouncing the same syllable as that on such representations. Deaf children generally have delivered acoustically than to faces pronouncing a dif- poor metaphonological abilities (Hanson & McGarr, ferent syllable (Kuhl & Meltzoff, 1982; MacKain, 1987; Campbell & Wright, 1988), reduced memory Studdert-Kennedy, Spieker, & Stern, 1983; Walton & span for linguistic stimuli (Conrad, 1979; Hanson, Bower, 1993). 1982), low reading and spelling levels (Conrad, 1979) The research on audio-visual speech thus gives and deficits in the retention and interpretation of free weight to Dodd's assumption that lipread information morphemes and articles (Taeschner, Devescovi, & may be dealt with similarly to acoustic information by Volterra, 1988; Volterra & Bates, 1989). Since these the brain. This conclusion leads to the supposition that deficits are observed even in native users, visual information about speech contributes to the for- they cannot be ascribed to the lack of linguistic experi- mation of phonological representations in hearing sub- ence per se, but rather to the lack of complete, well- jects. Mills's (1987) work about the phonological devel- specified perception of the phonetic contrasts of spo- opment of blind children illustrates this point nicely. ken language. For seeing babies, but not for blind ones, bilabials are easy to imitate, because the articulatory gesture is Cued-Speech: A System Providing Visually Well- clearly visible. Mills found that blind children do not show the predominance of labials in their first oral pro- Specified Speech Information ductions, which characterize the early productions of To overcome these deficiencies, several systems have seeing babies. been created aimed at disambiguating lipreading by The consequence of bimodal (i.e., audio-visual) adding visual information carried by the hands. The speech perception is that deafness per se does not pre- most widespread among them is cued-speech (CS), de- clude the development of phonological representa- vised by Cornett (1967). In CS, the speaker holds one 236 Journal of Deaf Studies and 1:4 Fall 1996 hand near the mouth while speaking to complement The consonants The vowels lipreading with a manual cue. A cue is made of two pa- l J P p (pas) A A a(papa) M rameters: hand shape and place of execution around [d] d (dis) au(peau) [0] the mouth. The hand can adopt several shapes (eight e (petit) . j in the French version of CS) at different positions , k(cou) around the mouth (five in French). disam- ivJ v(vu) A A [z] z(maison) biguate the consonants and hand positions the vowels. in (pain) [?] eu (deux) ^j Shapes and places are assigned to groups of two or ts] s (Sur) three consonants (or vowels) in such a way that pho- [r] < (nt) nemes easy to discriminate by lipreading share the b(bon) e(sel) te] Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 same or place while items that are difficult n(non) ou (loup) [u] to discriminate belong to different groups. For ex- w (cuistne) o (porte) ^ ample, a particular handshape is shared by /p, d, 3/, [m] m (maman) A\ another one by /b, n/ and so on. For the vowels, one [t] t (toot) [i] [f] ' (feu) • 1(nz) place is shared by /i, 5, a/, a second one by /a, o, 3 /, on(mon) [3] and so on (see Figure 1). Each time the speaker pro- an (sang) jgj duces a cue (a particular handshape at a specific place) [J. I doup) A A while pronouncing a CV syllable, he is visually giving [JJ ch(chat) [w] w(oui, quo) e(b<5be) te] unambiguous information about this syllable and its . , gn(cogne) u(su) [y] un (brun) j~-j . Syllabic structures like VC, CCV, CVC re- quire additional cues to reveal the supplementary pho- [g] g(guO nemes. CS has been designed in an oralist perspective. Hands alone do not give any useful information: No \*\ y (fine) handshape nor hand place can be interpreted without lnJ ng (parking) taking the lips into account. Although the system may appear somewhat complex and artificial to uninitiated • and any not preceded by a conjonant (arrete) persons, it is easy to learn and, with some practice, nor- • • and any isolated (sec, prof) or followed by a silent e (lune) mally hearing adults succeed at using it while speaking at a natural rate. Figure 1 French version of cued-speech. Reproduced with authorization of the Belgian LPC-Association. Initially, CS was aimed at improving the speech re- ception skills of deaf children. This improvement has been evaluated both for English- and French-speaking children. Nicholls and Ling (1982) studied a group of after CS will be used instead of Langage Parle Com- profoundly deaf children taught at school with CS plete to describe the experiments run in French.) The since at least three years of age. They found that the subjects (mean age: 10;7 years; range: 5;11—16; 1 years) speech reception scores of these children increased were divided in two groups: deaf children exposed to from about 30% for both syllables and words in the lip- CS both at home since an early age and at school (the reading condition, to more than 80% in the lipreading home group) and deaf children exposed to CS only at + cues condition. The authors underlined that the school (the school group). All children showed better children's average scores in the lipreading + cues con- understanding for sentences presented in the lipread- dition are within the range of normal hearing listeners' ing + cues condition than in the lipreading alone con- reception of similar materials through audition. Perier, dition. However, the advantage provided by CS was Charlier, Hage, and Alegria (1988; see also Alegria, larger for the home group than for the school group. Leybaert, Charlier, & Hage, 1992) investigated the These two studies thus strongly suggest that the addi- effect of exposure to the French version of CS, called tion of manual cues has the expected effect of disam- Langage Parle Complete. (For the sake of clarity, here- biguating lipread information. This positive effect is Visual Speech in the Head 237

particularly significant for children whose parents the global similarity between words. The first-graders use CS. likely analyze the word constituents and base their de- Given the fact that CS allows deaf children to per- cisions upon the identity of phonological segments ceive visually the same amount of phonetic information rather than on global similarity. as that perceived through audition by hearing children, Are the profoundly and congenitally deaf children a strong effect of exposure to CS on the acquisition of able to detect that two words rhyme as do hearing chil- phonological representations might be expected. The dren? In a pioneer study, Dodd and Hermelin (1977) general question we were interested in was whether found that 12- to 14-year-old deaf children are able to deaf children educated with CS rely on phonological judge above chance level whether pairs of lipread words representations in a different way than do deaf children rhyme. They argued that deaf children's notion of

who are not exposed to CS. We first tested whether rhyme is based on visual lipread similarity between Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 deaf children derive accurate phonological representa- spoken words rather than on articulatory cues. Al- tions from CS, suitable for rhyme decision tasks. Next, though of primary interest, this experiment did not de- we investigated whether these children rely on inner termine whether deaf children distinguish between the speech when involved in an immediate short-term notion of rhyme and the notion of lipread similarity. memory task. Finally, we tested whether they use accu- Indeed, all rhyming words automatically end with the rate -grapheme correspondences for word same lipread image (e.g., in French LIT /H/-NID spelling. The results obtained in these three situations /ni/), but all pairs of words ending with identical li- are summarized below. pread images do not necessarily rhyme (e.g., in French LIT /li/- NEZ /ne/). An important question then is to determine whether deaf people do discriminate be- Rhyming tween lipread similarity and rhyme. Unfortunately, this One way to explore the nature and the quality of pho- question has not been investigated empirically up to nological representations consists of evaluating the ca- now. In all studies that have followed Dodd and Her- pacity to judge whether two words rhyme or to gener- melin's, rhyming is confounded with lipread similarity. ate rhymes in response to a word target. These tasks Studies conducted after Dodd and Hermelin's require access to lexical phonological representations (1977) have investigated the effect of another variable and comparison of the words' final vowel (or vowel +' on rhyming judgment in deaf subjects, spelling similar- coda). ity (Campbell & Wright, 1988; Hanson & Fowler, 1987; If a hearing child is asked whether two words Hanson & MacGarr, 1988). Deaf youngsters, educated rhyme, what will determine his or her answer? First of orally or with sign language, generally achieve reason- all, the child's age. Under three years, children will ably good performance when the two items rhyme and probably respond at chance level; older, say around are spelled similarly (e.g., BAT/HAT) but fall nearly five, they will probably respond correctly. The emer- to chance level when the items are spelled differently gence of the notion of rhyme is correlated with the (e.g., HAIR/BEAR). Surprisingly enough, the deaf quality of the child's oral productions (Webster & students were sensitive to word spelling not only when Plante, 1995). The degree of phonological similarity written words were used as targets, but also when pic- between the words is a second factor determining per- tures were used. This differentiates them from their formance. Kindergartners are prone to judge assonant hearing controls, who showed a slight effect of spelling words like MASSA and LACA (in Portuguese) as congruency, but only for written targets. Campbell and rhyming as they do for SOLA and MOLA; first- Wright sugggested that deaf people spontaneously rely graders, however, correctly categorize MASSA and more on spelling than hearing people, because the pho- LACA as nonrhyming and SOLA and MOLA as nological representations derived from lipreading are rhyming (Cardoso-Martins, 1994). Children of differ- *not clearly enough specified to support rhyming judg- ent reading-levels seem thus to rely on different cues ment. However, reliance on spelling information varies to take their decision. The kindergartners probably use among deaf individuals. In the three studies mentioned 238 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996 above, the deaf youngsters who had a better speech in- This forces subjects to make comparisons between lexi- telligibility and those who were better readers reached cal phonological representations. New groups of hear- higher scores in rhyming judgments about word pairs ing (n = 12; mean age: 8;7 years), CS+ (n = 16; mean spelled differently. age: 10; 1 years), CS- (n = 18; mean age: 12;7 years) If the underspecified nature of lipreading is re- were recruited. In addition, new groups were included: sponsible for the poor quality of the phonological rep- children educated in spoken language only (the oral resentations of deaf individuals, a system disambiguat- group; n = 29; mean age: 13;3 years) and children who ing lipreading, like CS, should have a positive effect used sign language, who were divided up into two both on the quality of speech representations and on groups. The children of the SL+ group (n = 12; mean rhyming skills. Children exposed to CS would be able age: 10;4 years) were all native signers: they had deaf

to judge that the French words LIT and NID are parents and sign was used at home. At school they re- Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 rhyming, even if the rhyme is not spelled identically, ceived an oral education mixed with sign language. and that LIT and NEZ do not rhyme although sharing Subjects of the SL- group (n = 20; mean age: 10; 1 the same lipread image. years) attended the same schools as the SL+ group, This hypothesis was investigated in a series of ex- but had hearing parents and were educated in spoken periments (Charlier & Leybaert, Hearing by eye: The language at home. In order to tap the processes used rhyming skills of deaf children educated with cued- by the deaf children, two experimental variables were speech, submitted). The first one involved rhyming manipulated. The first one was spelling similarity. In judgment about pairs of written words. In addition to the rhyming pairs, the pictures represented either the hearing subjects, three groups of deaf subjects, words with similar spelling (RO+: e.g., TRAIN matched for grade level with the hearing children, were /tr£ /- MAIN /mi /) or words with different spelling tested. The cued-speech+ group (CS + ) included 15 (RO-: e.g., LIT /li/ - RIZ ri/). The second variable children (mean age: 11;2 years) having the French ver- was lipread similarity. Up to now, studies showing sion of CS at home, meaning that at least one of their rhyme sensitivity in deaf subjects have confounded parents used it in daily communication. CS had been phonological and lipread similarity. To distinguish be- introduced to this group at a mean age of three years. tween these two factors, performance for nonrhyming Most of these children were also provided with CS at pairs in which the names of the pictures have similar school. The CS-group (CS-) included 20 children lipread images when spoken (NRLR: e.g., LIT /li/ - (mean age: 13;0 years) exposed to CS only in their NEZ /ne/) was compared with performance for non- school environment, since the mean age of six years. rhyming pairs with different lipread images (NRC: e.g., The oral group included 28 children (mean age: 13;5 LUNE /lyn/ - FLEUR /floer/). Subjects who believe years) educated exclusively with the oral/aural method that rhyme and similar lipread images are the same both at home and at school. The results showed no thing will have a tendency to judge erroneously that difference between the groups of subjects: All children NRLR pairs rhyme. Subjects who have developped were able to make rhyme judgment with the same accu- well-specified phonological representations would not racy and all were influenced by spelling similarity. We be affected by lipread similarity. concluded that this situation did not test access to lexi- Children were presented with pairs of drawings. cal phonological representations for two reasons. First, For each pair, they had to say whether the pictures the presentation of the written words may force all sub- "were friends" or not. In a pretest, the notion of jects to process the word spelling. Second, subjects may "friend" (i.e., rhyme) was assessed for each child, and derive the word prononciation from the word spelling only children who show a comprehension of this no- itself instead of recovering the phonological informa- tion were included as subjects. tion from the mental lexicon. The results (see Table 1) revealed interesting In a second experiment pairs of pictures were used. differences between the groups. A significant effect of The reason for using pictures was that they do not pro- experimental conditions was obtained in all deaf vide any direct information about word pronunciation. groups, but not for the hearing group. Post-hoc con- Visual Speech in the Head 239

Table 1 Mean accuracy of deaf CS+, deaf CS —, deaf specified representations. This is very likely the case oral, deaf SL+, deaf SL—, and preliterate deaf CS+ and for children educated orally or with sign language, but hearing subjects for rhyme judgment about pictures not for the CS+ children. A second possible locus of RO RNO NRLR NRC the lipread similarity effect is the articulatory loop. Readers After getting access to the word representations, sub- DeafCS+ 97.4 94.4 93.8. 100 jects hold them in working memory in order to DeafCS- 86.4 73.9 68.3 95.6 compare them. They generally pronounced the items, Deaforal 92.2 71.6 73.8 96.6 DeafSL+ 89.3 58.2 74.7 96.7 aloud or silently, in order to compare them. Stimuli DeafSL- 82.9 66.6 58.6 92.0 with similar lipread images also share some articulatory Hearing 95.8 97.0 • 97.7 99.2 features. This articulatory similarity may induce a ten- Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 Prereaders dency to say that the two items rhyme. This tendency CS+ 100 94.4 89.0 100 may be stronger in deaf than in hearing subjects, be- Hearing 91.3 91.0 78.6 94.0 cause the former do not have acoustic images that can RO Rhyming pairs orthographically jimilir, RNO rhyming pairs ortho- confirm or deny the articulatory clue. Hearing subjects graphically different; NRLR: nonrhyming pairs with simitir lip-read im- ages; NRC; nonrhyming control pairs. Deaf CS+: Children educated may be supposed to access, through articulation, an with cued-speech both at home and at school; Deaf CS—: Children ex- acoustic image of the word, in the same way as they can posed to cued-speech at school only; Deaf oral: Children educated with spoken language only; Deaf SL+: Children educated with sign language evoke the sound of a wave, of a music instrument, or of both at home and at school; Deaf SL—: Children educated with spoken a bird singing. Profoundly deaf people probably do not language at home and with sign language at school (adapted from Char- Her & Leybaert, submitted). have the same capacity to evoke, mentally, the sound of a word (or of the wave, etc), although they may know a lot of things about the sounds related to different trasts revealed that all groups of deaf subjects except events (Marschark, 1995). The importance of acoustic the CS+ one had a lower performance for RO-pairs images is supported by the study of deafened adults than for RO+ pairs. The difference between RO+ and who show depressed performance for rhyming judg- RO-conditions reached 12.5% for the CS- group, ment about word pairs spelled differently (Lyxell, 20.6% for the oral group, 31.2% for the SL+ group, Arlinger, Andersson, Bredberg, Harder, & Ronnberg, 16.3% for the SL- group, and only 3.0% for the CS + 1996). It is thus possible that the slight effect of lipread group. This orthographic effect confirms that deaf >: similarity observed in the CS+ gToup is located at the children educated orally or with sign language use their working memory stage of comparison, while the larger knowledge of spelling to support rhyme judgment, effect displayed by the other groups is located both at probably because their phonological representations the representational and at the comparison level. are underspecified. This seems also to be the case for A further advantage of using pictures as test mate- the CS- children. By contrast, CS+ children and hear- rial is that it allows the evaluation of very young chil- ing children do not show any sensitivity to word spell- dren. An attempt to do so was made with deaf preread- ing. This differentiates them from the other groups of ers, which succeeded only in the case of CS+ children. deaf children tested in the experiment as well as from The fact that only CS+ children were able to under- the deaf subjects evaluated in the literature. stand the notion of rhyme is already interesting. In- The NRLR condition induced more errors than deed, although the development of metalinguistic abili- the NRC condition in all groups of deaf children, the ties is supposed to follow the acquisition of reading and CS+ one included. However, the effect of lipread simi- spelling in children educated orally or with sign lan- larity was much lower in the CS+ group (in which the guage (Campbell, 1991), it is clearly not the case for difference between NRLR and NRC equals 6%) than CS+ children, in which sensitivity to rhyme develops in the other groups (CS-: 27%; oral: 28%; SL+:22%; before literacy. SL-: 33%). We thought that the NRLR condition may A group of CS+ prereaders was thus compared to mislead subjects at two distinct stages of processing. a control group of hearing children. Both groups First, it is possible that subjects get access to under- showed the same pattern of results: no effect of spell- 240 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996 ing, but a significant effect of lipread similarity. Remembering This effect is compatible with the two interpretations sketched above. Possibly the representations of pre- Let us suppose somebody showed you a phone number readers are, to a certain extent, different from those written on a paper; you cannot write it down. What will of readers (e.g., Studdert-Kennedy, 1986). Or perhaps you do in order to retain it? As a hearing person, you prereaders are mislead by articulatory similarity, be- will very probably repeat it to yourself, again and again. cause they detect rhyme on the basis of a global phono- You may repeat it aloud, or mentally, without produc- logical similarity rather than on the basis of detailed ing any sound. This particular kind of memory will be segmental information (Treiman & Breaux, 1982; discussed in this section, that is, the ability to retain Cardoso-Martins, 1994). during a brief interval a series of verbal items (conso- The effect of lipread/articulatory similarity and of nants, words, pictures) in order to repeat them in the Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 orthographic similarity cannot be interpreted as show- same order in which they were presented. ing that subjects rely only on one of these clues. If this Conrad (1962, 1964) demonstrated that when hear- had been the case, subjects would have reached a 0% ing people memorize series of consonants presented level of correct responses in the critical condition. This visually, they show confusions that have an acoustic was never observed. These effects show only a ten- rather than a visual nature. For example, hearing sub- dency to prefer lipread/articulatory and/or spelling jects will have a tendency to confuse B, D, G, but not information for rhyme judgment in particular tasks. B and F. These confusions indicate that verbal infor- To conclude up to this point: Exposure to visual mation is held in short-term memory in a speech- speech information that specifies all phonological con- related form, even if it is presented visually and if the trasts leads deaf children to internalize representations recall is to be written. that are entirely specified. The notion of rhyme seems Baddeley and his collaborators (Baddeley & Hitch, to emerge as a natural consequence of the acquisition 1974; Baddeley, 1986) have elaborated a model of "ver- of such representations, independently of the modality bal working memory." There is no room here to go into (visual versus acoustic) through which language has all the details of this model but only to point to some been perceived. Although the rhyming ability of deaf of its characteristics. Working memory involves an children educated with CS seems to be well estab- auditory-verbal component, the "articulatory loop." lished, the detailed processes they use to perform such This system consists of two elements: a phonological tasks remain to be clarified. Hearing children have been input store within which memory traces fade if not re- found to use the articulatory loop (see next section) freshed, and a rehearsal process that serves to refresh when performing rhyme judgment; their rhyming the memory traces by means of subvocal rehearsal. The scores decrease significantly under articulatory sup- items presented auditorily enter directly into the pho- pression (Arthur, Hitch, & Halliday, 1994). The CS+ nological store. The items presented visually (like writ- deaf children are also likely to use an articulatory loop. ten words or pictures) are fed into the phonological There is, however, one main difference between the two store provided that they are encoded phonologically populations, namely the absence of acoustic images in and subvocalized through the rehearsal process. deaf subjects. Most of them say, during the rhyming Two main effects have been considered as the sig- judgment experiments: "It's difficult for me, because nature of the articulatory loop: phonological similarity I'm deaf?' What kind of knowledge do they use, then, and length effects. The phonological similarity effect to decide that two items rhyme? Is there a "visual im- refers to the fact that rhyming items are harder to age," made up of lipread + cues information, that is memorize than nonrhyming ones. The source of this equivalent to the acoustic image? Or do they access a effect is located at the phonological store, where the more abstract knowledge? Further experimental work, representations of phonologically similar words are ac- involving articulatory suppression, is needed to address tivated at the same time. The candidate representation these questions. for a particular serial recall position is confused with Visual Speech in the Head 241 other, phonologically similar, representations. Main- the non-rhyming set share the same visual envelope.) taining a sequence of such representations in correct Some of these deaf teenagers made more errors in the order is therefore more difficult and error-prone than rhyming than in the nonrhyming set, evidencing for maintaining a sequence of dissimilar representations in the use of a phonological code; the magnitude of correct order. the rhyming effect was, however, lower in deaf than in The word length effect refers to the fact that longer hearing subjects. Other deaf youngsters made more er- words are harder to retain than shorter words. Follow- rors on the nonrhyming visually similar set than on the ing the standard view, this effect is a direct consequence rhyming visually dissimilar set, showing evidence for of the articulatory nature of rehearsal. Longer words the use of a visual code. The use of a phonological code take longer to articulate, and longer to rehearse, than was strongly related to hearing loss; severely deaf shorter words. Consequently, their phonological store youngsters develop internal speech more often than Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 traces can be refreshed less frequently than is possible profoundly deaf ones. Other studies (Hanson, 1982; for short words (but see Brown & Hulme, 1995, for a Hanson, Liberman, & Shankweiler, 1984; Waters & model simulating the length effect without any re- Doehring, 1990; Wandel, 1990) have confirmed that hearsal process). some of the deaf youngsters exposed only to spoken Empirical studies have demonstrated that the artic- language or exposed to sign language rely on inner ulatory loop is involved both in short-term memory speech to memorize series of written words. and in rhyme judgment, in adults (Besner, Davies, & The use of phonological coding for memorizing set Daniels, 1981) as well as in children (Arthur, Hitch, & of pictures has been investigated by Campbell and Halliday, 1994). There is, however, one main difference Wright (1990). In one experiment, children had to- between the two tasks. Whereas in rhyme judgment the memorize pairs of pictures that either rhyme (e.g., articulatory loop is necessarily requested to perform CAT-HAT or TIE-FLY), were semantically related a phonologically-based decision, it is spontaneously (e.g., KNIFE-FORK), or were randomly associated used in a short-term memory task. (e.g., BELL-LEAF). The semantic relationship was Do deaf people use their articulatory loop sponta- used very efficiently both by hearing and deaf subjects. neously when memorizing a series of written words or The rhyme cue, however, was used only by the hearing pictures? Conrad (1970) was the first researcher to in- subjects. At first sight, this finding may be surprising' vestigate this issue. He showed that the errors of some in regard to the studies showing that at least some deaf- deaf children exposed only to spoken language were people are sensitive to rhyme when memorizing a series phonologically based (e.g., the B/D confusion), while of written words. A critical difference, underlined by for others children they were visually motivated (e.g., the authors, is the nature of the material: For the deaf, the N/V/X confusion). The quality of the deaf sub- written words, but not pictures, may have a favored jects' speech was correlated with the type of errors connection to their phonological form (see also Ley- made. Children with good oral skills made mainly pho- baert, Alegria, & Fonck, 1983; Leybaert & Alegria, nological confusions whereas children with poor oral 1993, for Stroop-like evidence about the strong rela- skills made essentially visual confusions. These data tionship between written and spoken words in deaf were replicated in a large-scale study involving nearly people). all deaf school-leavers (age: 15;6 to 16;6) of England In a second experiment, the authors examined the and Wales. In this study, the ability to retain series of influence of word length on memory for series of pic- written items extracted from a rhyming set (DO, FEW, tures. Two recall modes were tested: reconstruction of BLUE, WHO, ZOO, TRUE, SCREW, THROUGH) picture pairs and naming. Deaf teenagers were com- was compared to the ability to retain a series extracted pared to two groups of hearing controls: younger chil- from a visually similar, nonrhyming set (BARE, dren, and chronologically age-matched children. Deaf BEAN, DOOR, FURS, HAVE, HOME, FARM, teenagers differed from the hearing group matched for LANE). (When presented in lower case, the words of chronological age, but had the same pattern of perfor- 242 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996

Table 2 Mean accuracy of deaf CS+, deaf CS—, deaf 9; 11 years), and hearing children (« = 30; mean age: oral, and hearing children for memorizing series of pictures 9;4 years) were tested. The performance for the mono- Monosyllabic Rhyming Polysyllabic syllabic and nonrhyming condition did not differ sig- DeafCS+ 63.8 48.9 49.7 nificantly between groups, indicating that all children DeafCS- 54.5 57.1 57.6 were performing at an equal level of difficulty (see Deaforal 45.9 45.3 51.5 Table 2). The experimental variables, however, affected Hearing 63.6 47.4 57.6 the groups of children very differently. First, both the Monosyllabic: monosyllabic nonrbyming words; Rhyming: monosyllabic rhyming words; Polysyllabic poryiyllabic nonrhyming word*. De»f CS+, CS+ and the hearing groups displayed a large rhyming De«f CS-: see legend of Table 1. Adapted from Charlier (1994). effect, while the CS- and the oral groups display no sensitivity to phonological similarity. The difference of mance as the younger hearing group. Whereas the accuracy between the control and the rhyming condi- Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 older hearing subjects show a length effect, both in tion reached 6.3% for the hearing and 5.0% for the the naming and in the picture reconstruction modes, CS+ group, but -2.6% for the CS- and .6% for the the deaf and the younger hearing children displayed a oral groups. Second, both the CS+ and the hearing length effect only for naming. There was dius no clear groups exhibited a clear effect of spoken length, indication that the deaf and the younger hearing chil- whereas the CS- and the oral groups did not display dren spontaneously relied on inner speech when mem- any sensitivity to word length. The difference between orizing series of pictures. Indeed, the length effect in accuracy in the control and die polysyllabic conditions the naming condition can be ascribed to the rehearsal, reached 6.1% for the hearing and 14.1% for the CS + , but also to the naming process itself. but -3.0% for the CS- and —5.6% for the oral group. It is thus worthwhile to ask whether children edu- The results of the CS- children suggested that their cated from an early age with CS spontaneously rely on recall of series of pictures was not supported by phono- the articulatory loop in a serial recall task involving pic- logical representations. This confirms Campbell and tures. If so, they would exhibit the phonological simi- Wright's (1990) results. The data of the CS+ contrast larity effect and the length effect. To this end, Charlier with those of these two groups. The phonological simi- (1994) compared the performance of CS+ children, larity effect and the length effect indicated that the CS- children, and hearing children. The experiment CS+ children spontaneously used the articulatory included three experimental conditions: monosyllabic loop, based on phonological representations. Access to nonrhyming words (e.g., CHAT //a/, BANC /ba)/, phonological representations seems to be as unavoid- FEU /fe/), monosyllabic rhyming words (e.g., DE able in this situation for CS+ subjects as it is for hear- /de/, FEE /fe/, NEZ /ne/), and polysyllabic non- ing subjects. Indeed, this access limits their perfor- rhyming words (e.g., HELICOPTERE /elikopter/, mance in both rhyming and polysyllabic conditions. It CROCODILE /krokDdil/, TELEPHONE /telefDn/). is also interesting to note that the use of inner speech Pictures of these words were drawn on cards that seems to enhance their memory span. Indeed, the num- served to present the stimuli to the subjects. The num- ber of items used to run the experiment was higher for ber of items in a series was determined individually on the CS+ and the hearing groups (5.17 in average) than the basis of a pretest so that the subjects achieved for the CS- (4.86) and 4.67 for the oral group. around 60% of correct responses with nonrhyming monosyllabic words. The experimenter showed the de- Reading and Spelling termined number of cards to the subject, at the rate of The development of well-specified phonological repre- one per second. Then he turned over the cards. After sentations in deaf children may also have important the last card was turned over, the children had to re- consequences for their acquisition of reading and spell- construct the series of pictures in the order in which ing. It is generally assumed that hearing subjects recog- they were presented. nize printed words in two ways. Familiar words may be Groups of CS+ (« = 12; mean age: 8;8 years), CS- identified very rapidly, on the basis of the activation of (JI = 14; mean age: 10;9 years), oral (n = 11; mean age: an orthographic representation stored in long-term Visual Speech in the Head 243

memory. Unfamiliar words, for which no orthographic 1990, for negative results). In our study (Leybaert & representations are available, may be identified through Alegria, 1995), the regularity effect was less marked in the application of grapheme-phoneme conversion deaf than in hearing subjects, not because the deaf were rules, leading to the assembly of the word's phonologi- better with irregular words, but because they were cal form. The sound of the word then gives access to its poorer with regular words. The deaf subjects thus took meaning. Similar procedures are available for spelling. less advantage than the hearing subjects of the exis- Hearing children may use their lexical knowledge tence of regularities between phoneme and grapheme. about how a particular word is spelled, or they may Another indicator of the use of phoneme-to- construct a spelling output by applying phoneme-to- grapheme regularities is the percentage of phonologi- grapheme rules to the word's phonological form. The- cally accurate errors. When a hearing child ignores how

orists have underlined that the use of sublexical corre- to spell a particular word, he produces, most of the Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 spondences between graphemes and speech segments time, a spelling compatible with the word pronuncia- plays an important role at the beginning of reading and tion (e.g., BRANE for BRAIN). This kind of mis- spelling acquisition (Frith, 1985; Seymour & McGre- spelling reveals that the child starts from an accurate gor, 1984). phonological representation to which he applies One may wonder to what extent deaf children use phoneme-to-grapheme rules. Deaf people also make their phonological representations for reading and such misspellings, but less than their hearing peers. For spelling. In other words, are the deaf limited to the ex- example, in our study, the percentage of phonologically clusive use of lexical word-specific knowledge, available accurate errors was 21.8% for young deaf children only for familiar words, or are they also able to use su- against 90.6% for young hearing children. blexical relationships between phoneme and graph- These observations may be interpreted in two ways. eme? In the present article, we will limit ourselves to First, deaf children may rely less than the hearing on the discussion of the procedures used for spelling. phoneme-to-gTapheme rules. Second, they may rely on The typical method for determining whether deaf phoneme-to-grapheme rules, but start from inaccurate children use phoneme-to-grapheme knowledge con- phonological representations. Evidence supporting sists in comparing their performance for regular and both hypotheses was found in Leybaert and Alegria's irregular words. Regular words are those for which study. The fact that deaf children start from inaccurate,

the correct spelling can be straightforwardly derived underspecified phonological representations is exemT from their pronunciation, by applying phoneme-to- plified by the systematic part of their phonologically grapheme rules. Irregular words are those for which inaccurate errors. For example, they often reduced the correct spelling violates the phoneme-to-grapheme consonantic clusters by deleting the liquid, misspelling rules and must thus be retained by rote. The rationale ARMOIRE as AMOIRE, CARTABLE as CATABLE. of the comparison is the following: An advantage for They substituted voice, voiceless, and nasal conso- regular over irregular words indicates that children nants, misspelling OUVERT as OUFERT, AVION as rely on phoneme-grapheme rules for spelling. On the AFEO. These errors indicate that the deaf apply con- other hand, if children rely exclusively on word- version rules to inaccurate representations (see also specific knowledge, no difference is expected in the Burden & Campbell, 1994, for similar observations). spelling of the two types of words. On the other hand, lower reliance on phoneme- In several studies, deaf youngsters have been found grapheme rules is indicated by the occurrence of mis- to show a regularity effect, that is, better performance spellings consisting of keeping the letters of a word, but for regular than for irregular words. This is true not in a wrong order (e.g., SPTE instead of SEPT). These only for deaf students who have achieved an excep- misspellings occurred only in the deaf group. They tional reading level (Hanson, Shankweiler, & Fischer, probably result from a tendency to retain the word 1983), but also for children with a reading age of ap- spelling by rote, without reference to its pronunciation. proximately nine years (Burden & Campbell, 1994; To summarize up to this point, although deaf Leybaert & Alegria, 1995; but see Waters & Doehring, youngsters do use phoneme-to-grapheme mappings, 244 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996

Table 3 Characteristics of the subjects of the spelling experiment Hearing DeafCS+ DeafCS- n 28 15 13 Mean age 9;2 9;8 12;2 % Correct responses spelling test 85.1 86.0 75.0 Score at the reading test 23.6 23.3 10.9 Mean school years (in years; months) 3:1 3:4 5:5 they are unable to take a full advantage of the transpar- with the CS- children who made, besides phonologi- ency of alphabetic orthographies by the underspecified cally accurate misspellings (30.5%), a large amount

nature of their phonological representations. Exposure (32.1%) of phonologically inaccurate misspellings Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 to visual systems that entirely specify the phonological (e.g., COPAT for COPAIN, JOU or ZUS for JUS). information, like CS, may thus have a strong impact at The error analysis thus confirms that CS+ children this level. Deaf children who have internalized accurate start from accurate phonological representations for phonological representations may benefit more from spelling. the regularities between phonemes and graphemes Given that CS+ children seem to access accurate than do other deaf children not using CS. phonological representations, it is possible that they The effect of exposure to CS on the development develop spelling along lines similar to those of their of spelling has been investigated in an experiment still hearing peers. The development of two aspects of their in progress (Leybaert, The effect of cued-speech on spelling is summarized below: spelling of words con- the development of deaf children's spelling, in prep.). taining dominant and nondominant phoneme- Again, three groups of children were compared: deaf grapheme transcriptions and spelling of words con- children educated with CS at home (the CS+ group), taining consonant clusters. deaf children exposed to CS only at school (the CS- group), and hearing children (see Table 3). These chil- Phoneme-to-grapheme dominance. Although French or- dren were assessed on a spelling test, which consisted thography is quite consistent for reading, this is not the of writing the names of pictures representing simple case for spelling; most of the phonemes can be spelled concrete words. They also passed a reading sentence in different ways. They have a statistically dominant comprehension test. The preliminary results show that transcription (e.g., /s/ at the beginning of words, and the CS+ group achieved the same spelling level and before e and i is usually written as the letter S) and one the same reading level as the hearing controls at ap- or several nondominant transcriptions (i.e., the pho- proximately the same chronological age. This contrasts neme /s/ is also transcribed, but less frequently, by the with the CS- children who, although older than their letter C). The interest in using such variables experi- hearing peers, were delayed in their spelling and even mentally is the following. The spelling of the words more in their reading achievement. These descriptive containing a dominant transcription constitutes a good data already indicate a positive effect of exposure to CS evaluation of the degree of mastery of the frequent at home on the development of literacy. phoneme-grapheme correspondences. On the other What interests us primarily, however, is the effect hand, the percentage of correct spelling for words con- of exposure to CS on the procedures used for spelling taining nondominant transcriptions is an indicator of and, more precisely, on the use of phoneme-to- the development of the orthographic lexicon (Ale- grapheme correspondences. The type of errors consti- gria & Mousty, 1994). tutes one indicator of this. In the hearing group, 92.9% The children were thus asked to spell words con- of the errors were phonologicaJly accurate (eg., taining either dominant (eg., SEL, SECRET) or non- SIGARETTE for CIGARETTE; TRIN for TRAIN). dominant transcriptions (e.g., CTET., CIGARETTE). As expected, this was also the case for most of the mis- The words were varied in frequency. The score was the spellings of the CS+ children (74.3%). This contrasts percentage of correct spelling for the grapheme under Visual Speech in the Head 245

Table 4 Percentage of correct responses in spelling words containing dominant and nondominant phoneme-grapheme transcriptions Frequent Rare Dominant Nondominant Dominant Nondominant Hearing 87.2 51.5 82.8 24.5 DeafCS+ 74.1 50.9 89.8 38.0 DeafCS- 96.8 82.4 69.4 51.9 Deaf CS+, Deaf CS-: see legend of Table 1. From Leybaert (in prepi).

Table 5 Percentage of correct responses for spelling words containing consonant

clusters and control words Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 Frequent Rare dusters Controls Ousters Controls Hearing 94.3 100 90.5 100 DeafCS + 89.0 100 85.7 100 DeafCS- 80.5 96.7 55.6 80.6 Deaf CS+, Deaf CS-: see legend of Table 1. From Leybaert (in prep.). investigation. For example, if CIEL was spelled with a Their whole pattern was more determined by word fre- C, a correct response was attributed, independently of quency than by the use of phoneme-to-grapheme dom- the way the other phonemes were spelled. inance, which indicates a main reliance on lexical The CS + children behaved similarly to the hearing knowledge. controls (see Table 4). (Because this experiment is un- finished, the results described here are only prelimi- Consonant clusters. Bruck and Treiman (1990) have nary, and we limit ourselves to describing the patterns, shown that consonant clusters like BR, TR, GR are without giving the statistics.) At the beginning of difficult to spell for young hearing children who fre- learning to spell, they spell correctly words containing quently omit the liquid in their written productions. the dominant rule, irrespective of word frequency. Our previous study of the spelling of deaf children ed- They also made many errors for words including a ucated orally (Leybaert & Alegria, 1995) showed that nondominant rule, which were usually spelled with the they have persistent difficulties in spelling consonant dominant transcription (e.g., SIEL instead of CIEL). clusters, probably because the two consonants do not Their pattern of performance, at this first stage of have distinct images on the lips. The delivery of precise learning to spell, is clearly more influenced by the phonological information by CS may play a positive dominance of the transcription than by word fre- role on spelling such syllabic structures, because each quency, which indicates a main reliance on phoneme- consonant is indicated by a manual cue. In the same to^grapheme mapping rules. study (Leybaert, in prep.), the spelling of R and L ei- The results of the CS- group contrasted with those ther in a consonant cluster (e.g., TRAIN, FLEUR, of the other two groups (see Table 4). At the first level TROMPETTE) or at the beginning of a syllable (e.g., of spelling achievement, the CS- children had higher LAPIN, ROUGE, REVOLVER) was thus compared in scores than the other two groups for frequent words, the three groups of subjects. Again, target words were containing dominant as well as nondominant transcrip- varied in frequency. A correct response was scored if tions. This probably resulted from the fact that the R or the L was present in the subject's response. they were older and had had a longer exposure to or- The data of the hearing beginners (see Table 5) thographic material. Their scores for rare words, al- show that words containing consonant clusters entail though higher for the nondominant transcriptions, slightly more errors than control words, thus confirm- were clearly depressed for the dominant transcriptions. ing Bruck and Treiman's data. Of course, no effect of 246 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996 word frequency appears: the graphemes R or L, either ties. What seems to be the critical factor is the delivery in consonantic clusters or in control words, could be of accurate, well-specified information about the pho- spelled on the basis of consistent phoneme-to- nological contrasts of spoken language. Indeed, the grapheme rules. The CS+ group, although achieving performance of the children exposed to slightly lower performance for the clusters, show the clearly shows only that exposure to an underspecified same pattern as the hearing group. This contrasts with input is not sufficient to start the development of inner the data of the CS- children, who were more impaired speech processes. by consonant clusters, especially for rare words, and In addition, the differences observed between CS + who showed a stronger effect of word frequency. children and CS- children suggest that the critical fac- To sum up, the CS+ children exhibit spelling de- tor for acquisition of accurate phonological representa- velopment qualitatively and quantitatively similar to tions is the exposure to CS at home. Indeed, children Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 that of hearing children. This indicates that deaf chil- exposed to CS only at school do not use phonological dren can establish correspondences between visual rep- representations to the same extent and with the same resentations of speech and the alphabet, provided that accuracy as the CS+ children. The two groups may these representations are entirely specified. differ in a number of aspects. It is possible that the chil- dren exposed to CS at school suffer only from a lack of linguistic experience in terms of quantity of language. Conclusion Another possibility is that the quality of the language addressed to the child is different at school and at The fact that the CS+ children rely on inner speech home. A third explanation could be that the preco- processes to the same extent as their hearing peers for ciousness of exposure to CS plays a determinant role rhyming, remembering, and spelling has important im- in the emergence of phonological abilities. Our data do plications for several theoretical and educational issues. not allow us to choose between these explanations be- First, the observations made for the CS+ children cause the children educated with CS at home were also are relevant to the understanding of the potentialities exposed to it at an early age. Further experimental re- available to deaf people in general. Because of their search is needed to determine whether these factors sensory impairment, deaf children are often thought to can be disentangled in terms of their effect on deaf sub- develop necessarily different cognitive functioning. At jects' phonological skills. first sight, this opinion may seem supported by the Our data also confirm the close relationship be- available data. Up to now, the studies have shown that tween the use of inner speech in the activities that have some deaf people may develop phonological and meta- been called the three R's (rhyming, remembering, read- phonological abilities, but not to the same degree as ing, and spelling). The phonological representations hearing people (see e.g., Conrad, 1979). The fact that derived from CS are recruited in cognitive functioning the CS+ children, despite profound and prelingual sometimes automatically, as for memorizing, some- hearing loss, behave similarly to the hearing in a num- times perhaps more deliberately, as for rhyming or ber of situations indicates clearly that it is possible for spelling. Further work is necessary to determine if deaf children to develop phonological representations phonological information also becomes automatically and inner speech processes, provided that they are ex- available when these children recognize printed words. posed to an adequate input. A second question concerns the critical conditions for the development of such inner speech processes. References The case of CS+ children indicates that acoustic expe- rience of speech is in no way a necessary condition for Alegria, J., Leybaert, J., Charlicr, B., & Hage, C (1992). On the the development of such processes. Information deliv- origin of phonological representations in the deaf: hearing lips and hands. In J. Alegria, D. Holender, J. Morais & M. ered exclusively through the visual modality, as in CS, Radeau (Eds), Analytic approaches to human cognition (pp. may trigger off the development of phonological abili- 107-132). Amsterdam: North-Holland. Visual Speech in the Head 247

Alegria, J., & Mousty, P. (1994). On the development of lexical Conrad, R. (1970). Short-term memory processes in the deaf. and non-lexical spelling procedures of French-speaking nor- British Journal of Psychology, 61, 179-195. mal and disabled children. In G. D. A. Brown & N. C Ellis Conrad, R. (1979). The deaf school child. London: Harper & Row. (Eds.), Handbook of spelling: theory, process and intervention Cornett, O. (1967). Cued speech. American Annals of the Deaf, (pp. 211-227). Chichester John Wiley & Sons. 112, 3-13. Arthur, T. A. A., Hitch, G. J., & Halliday, M. S. (1994). Articu- Dodd, B. (1987). Lip-reading, phonological coding and deafness. btory loop and children's reading. British Journal ofPsychol- In B. Dodd & R. Campbell (Eds), Hearing by eye: the psy- ogy, 85, 283-300. chology of lip-reading (pp. 163-175). Hillsdale, NJ: Lawrence Baddeley, A. (1986). Working memory. London: Oxford Univer- Erlbaum Associates. sity Press. Dodd, B., & Campbell, R. (Eds). (1987). Hearing by eye: the psy- Baddeley, A., & Hitch, G.J. (1974). Working memory. In G. chology of lip-reading. Hillsdale, NJ: Lawrence Erlbaum As- Bower (Ed.), The psychology of learning and motivation, vol. 8 sociates. (pp. 47-90). New York: Academic Press. Dodd, B., & Hermelin, B. (1977). Phonological coding by the

Besner, D. (1987). , lexical access in reading, and artic- prelinguistically deaf. Perception and Psychophysus, 21, Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 ulatory suppression: A critical review. Quarterly Journal of 413-417. Experimental Psychology, 39A, 467-478. Erber, N. P. (1969). Interaction of audition and vision in the rec- Besner, D., Davies, J. & Daniels, S. (1981). Reading for meaning: ognition of oral speech stimuli. Journal of Speech and Hear- The effects of concurrent articulation. Quarterly Journal of ing Research, 12, 423-425. Experimental Psychology, 33A, 415-438 Frith, U. (1985). Beneath the surface of developmental dyslexia. Binnie, C A., Montgomery, A. A., & Jackson, P. L. (1974). Au- In K. E. Patterson, J. C Marshall, & M. Coltheart (Eds.), ditory and visual contributions to the perception of conso- Surface dyslexia (pp 301-330). London: Erlbaum. nants. Journal of Speech and Hearing Research, 17, 619—630. Hanson, V. L. (1982). Short-term recall by deaf signers of Amer- Brown, G. D. A., & Hulmc, C (1995). Modeling item length ican Sign Language: implications of encoding strategy for effects in memory span: no rehearsal needed? Journal of order recall. Journal of Experimental Psychology: Learning, Memory and Language, 34, 594—621. Memory and Cognition, 8, 572—583. Brack, M., & Treiman, R. (1990). Phonological awareness and Hanson, V. L., & Fowler, C A. (1987). Phonological coding in spelling in normal children and dyslexics: The case of initial word reading: Evidence from hearing and deaf readers. consonant clusters. Journal of Experimental Child Psychology, Memory and Cognition, 15, 199-207. 50, 156-178. Hanson, V. L., Liberman, I. Y, & Shankweiler, D. (1984). Lin- Burden, V, & Campbell, R. (1994). The development of word- guistic coding by deaf children in relation to beginning read- coding skills in the born deaf: An experimental study of deaf ing success. Journal of Experimental Child Psychology, 37, school-leavers. British Journal of Developmental Psychology, 378-393. 12, 331-349. Hanson, V. L., & McGarr, N. S. (1987). Rhyme generation Campbell, R. (1991). Speech in the head? Rhyme skill, reading by deaf adults. Status Report on Speech Research, S-R, 92, and immediate memory in the deaf. In D. Reisberg (Ed.), 137-156. Auditory imagery (pp. 73-93). Hillsdale, NJ: Lawrence Erl- Hanson, V. L., Shankweiler, D, & Fischer, F. W. (1983). Deter? baum Associates. minants of spelling ability in deaf and hearing adults: Access Campbell, R., & Dodd, B. (1980). Hearing by eye. Quarterly to linguistic structure. Cognition, 14, 323—344. Journal of Experimental Psychology, 32, 85-99. Kuhl, P. K., & Meltzoff, A. N. (1982). The bimodal perception Campbell, R., & Wright, H. (1988). Deafness, spelling and of speech in infancy. Science, 218, 1138-1141. rhyme: How spelling supports written words and picture Leybaert, J., & Alegria, J. (1993). Is word processing involuntary rhyming skills in deaf sujects. The Quarterly Journal of Ex- in deaf children? British Journal of Developmental Psychology, perimental Psychology, 40A, 771-788. 11, 1-29. Campbell, R., & Wright, H. (1990). Deafness and immediate Leybaert, J., & Alegria, J. (1995). Spelling development in deaf memory for pictures: Dissociations between "inner voice" and hearing children: Evidence for use of morpho- and "inner ear"? Journal of Experimental Child Psychology, phonological regularities in French. Reading and Writing, 7, 32, 259-286. 89-109. Cardoso-Martins, C (1994). Rhyme perception: global or ana- Leybaert, J., Alegria, J., & Fonck, E. (1983). Automaticity in lytical. Journal of Experimental Child Psychology, 57, 26-41. word recognition and word naming by the deaf. Cahiers de Charlier, B. L. (1994). Le developpement des representations Psychologic Cognitive, 3, 255-272. phonologiques chez 1'enfant sourd: Etude comparative du Lyxell, B., Arlinger, S., Anderson, J., Bredberg, G., Harder, Langage Parle Complete avec d'autres outils de communica- H., & Ronnberg, J. (1996). Verbal-information processing tion. Brussels: U.L.B., Unpublished Ph.D. dissertation. capabilities and cochlear implants: Implications for pre- Conrad, R. (1962). An association between memory errors and operative predictors of speech understanding. Journal of errors due to acoustic masking of speech. Nature, 193, 1314- Deaf Studies and DeafEducation, 1, 190-201. 1315. MacKain, K. S., Studdert-Kennedy, M., Spieker, S., & Stern, Conrad, R. (1964). Acoustic confusions in immediate memory. D. (1983). Infant intermodal speech perception is a left British Journal of Psychology, 55, 179-185. hemisphere function. Science, 219, 1347-1349. 248 Journal of Deaf Studies and Deaf Education 1:4 Fall 1996

Marschark, M. (1995). Auditory imagery in cognitive processing R. Campbell (Eds), Hearing by eye: the psychology of lip- by deaf adults. Paper presented at the VTTI ESCOP meeting, reading (pp. 3-51). Hillsdale, NJ: Lawrence Erlbaum Asso- Rome, Italy. ciates. McGurk, H., & McDonald, J. (1976). Hearing lips and seeing Taeschner, T, Devescovi, A., & Volterra, V. (1988). Affixes and voices. Nature, 264, 746-748. function words in written language of deaf children. Applied Mills, A. E. (1987). The development of phonology in the blind Psycholinguists, 9, 385-401. child. In B. Dodd & R. Campbell (Eds.), Hearing by eye: The Treiman, R., & Breaiw, A. M. (1982). Common phoneme and psychology of lipreading (pp. 145-171). London: Lawrence overall similarity relations among spoken syllables: their use Erlbaum Associates. by children and adults. Journal of Psycholinguistic Research, Nicholls, G., & Ling, D. (1982). Cued speech and the reception II, 569-598. of spoken language. Journal of Speech and Hearing Research, Volterra, V, & Bates, E. (1989). Selective impairment of Italian 25, 262-269. grammatical morphology in the congenitally deaf: A case Perier, Q, Charlier, B. L., Hage, C, & Alegria, J. (1988). Evalua- study. Cognitive Neuropsychology, 6, 273-308.

tion of the effects of prolonged cued speech practice upon Walden, B. E., Prosek, R. A., Montgomery, A. A., Scherr, Downloaded from https://academic.oup.com/jdsde/article/1/4/234/486391 by guest on 25 September 2021 the reception of spoken language. In I. G. Taylor (Ed.), The C K., & Johs, C J. (1977). Effects of training on the visual education of the deaf: Current perspectives, vol.1, 1985 Inter- recognition of consonants. Journal of Speech and Hearing Re- national Congress on Education for the Deaf. Beckenham, search, 20, 130-145. Kent, UK: Croom Helm, LTD. Walton, G. E., & Bower, T. G. R. (1993). A modal representation Seymour, P. H. K., & MacGregor, C. J. (1984). Developmental of speech in infants. Infant Behavior and Development, 16, dyslexia: a cognitive experimental analysis of phonological, 233-243. morphemic and visual impairments. Cognitive Neuropsychol- Wandel, J. E. (1990). Use of internal speech in reading by hear- ogy, 1, 43-82. ing impaired students in oral, total communication and cued Spoehr, K. T. ,& Conn, W. S. (1978). The stimulus suffix as speech programs. Unpublished master's thesis, Colombia a memory code phenomenon. Memory C Cognition, 6, University. 583-589. Waters, G. S., & Doehring, D. G. (1990). Reading acquisition in Sruddert-Kennedy, M. (1986). Sources of variability in early congenitally deaf children who communicate orally: Insight speech development In J. S. Perkell & D. H. Klatt (Eds.), from an analysis of component reading, language, and mem- Invanance and variability in speech processing (pp. 58-76). ory skills. In T Carr & B. A. Levy (Eds), Reading and its Hillsdale, NJ: Lawrence Erlbaum Associates. development: component skills approaches (pp. 323—373). New Sumby, W., & Pollack, I. (1954). Visual contributions to speech York: Academic Press. visibility in noise. Journal of Acoustic Society of America, 26, Webster, P. E., & Plante, A. S. (1995). Productive phonology and 212-215. phonological awareness in preschool children. Applied Psy- Summerfield, Q; (1987). Some preliminaries to a comprehensive cholinguistics, 16, 43—57. account of audio-visual speech perception. In: B. Dodd &