Cognitive development in deaf children: the interface of...

2002 Elsevier Science B.V. All rights reservedHandbook of Neuropsychology, 2nd Edition, Vol. 8, Part IIS.J. Segalowitz and I. Rapin (Eds)

CHAPTER 4

Cognitive development in deaf children: the interface oflanguage and perception in neuropsychology

Rachel I. Mayberry *

School of Communication Sciences and Disorders, McGill University, 1266 Pine Avenue West, Montreal, PQ H3G 1A8, Canada

Introduction

What does the sense of hearing contribute to humandevelopment? To answer the question, we must askwhat the sense of hearing gives the child. Hearinggives the child the acoustic correlates of the physicalworld: approaching footsteps, dog barks, car horns,and the pitter-patter of rain. Hearing also allowsthe child to revel in the patterned complexity of aBeethoven symphony or a mothers lullaby. Childrenwho are born deaf clearly miss a great deal. How-ever, hearing conveys much more to the growingchild than the acoustics of the physical world. Hear-ing is the sensory modality through which childrenperceive speech the universe of talk that ties in-dividuals, families and societies together. Childrenborn with bilateral hearing losses that are severe(7089 dB loss) or profound (>90 dB loss) are re-ferred to as deaf. They cannot hear conversationalspeech (approximately 60 dB) and consequently donot spontaneously learn to talk. Indeed, not talkingat the right age is one of the first signs that a childcannot hear.

The primary consequence of childhood deafnessis that it blocks the development of spoken language both the acts of speaking and comprehending.This fact leads us to ask what spoken languagecontributes to the childs cognitive development. Be-

Corresponding author. Tel.: +1-514-398-4141;E-mail: [email protected]

cause deafness impedes the development of spokenlanguage, we must ask whether complex and logi-cal thought can develop in the absence of spokenlanguage. Can the child develop inner thought orworking memory without the ability to hear? Con-sider sign language. Can sign language foster thesame kinds of abstract mental development and com-plex thought as speech? Now consider an even morecomplex situation, namely, the cognitive develop-ment of children who grow up with little or noexposure to any language in any form, be it signed orspoken, as a simple consequence of being born deaf.What are the effects of such linguistic and socialisolation on the childs development of a mental life?

Each of these questions has been asked aboutdeaf children in one form or another since the be-ginning of philosophical inquiry (Lane, 1984). Atfirst glance, the answers would seem to be straight-forward and, historically, have been treated as such.However, as we shall see, the answers to these ques-tions quickly become complex for several reasons.First, cognitive development entails more than matu-ration of the childs brain. Cognitive development isthe product of the childs attempts to understand thefamily, neighborhood, school and the world at largeduring this period of rapid brain growth and learn-ing. The effects of deafness on cognitive develop-ment are, therefore, quite diverse and complex due tothe multitudinous ways in which families, societies,and cultures, react to and interact with children whoare born deaf and hence do not spontaneously learnto talk and comprehend speech. Against this enor-

CICERO/GALAYAA B.V./SEGA2-4: pp. 71-107

71

Ch. 4 R.I. Mayberry

mously varied backdrop, we explore here researchon the cognitive development of deaf children.

We begin with the premise that cognition, orintelligence, is multi-faceted and reflected in the co-ordinated performance of numerous language andnon-language tasks, including perception, memory,mental imagery, concept formation, problem solv-ing, language learning, academic achievement, andnavigating everyday life (Sternberg, 1989). Our fo-cus here is on whether and how deafness affectsthe childs cognitive development across several do-mains that have been studied in some, but not equal,detail. The cognitive aspects we consider are thefollowing: (1) Academic achievement; (2) readingdevelopment, (3) language development and the fac-tors known to affect it; (4) performance on stan-dardized intelligence tests; (5) visualspatial andmemory skills; (6) conceptual development, and (7)neuropsychological function. For each domain, wediscuss the development of deaf children who use ei-ther signed or spoken language. However, because solittle is known about them, we will only briefly dis-cuss the cognitive development of deaf children whogrow up with little or no exposure to conventionallanguage (in any form, spoken or signed). We beginwith academic achievement because it illustrates theunique developmental challenge congenital deafnessposes for the young child.

Academic achievement

Population profiles

The Gallaudet Research Institute regularly collectsand analyzes demographic data on the academicachievement of deaf children in the United Statesbased on the Stanford Achievement Test (Allen,1994). The median math computation skills of 15-year-old deaf children in the USA are at the 7th gradelevel. Age-matched hearing children perform at the10th grade level (Allen, 1989). These statistics showthat deafness, by itself, does not impede the childsability to learn and manipulate abstract symbols andsymbolic relations. By contrast, the median readingachievement of 1721-year-old deaf students leavingAmerican secondary schools is at the 4th grade level(Allen, 1994). This wide performance gap betweenlanguage tasks as compared to non-language tasks

is a common profile among deaf children world-wide (Conrad, 1979). These academic performancepatterns illustrate the great difficulty experienced bydeaf children perceiving and learning spoken lan-guage and visual representations of speech, namelywritten and read language. Indeed, the effects ofdeafness on spoken language development increaseas degree of hearing loss increases. For example,students with mild to moderate hearing losses readat lower levels than do students with normal hearing.Furthermore, students with severe to profound hear-ing losses read more poorly than do students withmoderate losses but on math computation they showequivalent achievement (Allen and Schoem, 1997).

The primary effect of degree of hearing loss onlanguage development, in turn, interacts with factorsextraneous to deafness, such as socioeconomic andethnic status and additional handicaps. Deaf childrenfrom lower socioeconomic status homes perform lesswell than middle-class deaf children. For example,only 5% of Black and 6% of Hispanic deaf stu-dents (1721 years old, severely to profoundly deaf)read at or above the 6th grade level whereas 15%of White deaf students read above this level (Allen,1994). Deaf children who have motor or sensoryimpairments in addition to deafness, such as poorvision or cerebral palsy, perform less well as a groupthan deaf children without additional impairments(Allen, 1989). Together these data indicate that theacademic achievement of deaf students is predictedto a large extent by the same factors that predictthe academic achievement of normally hearing stu-dents in North America, that is, social class, ethnicand racial background, and other handicapping con-ditions. This means that deafness, by itself, doesnot determine academic success or failure but ratherinteracts with many other factors in complex ways.

Reading development

The median reading level of the deaf, high schoolpopulation does not reach the level required fora person to be considered literate (i.e., the 6th to8th grade level and beyond). Indeed, the medianreading levels of the deaf student population have notchanged much over the past century (Chamberlainand Mayberry, 2000). This discouraging, but oftenreplicated, finding suggests that something about

72


Cognitive development in deaf children Ch. 4

deafness creates a barrier to reading development.However, if the barrier were insurmountable, no deafstudents would read proficiently. It is important toremember that these reading statistics are medianreading levels. Half of deaf high school students readbelow the fourth grade level but half also read abovethis level.

Factors in reading development

Spoken languageMany deaf students read well. For example, in astudy of 100 profoundly deaf 1617-year-olds, Geersand Moog (1989) found 57% to read at or above thegrade 7 level and 30% to read at the grade 10 level.The students shared several traits: above averageperformance on nonverbal IQ tests, parents who werecollege educated and professionally employed, andspecial education begun at or before preschool age.These facts mean that the students in this study allhad access at an early age to high quality interventionservices. In addition to high reading achievement,the students performed at high levels on a battery ofspoken language tests.

Indeed, the spoken language development of deafchildren, as measured by syntactic and vocabularyskills (independent of word decoding skills), has of-ten been postulated to be a causal factor in readingdevelopment in the deaf population. Spoken lan-guage skills have been found to predict readinglevels in deaf students, both those who use sign lan-guage (r = +0.70, Lichtenstein, 1983; Moores andSweet, 1990) and those who use speech (r = +0.83,Geers and Moog, 1989). Sign language has histor-ically been conceptualized as being unimportant oreven detrimental to the reading development of deafstudents, either because it has no sound patterningor because its grammatical structure is different fromthat of spoken language (Mayer and Wells, 1996;Moores and Sweet, 1990). In fact, whether signlanguage can provide the cognitive foundation thatspoken language provides for reading developmenthas been a matter of considerable debate for decades.Only recently has the question been investigated ina systematic fashion (Chamberlain and Mayberry,2000). If sign language development interferes withreading development, then there should be a negativerelation between deaf childrens sign language skills

and reading ability but recent research has found theopposite relation. Recent research shows a positivecorrelation between sign language skills and readingdevelopment.

Sign languageMayberry and her colleagues (Mayberry, Chamber-lain, Waters and Doehring, 2001) investigated thereading and sign skills of 48 severely and profoundlydeaf students aged 715 years. They found 42% toread at the expected grade level and 67% to readabove the fourth grade level. All the children wereeducated with sign language and spoken language(specifically Total Communication). All the chil-dren additionally had average or above average per-formance on selected nonverbal IQ sub-tests and hadbegun to learn sign language through early interven-tion by age three. Unlike the orally trained childrenstudied by Geers and Moog (1989), the readinglevels of the children educated with total communi-cation were predicted by their ability to comprehendsign language, not spoken language. (We define anddiscuss total communication below). These findingssuggest that factors which promote the developmentof language comprehension in general, independentof sensorymotor modality, also promote reading de-velopment. In fact, the deaf children who showed thehighest levels of reading and sign language compre-hension tended to have parents who knew and usedsign language with them from a very young age 3 years or before. Parental use of sign language alsomeant that the deaf child was able to converse insigned language in nearly all domains of his or herlife both at home and school which wouldmean that the child had abundant amounts of lan-guage input during childhood (Hoffmeister, 2000).

Other studies have also measured sign languageskill in relation to reading development and found apositive relation between the two kinds of language.The sign language measures have included the fol-lowing kinds of American Sign Language (ASL)knowledge: plural markers and synonyms/antonyms,(r = +0.45 to +0.64, Hoffmeister, 2000); compre-hension of ASL classifiers, stories, and time mark-ers (combined r = +0.58, Strong and Prinz, 2000);memory for ASL sentences, classifier production,and recognition of fingerspelling in ASL sentences(r = +0.53 to +0.57, Padden and Ramsey, 2000);


73

Ch. 4 R.I. Mayberry

and comprehension of ASL stories and workingmemory for ASL sentences (r = +0.49 to 0.69Chamberlain and Mayberry, 2000; Mayberry et al.,2001). One unique feature of these studies is that thehigh and positive correlation between sign languageskill and English reading was found specifically forASL, a natural sign language with a linguistic struc-ture different from spoken English. These findingsfurther show that the major impediment to deaf chil-drens reading development is not simply an inabilityto speak English but rather impoverished languagedevelopment, in any form, signed or spoken, as wediscuss below.

Good and poor readersThe relation between ASL language ability and read-ing achievement was further investigated by Cham-berlain (2001) who hypothesized that well developedreading skill is predicated on well developed lan-guage skill, even in sign language. To test the hy-pothesis, she measured the ASL and reading skillsof a random sample of 35 deaf adults who reportedusing ASL as a primary language. Sign languageskill was measured with two tasks, ASL grammaticaljudgement (Boudreault, 1998) and narrative compre-hension in ASL and MCE, Manually Coded Englishdescribed below (Mayberry et al., 2001). Readingskills were measured with two reading tests (theGatesMacGinite and the Stanford). Performancedistribution on the ASL and reading measures sig-nificantly overlapped. Half the adults performed athigh levels on the sign language tasks, at near-nativelevels. These same individuals also could read well,reading at the 8th grade level or above. Most of theremaining participants performed at low levels onthe sign language measures. These same individualsperformed poorly on the reading tests, reading belowgrade 4 (Chamberlain, 2001) as Fig. 1 shows.

A few deaf adults read well but performed poorlyon the sign language tasks. These individuals were,in actual fact, successful speakers of English. Al-though they considered sign language to be theirprimary mode of communication, they had success-fully acquired spoken English in childhood and sub-sequently learned ASL as a second language inlate adolescence. In other words, they had well-developed primary language skills. A few deaf adultsperformed at high levels on the sign language tasks

but could not read well, constituting about 8.5% ofthe experimental sample. This figure is very similarto the proportion of the normally hearing populationreported to have dyslexia, around 10%, that is, peo-ple who have good face-to-face language skills butcannot read (Chamberlain, 2001).

These results show that highly developed signlanguage skill is related to high levels of readingachievement in deaf individuals for whom sign lan-guage is a primary means of communication. Spokenlanguage is not the only path to literacy develop-ment. These results also suggest that the low medianreading levels of the deaf school-aged populationare likely caused by low levels of primary languagedevelopment, in sign language as well as in spokenlanguage. Reading must be taught to deaf children,certainly. But in order to benefit from reading in-struction, deaf children must have a well-developedprimary language upon which to base the readingtask.

Additional factors in reading achievement

Returning to the factor of socioeconomic status,Geers and Moog (1989) found in their sample oforally trained deaf students that those who achievedthe highest reading levels were from middle classfamilies. More specifically, they had parents whowere college educated and professionally employed.They concluded that socioeconomic status predictsdeaf childrens reading achievement. This finding issupported by demographic data showing that His-panic and Black deaf children show significantlylower levels of reading achievement than do Whitechildren (Allen, 1994). Socioeconomic status mayhave a greater impact on the academic attainmentof deaf children than that of hearing children. Thisis because hearing children, no matter how poor,spontaneously acquire language by merely listeningto family members who speak to them from infancy.By contrast, poor deaf children are at a high riskfor not being exposed to accessible language at theright time in early childhood. This is because in mostcountries poverty translates into a lack of access tothe educational and clinical services that expose deafchildren to language at the appropriate age. Thesefactors are early diagnosis of hearing loss, parentinfant and preschool programs, availability of hear-

74



Fig. 1. The relation between English reading level and comprehension of American Sign Language (ASL) in a sample of 35 randomlyselected deaf adults reporting ASL to be their primary language. Good ASL skills indicate nearly complete comprehension of ASLnarratives and near-native control over ASL syntactic structures; poor ASL skills indicate minimal control of ASL syntactic structuresand comprehension of less than half of an ASL narrative (extrapolated from Chamberlain, 2001).

ing aids, parentinfant programs that promote andteach sign language, parental sign language instruc-tion, parent counseling, and so forth. This leads usdirectly to the question of why primary languageskill is so difficult for deaf children to develop inde-pendent of socioeconomic status.

Language development

Because deaf children cannot hear the language spo-ken around them, they do not learn it spontaneously.In fact, one of the first signs that a child is deaf,aside from not responding to sound, is not beginningto talk at the expected age, 1018 months (Meadow-Orlans, 1987). Making generalizations about the lan-guage development of deaf children is complex. Thisis due to the multiple sensory forms of linguisticinput deaf children receive and the varying kinds andamounts of language input available to them duringearly childhood. For example, some deaf childrenreceive no accessible language input of any kind(i.e., children who have severe to profound hearinglosses that go undetected and thus receive no specialintervention). Other deaf children receive incompletespoken language input (incomplete because hearingaids and cochlear implants do not restore hearing tonormal levels). Some other children receive sign lan-

guage input in the form of MCE (Manually CodedEnglish) or ASL. Signed language input is, in princi-ple, fully accessible to deaf children due to its visualnature. Unfortunately, familial and educational cir-cumstances often inadvertently conspire to yield bothincomplete and reduced amounts of signed and spo-ken linguistic input throughout the deaf childs earlylife when normal language development occurs, aswe discuss in detail below.

Speech

When considering deaf childrens language devel-opment, it is important to distinguish the childsability to speak from the childs ability to understandand produce language, that is, linguistic competence.The two factors of speech and language are clearlydissociable in deaf childrens development. Indeed,intelligible speech is extremely difficult for deaf chil-dren to achieve. This is understandable for two rea-sons. First, deaf children either do not hear speechat all or hear it in a highly distorted fashion dueto the sensori-neural nature of their hearing losses.Second, the visual facial movements people makewhen speaking provide only limited clues as to howspeech is produced. Deaf children as a whole achievelow levels of speech intelligibility (for a review see


75

Ch. 4 R.I. Mayberry

Seyfried and Kricos, 1989). Some proportion of deafchildren learn to speak intelligibly. Such childrentypically have received special education at an earlyage (before preschool) with successful use of hearingaids, speech training, and very high levels of parentalinvolvement in the childs speech and auditory devel-opment (Meadow-Orlans, 1987). Three other factorscommon to deaf children who successfully learn tospeak are the following: (1) the ability to auditorilydiscriminate speech patterns, (2) higher than averagenonverbal IQ, and (3) higher than average socioeco-nomic status (Geers and Moog, 1987).

Vocabulary

In a pioneering study of deaf childrens cognitivedevelopment, Katheryn Meadow observed that, Thebasic impoverishment of deafness is not lack ofhearing, but lack of language. To illustrate this,we have only to compare the 4-year-old hearingchild, with a working vocabulary of between twoand three thousand words, to a child of the same ageprofoundly deaf since infancy, who may have only afew words at his command, (Meadow, 1968: 29).

The major developmental hurdle facing deaf chil-dren is not simply learning to speak intelligibly, butacquiring language namely the lexicon, morphol-ogy, syntax, and semantics of language. Substantialdelays in language development are the primary hall-mark of childhood deafness without early and appro-priate intervention. The reasons for this are simple.The average intensity of conversational speech bya male talker is around 60 dB. Even children withmoderate hearing losses (i.e., 5670 dB) show a1-year delay in vocabulary development comparedto age-matched children with no hearing loss. Chil-dren with severe hearing losses (7190 dB) show a3-year lag in vocabulary development (Davis, Elfen-bein, Chum and Bentler, 1986). In turn, profoundhearing loss (


Fig. 2. The vocabulary comprehension test scores of 150 deaf children plotted as a function of expected performance for chronologicalage (with norms from hearing children added; redrawn from fig. 19, Moeller et al., 1986).

The striking delays in deaf childrens vocabularyacquisition leads to the question of whether deafnessaffects the cognitive processes that underlie wordlearning. However, this appears not to be the case.Lederberg, Prezbindowski and Spencer (2000) foundthat overall vocabulary size, not age, predicted deafchildrens word learning strategies. Young deaf chil-drens ability to learn new words from context withminimal exposure was similar to that of hearing chil-dren, but only when vocabulary size was taken intoaccount. This finding suggests that some kinds oflanguage learning abilities emerge as a consequenceof language development. Thus, deaf children whoselanguage is very delayed may show learning patternssimilar to younger hearing children, not due to deaf-ness per se, but due instead to a significantly delayedrate of language growth. This is an important conceptto which we return below.

Grammar

Given the highly delayed vocabulary acquisition ofdeaf children, it is not surprising to learn that theiracquisition of syntax and grammatical morphologyis also very delayed. In a study of orally-trained, 415-year-olds (75% of whom were 7 years and older),Geers and Moog (1978) found 50% of the children

to have a level of expressive grammar lower thanwould be expected for hearing 3-year-olds (but withmore mature content). In a study of 150 deaf childrenbetween the ages of 4 and 20, Moeller et al. (1986)found grammatical comprehension to be even moredelayed than vocabulary comprehension. Few of thestudents tested achieved grammatical comprehensionscores higher than would be expected for normallyhearing 57-year-olds. Some researchers concludethat deaf children taught exclusively through spokenlanguage appear to pass through the same generalstages of language acquisition as their hearing peersbut without reaching the same ultimate level of pro-ficiency (Mogford, 1988).

Many similarities between deaf and hearing chil-drens syntactic acquisition were documented in oneof the most detailed investigations of deaf studentssyntactic skills ever conducted. Quigley and his col-leagues investigated the English syntactic skills ofmore than 400 deaf students on a variety of En-glish grammatical structures including, for exam-ple, negation, conjunction, determiners, questions,pronominalization, relativization, and complementa-tion, among others (for a review see Quigley andKing, 1980). The research team first collected nu-merous samples of deaf childrens writing to de-termine the most common types of grammatical


77

Ch. 4 R.I. Mayberry

errors they committed. In follow up studies, the teaminvestigated the performance patterns of deaf stu-dents on a wide range of syntactic structures. Thenet result was both a screening test and a normeddiagnostic test of grammatical skill, Test of Syntac-tic Abilities (Quigley, Steinkamp, Power and Jones,1978).

These studies produced two main findings. First,deaf childrens control of English syntax decreaseswith decreasing hearing levels. Second, the orderof difficulty of English syntactic structures for deafstudents, even those with profound hearing losses, isfor the most part highly similar to those of Englishhearing students and second language learners ofEnglish (Quigley and King, 1980), as shown inFig. 3. It is important to note, however, that thissyntactic research was conducted solely through themodes of reading and writing. The extent to whichthese findings generalize to deaf childrens face-to-face language in sign and/or speech is unclear.Nonetheless, these studies demonstrate the low levelof control many deaf students have over Englishsyntax when reading.

Fig. 3. Performance accuracy on selected English syntactic struc-tures for 813-year-old hearing students, deaf students betweenthe ages of 1018 years in both the USA and Australia and deafcollege students (fig. 7 from Quigley and King, 1980).

The achievement patterns deaf children show forspoken and written language indicate that they donot easily learn the grammar of a spoken languageeven in a visual form, i.e., reading and writing.Nearly a century of teaching and research with deafchildren has also demonstrated that learning to readcannot serve as a primary language for deaf childrenfor reasons we do not fully understand. Learningto read and learning a primary language are twodifferent skills. We know now that learning to readis based upon the prior acquisition of a primary anddynamic face-to-face language in either a signedor spoken form. Because primary language learningis often difficult and problematic for deaf children,educators have sought various means to facilitatelanguage acquisition through the use of gesture andsign language. We now consider the gesture and signlanguage development of deaf children.

Gesture and homesign

Deaf children, who have not yet been exposed tosign language and have not yet acquired spoken lan-guage, often spontaneously create their own meansof gesture communication for the purpose of self-expression. The phenomenon is known as homesign.Homesign is an ASL term referring to the idiosyn-cratic gesture systems used by deaf individuals whowere reared in isolation from other deaf signers. Thephenomenon is that the deaf child makes statementsand requests by combining points (to objects andpeople) with iconic gestures (called characterizinggestures by Goldin-Meadow and Feldman, 1977) inan effort to communicate with hearing family mem-bers. For example, the child might point to an appleand then open and close his mouth around his fistto mean, I eat an apple, (Goldin-Meadow, per-sonal communication). All the young deaf childrenin hearing families observed by Goldin-Meadow andMylander (1984) used gestures to communicate withtheir mothers with varying degrees of frequency.

In addition to creating a gesture lexicon, the deafchildren combine their gestures into multi-gestureutterances in a rule-governed fashion. The childrentend to gesture the patient (recipient of the action)followed by the action when the action is transi-tive (i.e., when the action acts on something as ineat, kick, twist, etc.). However, when the action

78



is intransitive (e.g., think, sleep, walk, etc), thechildren tend to gesture the agent followed by theaction (Goldin-Meadow and Feldman, 1977; Goldin-Meadow and Mylander, 1984).

In subsequent research these researchers foundthat the gesture lexicon and ordering rules of thegesture system originate from the deaf child and nottheir hearing parents. Indeed, the gestures of youngdeaf children differ from those of their hearing par-ents in frequency, complexity, and ordering patterns.The deaf child combines strings of gestures to con-vey ideas, or propositions (as the above exampleillustrates), similar to the way in which speakerscombine strings of words to make statements. Bycontrast, the gestures produced by hearing people asthey speak are much simpler. Speakers tend to pro-duce only a single iconic gesture for each major ideathey say (Mayberry and Nicholadis, 2000; McNeill,1992). In homesign, gestures carry all of the mean-ing. In speech, spoken words carry all the meaningand gestures act as a supplement (McNeill, 1992).In young normally hearing children, gesture com-plexity increases with spoken language developmentfrom the earliest stages (Mayberry and Nicholadis,2000). This means that gesture communication islinguistic for hearing children just as it is for deafchildren.

The characteristics of homesign appear to be uni-versal and transcend culture so far as is currentlyknown. Goldin-Meadow and Mylander (1998) ob-served that Chinese deaf children produce homesigntoo. The homesign gestures and ordering patternsof the Chinese deaf children were very similar tothose of American deaf children. Homesign may bea valuable diagnostic tool, although no research hasyet documented the possibility. How often and howelaborately the young deaf child communicates viagesture prior to having learned any sign language orspoken language, or how frequently and complexlyhe or she gestures when speaking probably indicatesincreased symbolic sophistication (in either spokenor signed forms).

We now know that deaf childrens gesture com-munication can be elaborate and shows some basicproperties shared by all languages. How is home-sign related to sign language? We turn now to thisquestion.

Sign language origins

A common misconception about the origins of signlanguages is that they were invented by hearingteachers. However, documents from some of the ear-liest teachers in the late 18th century show that theyintentionally borrowed the signs and gestures theirdeaf pupils used in the classroom to communicatewith one another and used them for instructionalpurposes (Lane, 1984). 20th century research inNicaragua demonstrates how quickly sign languagesemerge among communities of individuals who aredeaf from birth.

After the Sandinista revolution, literacy and ed-ucation became a national priority in Nicaragua,including the education of deaf children (Morfordand Kegl, 2000). Before the revolution, individualswho were deaf did not attend school. Instead, theylived with their hearing families in isolation fromone another. Like deaf children everywhere whoare unable to communicate with spoken language,these isolated deaf individuals developed idiosyn-cratic gesture systems, or homesign, to communicatewith their families. Their homesign systems hadmany of the properties described above (Morfordand Kegl, 2000).

When the first school for deaf students was estab-lished in Nicaragua, linguists who were present at thetime observed that the older students, namely ado-lescents and young adults, communicated with oneanother using their various homesign systems (Kegl,personal communication; Morford and Kegl, 2000).Recall that homesign consists of combinations oficonic and point gestures to create simple utterances,as described above. When the youngest deaf chil-dren arrived at the school (e.g., ages 45), they wereexposed to the homesign used by the older deaf stu-dents for communication. Surprisingly, however, theyounger children did not copy the homesign systemsof the older students. Instead, they were observedto use a sign language! The signing of the youngestdeaf students was quick, limited to the hands (i.e.,the children did not use pantomime), constrained inspace, and consisted of signs with sub-lexical struc-ture (i.e., meaningless phonetic units). They orderedtheir signs in relation to one another with syntaxand grammatical morphemes. In other words, theseyoung deaf children had created a sign language, or


79

Ch. 4 R.I. Mayberry

Idioma de Seas de Nicaragua, from homesign input(Morford and Kegl, 2000; Senghas, 1995; Senghas,Kegl, Senghas and Coppola, 1994).

The phenomenon of children creating linguisticstructure from fragmented language input is by nomeans unique to sign language. The phenomenon,known as creolization, has often been observed insituations of language contact where adults speaka pidgin (vocabulary words of a foreign languagestrung together with little or no syntax or morphol-ogy as means of communication between groups ofadults speaking different languages). Young childrenexposed to pidgins creolize them, that is, they cre-ate new syntax and morphology to fill the existinggaps in the pidgin to which they have been exposed(Bickterton, 1990). Indeed, some of the first linguiststo study ASL initially hypothesized that the sign lan-guages used by Deaf communities worldwide wereanalogous to spoken Creoles (Fischer, 1978).

The sudden appearance of Idioma de Seas deNicaraga provides insights into how sign languagesevolve. There are at least two necessary stages ofsign language evolution. First, from the sparse, butaccessible (i.e., visible), gesture input they receivefrom speaking people who gesture as they talk, deafchildren create gesture communication home-sign. This homesign shows rudiments of linguisticstructure, namely a limited lexicon and orderingrules (Goldin-Meadow and Mylander, 1998). Sec-ond, young deaf children exposed to the homesignof older deaf children, in turn, fill the grammaticalgaps in this rudimentary, quasi-language input to cre-ate a sign language (Morford and Kegl, 2000). Signlanguage is thus a remarkable human adaptation tochildhood deafness. Language capacity is such anintegral part of the human endowment that whenlanguage expression in the auditoryoral form isblocked by profound deafness, it reappears one gen-eration later in a readily accessible and expressibleform that circumvents the spoken language barrier ofdeafness.

American Sign Language

Most Deaf communities in North America use asign language known as American Sign Language,or ASL. ASL is a natural language that has evolvedamong deaf people in North America, probably in

the manner described above. Because ASL is per-ceived by the eyes and expressed with the handsand arms (in addition to the face and torso), itsgrammatical structure (phonology, lexicon, morphol-ogy, syntax, and semantics) is highly spatial (Em-morey, 2001). ASL is not a version of English onthe hands. However, ASL has rarely been used inclassrooms for deaf children. In fact, ASL has histor-ically been actively banned from classrooms for deafchildren because the goal of educators has alwaysbeen to teach spoken language. Most educators havetraditionally believed that ASL impedes this goal(Baynton, 1996; Lou, 1988). Nonetheless, ASL isthe native language of many deaf people who wereexposed to sign at a young age, especially those withdeaf parents. ASL is also the primary language ofmany deaf people who do not acquire spoken lan-guage, even though they often learn ASL at very lateages, as described below.

ASL acquisitionDeaf children exposed to ASL from birth by theirdeaf parents spontaneously acquire it in a fashioncomparable to that of hearing children who spon-taneously acquire spoken language. Beginning witha sign-babbling stage, the child begins with a one-word stage, moves to a two-word stage, and then be-gins to acquire the complex morphological system ofASL (Marentette and Mayberry, 2000; Newport andMeier, 1985; Petitto and Marentette, 1991; Schick,1990). The child does not fully master ASL gram-mar until around 56 years of age or later (for recentresearch on sign language acquisition, see Chamber-lain, Morford and Mayberry, 2000). The ease withwhich deaf children spontaneously acquire ASL un-equivocally demonstrates that congenital deafnessdoes not alter the childs ability to acquire grammar.Rather, the major problem congenital deafness posesfor the young deaf child is accessibility to suffi-cient language input at the right age (Morford andMayberry, 2000).

The age of exposure problemOne rationale behind early speech intervention pro-grams for deaf children (and bilingual education forhearing children) is the idea that languages are bestlearned early in life (Scovel, 1989). People generallybelieve that spoken languages can only be learned

80



spontaneously and perfectly in early childhood. Thisconcept is referred to as the critical period for lan-guage acquisition. Critical periods are common inthe development of living things and are defined asa time-bounded period during development when thedeveloping system, such as vision, is most sensi-tive to environmental input. A well-studied humanexample in is the development of binocular vision(Greenough and Black, 1992). Although most re-searchers think that spoken language acquisition isguided by critical period learning, sign languageshave traditionally been excluded from this principle.The prevailing conviction among many educators hasbeen that proficiency in sign language can be readilyattained by anyone at any age. However, recent re-search shows that the age constraints that guide theoutcome of second language acquisition also applyto the ultimate outcome of sign language acquisi-tion (Emmorey, Bellugi, Frederici and Horn, 1995;Newport, 1988).

In a series of studies, Mayberry and her col-leagues (Mayberry, 1994; Mayberry and Eichen,1991; Mayberry and Fischer, 1989) found numerouseffects associated with age of sign language expo-sure. For example, overall accuracy decreased withlater age of sign language acquisition. This was truefor memory and comprehension of ASL sentencesand stories as well as for sentences in a versionof Pidgin Sign English (PSE), a simplified formof ASL with some English elements used betweendeaf and hearing people. Deaf participants who werenot exposed to sign until late childhood and earlyadolescence, around 913 years, showed significantdeficits in ASL comprehension, despite the fact thatASL was their primary means of communication andthey had used ASL for all of their adult lives. Indeed,the performance of the deaf individuals who firstlearned ASL in adolescence was worse than whatwould be expected from second-language learners ofa spoken language. Why might this be so?

Recall that the major consequence of childhooddeafness is a significant delay in the development ofspoken language. This means that when deaf adoles-cents are first exposed to sign language, it is oftenbecause they acquired insufficient spoken languageto enable them to cope with everyday life. In otherwords, many deaf children are only allowed to usesign language after they prove that they are unable

to acquire spoken language. For these deaf adoles-cents, then, sign language exposure occurs at anage well beyond early childhood. Importantly, thesedeaf students have often acquired very little func-tional spoken language prior to learning to sign. Thissituation is not second-language learning. This situ-ation is akin to first language acquisition begun inadolescence. The important question then becomeswhether these deaf students ever catch up with re-spect to their sign language skills. Do they ultimatelyuse sign language as proficiently as second-languageor native-language (i.e., acquired from birth) learnersof ASL?

This question was investigated in another study(Mayberry, 1993). The ASL sentence comprehen-sion skills of deaf adults with contrasting earlylanguage backgrounds were compared. One exper-imental group consisted of congenitally deaf adultswho began to learn sign language between the agesof 9 and 13. They were exposed to sign languageat this late age because their spoken language skillswere not functional. Thus, their ASL acquisition wasanalogous to first-language learning. The compari-son group consisted of deaf adults who were bornwith normal hearing which they suddenly lost be-tween the ages of 9 and 11 due to various viralinfections. After becoming deaf, these individualswere enrolled in schools for deaf children where signlanguage was used. Thus, one group acquired spokenEnglish in early childhood (because they could hear)so ASL was clearly a second language for them.By contrast, the other group had acquired little lan-guage prior to learning ASL at the same older ages.The critical question was whether the two groupswould show comparable ASL comprehension. Theanswer was no. The ASL skills of the two groupswere very different. The second-language learnersperformed at near-native levels. By contrast, thelate, first-language learners showed significant ASLcomprehension deficits despite the fact that it wastheir primary language and they had been using itfor many years (Mayberry, 1993), as Fig. 4 shows.Importantly, the late-language learners were as in-telligent as the other groups in terms of non-verbalintelligence, as we discuss in detail below.

Thus, the postponement of first-language acqui-sition to ages beyond early childhood has perma-nent, deleterious effects on language comprehen-


81

Ch. 4 R.I. Mayberry

Fig. 4. ASL sentence comprehension and memory performanceof deaf adults as a function of the age at which they were firstbegan to learn ASL and whether it was their first or secondlanguage (fig. 4 from Mayberry, 1993).

sion in later adulthood. This finding was replicatedand extended in another study. Mayberry and Lock(2001) compared the English grammatical skills oftwo groups of second-language learners, one deafand one hearing, to a group of late, first-languagelearners who were deaf. The deaf second-languagelearners of English acquired ASL from their deafparents; they began to learn English as a secondlanguage in school between the ages of 5 and 7.The hearing second-language learners had a vari-ety of native languages (Urdu, Italian, German, andFrench) and began to learn English as a second-language in school between the ages of 5 and 9.By contrast, the late first-language learners began tolearn sign language and English at the same timebetween the ages of 5 and 9 when they enrolled in aschool that used sign language; they had not acquiredany language in early childhood. As predicted byprevious research, there were no performance differ-ences on the English syntax task for the two groupsof second-language learners of English. The deaf,second-language learners and the hearing second-language learners performed similarly across all theEnglish syntactic structures. However, the late-firstlanguage learners performed at low levels on all theEnglish syntactic structures and tasks (Mayberry andLock, 2001).

Together the results of these studies demon-strate that language exposure in early childhoodis necessary for language to develop fully in anylanguage, be it a first or second language. Deafchildren exposed to accessible language in earlylife show second-language grammatical and readingskills comparable to hearing children, even thoughtheir first language was in sign rather than in speech.Thus, the unique risk congenital deafness poses forinfants and children is that it prevents them fromgetting accessible language input at the right de-velopmental stage. Deaf people who, for whateverreasons, are not exposed to accessible language dur-ing early childhood (in any form, spoken or signed),consequently suffer from two permanent handicaps:they cannot hear sound and they do not readily com-prehend any language in any mode, be it in sign,speech, or writing. This second handicap, incom-plete language comprehension, is completely pre-ventable.

Educators have attempted to ameliorate the lan-guage problems common to deaf children througha variety of approaches. One commonly used edu-cational method that combines the use of signs andspeech is called Total Communication. We now turnto this means of classroom communication.

Total Communication

Two factors prompted a majority of Americanschools for deaf children to supplement a strictlyvocal mode of instruction (oralism) with signs.One factor was the dawning realization by linguiststhat the sign languages used by Deaf Communitiesworldwide were natural languages, such as AmericanSign Language, ASL, in North America (e.g., Klimaand Bellugi, 1979; Stokoe, Casterline and Cronebeg,1965). The second factor was widespread dissatis-faction with the academic achievement of deaf chil-dren taught exclusively through speech (Lou, 1988).The objective behind this philosophical change ofheart, known as Total Communication, or TC, wasto make the structure of spoken language visuallyaccessible to deaf children. The hope was that deafchildren would spontaneously acquire the grammarof spoken language by watching English in theair in much the same fashion as hearing chil-dren acquire the grammar of spoken language by

82



simply listening to English in the air (Moores,1982). 1

Manually coded englishTo implement TC, teachers and parents accompanytheir speech with a simultaneous rendition in sign.The sign portion is accomplished by borrowing vo-cabulary words/signs from ASL, signing them inEnglish word order, and then using invented signs torepresent grammatical English morphemes, such asdeterminers, plural and tense markers, and so forth.Several groups of educators began the task of invent-ing signed English morphology at the same time.Their various results are collectively known as Man-ually Coded English, or MCE (for a description, seeLou, 1988; Schick and Moeller, 1992; Wilbur, 1987).

Allen (1989) observed that, between 1974 and1983, American deaf students showed gains in meanreading and mathematics achievement, especially atyounger ages. Whether these gains were due to thewidespread adoption of TC, the greater availabilityof services for families with deaf infants, or somecombination of factors is, of course, impossible tosay. The hope that TC would lead to vastly im-proved facility in English grammar for the majorityof deaf children has not been fully realized, unfortu-nately. This is because mode of communication isnot the only factor that predicts deaf childrens En-glish learning and academic success, as we describebelow.

Geers, Moog and Schick (1984) examined theability of 327 profoundly deaf 58-year-olds to pro-duce English sentences containing a variety of earlydeveloping grammatical structures. Half the chil-dren were orally trained and half were TC trained,i.e. speech plus MCE manually coded English.Although there was considerable variation, the TCchildren as a group developed English grammar at

1 The signing used in Total Communication, commonly referredto as Manually Coded English (MCE), uses a grammar that isunlike ASL, the natural sign language used by deaf people inthe United States and most of Canada. The grammar of ASLis dissimilar from English because it evolved independent ofspoken English. ASL has evolved through generations of deafchildrens minds. By contrast, MCE was invented by committeesof educators to specifically mirror English grammar in sign. Fora more detailed description of MCE, see Wilbur (1987).

a slower rate compared to the orally trained chil-dren. However, it is important to remember thatorally trained children, as a group, are more likelyto be privileged than TC children, especially withrespect to hearing level and socioeconomic status,as described above. Leakey (1991) observed in onesample of TC trained children that they were correctless than half the time in their production of signedEnglish grammatical morphemes, such as articles,prepositions, and auxiliary verbs and tense markers.These results seem to indicate that English grammarcannot be learned through a combined speech andsign mode. Factors other than communication modemay contribute to poorer than expected English lan-guage outcome, however. These factors constituteproblems with respect to linguistic input and learn-ability. We consider these factors below.

The input problemEven when English grammar is made visually acces-sible in classrooms through the use of simultaneousEnglish and MCE, it is still often inaccessible tomany deaf children for large portions of the day,especially at home. This is the situation for deafchildren in day classes or schools when their hear-ing families make little effort to sign. For example,Mayberry et al. (2001) interviewed 24 severely andprofoundly deaf 715-year-olds with hearing par-ents who attended TC day classes in a large city.One-third of the children reported that they couldcommunicate with no one in their families becauseno family member knew or used any kind of signsor sign language with them. Another third reportedthat they could communicate with only one person intheir family, typically the mother. Less than one-thirdof the children reported that they could communicatewith family members through sign. These interviewdata suggest that the linguistic input many TC chil-dren receive, although visually accessible, is oftenrestricted to class time in contrast to the speechoverheard throughout the waking hours by hearingchildren.

Considerable success in learning English has beenreported for TC educated children when their schoolsand families have made a concerted effort to provideconsistent MCE input for the children in all aspectsof their lives. Schick and Moeller (1992) investigatedthe English and reading skills of 13 profoundly deaf


83

Ch. 4 R.I. Mayberry

children between the ages of 7 and 14. Unlike pre-vious reports, all these deaf children read at levelswithin normal limits for hearing children. In addi-tion, they performed at superior levels on a test ofEnglish grammar in comparison to norms based onorally trained deaf children (see Geers et al., 1984).Detailed analyses of the childrens expressive En-glish showed it to be very similar to that of matchednormally hearing children in terms of complexityand grammatical intactness. There was a tendencyfor the deaf children to delete some types of gram-matical morphology from their English expression,most notably those English morphemes that require aseparate sign in MCE but are fused to the root wordin spoken English (Schick and Moeller, 1992). 2

These findings show that the use of MCE can leadto high levels of functional language acquisition ina signed version of a spoken language, English inthis case, and to literacy levels typical for hearingchildren. The key factor appears to be ensuring thatthe frequency and consistency of the childs signedlanguage input is comparable in many ways to thatof hearing childrens spoken language input.

Other research shows that restricted language in-put may retard the development of signed languagecomprehension. Mayberry et al. (2001) studied 48severely and profoundly deaf 715-year-olds. Thechildren whose sign input was limited to classroomhours (that is, TC children with hearing, non-signingparents) understood stories given in sign language(both MCE and ASL) less well than children whosesign input was unlimited. The latter children had par-ents who signed with them at home, both deaf andhearing parents. The comprehension gap between thetwo groups, those with limited versus abundant sign

2 When using MCE, for example, the child would be likelyto accompany both the morphemes of the verb, going withtwo signs because each morpheme, go and -ing are stressedsyllables in speech. However, the same child would be lesslikely to accompany both the morphemes in the verb goeswith two signs because the English verb phrase is one syllablein speech. The verb goes requires two signs, however, onesign for the root go and another sign to mark the third-personsingular, -es. Thus it should be clear that this type of signmorphological deletion in childrens and adults production ofMCE is an adaptation to signing and speaking simultaneouslywhen the syllabic and stress patterns of two modalities are inconflict with respect to stress patterns.

input, became more marked with age (specifically, at1315 years as compared to 79 years). These find-ings suggest that the effects of restricted linguistic in-put on language development may be cumulative andhence not readily apparent in the young deaf child.

The facilitative effects of abundant sign input onlanguage development via MCE were also found byGeers and Schick (1988). They compared the abilityof two groups of TC children to produce a variety ofearly developing grammatical structures in English.One group of 50 profoundly deaf 58-year-olds haddeaf parents who had used sign language with themfrom birth. The other group (matched for age, hear-ing level, and nonverbal IQ) had hearing parentswhose sign skill varied considerably. The group withabundant sign input significantly outperformed thegroup with limited sign input on a test of Englishgrammatical skill (given in simultaneously presentedMCE and spoken English). Again, the differencesbetween the two input groups with respect to pro-duction of English grammar became more markedwith age. Indeed, the TC children who had receivedabundant sign input from birth produced Englishgrammatical structures (in MCE with speech) at alevel commensurate with a group of orally traineddeaf children who were previously tested by theresearchers (Geers and Schick, 1988).

To summarize, research results suggest that thecrucial factor in deaf childrens language develop-ment is not the sensory modality through which theyperceive and produce language (i.e., speech with-out sign, speech with sign, or sign without speech).Instead, the abundance and richness of consistentlinguistic input, accessible and available to the deafchild throughout early life and later schooling is akey factor in the childs cognitive outcome, bothin terms of language development and educationalachievement.

The learnability problemSome researchers pose yet another version of theinput problem, which we call the learnability prob-lem. Many linguists and educators criticize the useof MCE as the mode of instruction for deaf chil-dren because it is artificial, i.e., not a real language(Johnson, 1990; Johnson, Liddell and Erting, 1989).There are at least three linguistic and psychologi-cal consequences of this artificiality. By artificiality

84



Fig. 5. Visual acuity is greatest at the face where signers focusduring sign language perception. Visual acuity decreases sharplyas a function of number of degrees from the focus point, rangingfrom 0.50 just beside the face to only 0.10 at the shoulder andchest. Areas of low visual acuity in the signers visual focus relyon peripheral vision; peripheral vision, although poor in visualacuity, is dept at perception of movement trajectory and velocity(fig. 2 from Siple, 1978).

we mean using communication that has not evolvednaturally through the minds of generations of youngchildren, as have all natural languages, includingsigned languages.

The first consequences of using an artificial com-munication system are perceptual and phonological.Linguists now know that the signs of natural lan-guages have sublexical structure, which means thatall signs are constructed from a finite set of mean-ingless articulatory, or phonetic, units. One propertyof ASL phonological structure is that it evolved tofit the constraints of the human visual system. Siple(1978) first observed that ASL signers focus on oneanothers faces (and not on the hands and arms). Thismeans that signs made around the face fall into thefoveal area of the visual field, or central vision, asshown in Fig. 5. Central visual processes perceivesform and shape. In fact, ASL signs made around theface use contrasts of small changes in handshape andlocation. Signs made away from the face fall intoperipheral vision. Peripheral vision perceives the tra-jectory and velocity of movement. In fact, ASL signsmade away from the face use contrasts of movementpatterns.

Many of the invented MCE signs for Englishgrammatical morphology violate the perceptual fitbetween central and peripheral visual processes andthe articulatory structure of ASL signs. The poorfit between the visual processing and many MCEsigns makes it impossible for the signer to see theform of many invented signs. For example, in someMCE systems the various forms of English verb aux-iliary have, has, had are created with handshapechanges consisting of V, S, and D (finger spelledhandshapes) made inward to the chest with twohands as a derivation of the root verb root have inASL. However, when looking at a signers face, asall signers do, these handshape changes are invisible;only the movement and location of three inventedsigns can be detected, yielding only one sign ratherthan the three invented English grammatical auxil-iary verbs. This observation was predicted by studiesof sign identification in peripheral vision. Swisher,1993 (Swisher, Christie and Miller, 1989) foundthat the handshapes of unfamiliar signs could notbe recognized in peripheral vision. Supalla (1991)noted that the deaf children he observed who wereeducated with MCE tended to replace the inventedpronouns of the MCE system they used with modi-fications to the movements of the signs, reminiscentof ASL grammatical inflections that indicate person,even though the children had never seen ASL.

The second and third consequences of the linguis-tic artificiality of MCE systems are morphologicaland prosodic. Linguists now know that sign lan-guages have a strong tendency to produce grammati-cal morphology within various changes to movementtrajectories through space; these grammatical mor-phemes are produced simultaneously with the signroot rather than through a sequence of signs. Thiscreates visual intonation patterns through the air inwhich grammatical roots and morphemes are simul-taneously combined. The linear, sign for morpheme,speech/sign expression of MCE systems destroysthis visual intonation and significantly slows thepace of sign. In fact, many deaf adults remark thatthey cannot parse and grasp MCE signing. 3 Hearingadults do not have this problem because they listen to

3 Patrick Boudreault, Pamela Witcher and Patricia Viens, per-sonal communication.


85

Ch. 4 R.I. Mayberry

the speech portion of MCE and thus are not requiredto perceive MCE solely through vision.

SummaryIn asking how hearing affects the childs cogni-tive development, we have found that the primarydeficit posed by deafness is a high risk for impov-erished language acquisition in any form. We haveseen that deaf children, as a group, show heteroge-neous levels of language development related to boththe developmental timing and accessibility of theirlanguage exposure. Some deaf children show highlevels of language development (in signed or spokenlanguage) commensurate with those of hearing peersbut other deaf children show significantly delayedand depressed levels of language (in signed or spo-ken language). This state of affairs leads to two im-portant questions. First, what are the consequencesof this wide variation in language development (insigned or spoken language) for cognitive develop-ment? Second, does the sensory modality of thechilds primary language (i.e., signed versus spokenlanguage) have any specific effects on cognitive de-velopment? We begin our inquiry with a summary ofIQ test performance followed by a look at memory,symbolic play, and conceptual development.

IQ test performance

Intelligence tests were originally devised to predictacademic performance (Sternberg, 1989). Thus, oneimportant question is whether the lower than averageacademic achievement of the deaf population (rela-tive to the hearing population) is due to a lower thanaverage IQ. Two issues related to the question havebeen studied in some detail. One question is how thenonverbal IQ performance of deaf students comparesto the general population. Another issue is whetherdeaf children show unique performance patterns, thatis, cognitive strengths or weaknesses that are atypi-cal of normally hearing children. Our discussion ofthese issues will be facilitated if we first note whichIQ tests are commonly administered to deaf students.

Mean nonverbal IQ

The office of Demographic Studies at Gallaudet Uni-versity gathered IQ data on 41,000 deaf students

enrolled in special education in the United States.15,000 students had been given the performancescale of the WISC, the Leiter, and the HiskeyNebraska; the remaining students had been adminis-tered a variety of other nonverbal tests (Schildroth,1976). Average performance IQ for the general hear-ing population is 100. Mean nonverbal IQ for deafchildren with no additional handicap was 100.1.Deaf children with one additional handicap showed alower mean nonverbal IQ of 86.5 (Schildroth, 1976).

Braden (1992) replicated these findings with ameta-analysis of 285 studies that together tested171,517 deaf students from 19001988. Mean non-verbal performance IQ across the studies was 97.4with a SD of 15.33. Braden discovered that meannonverbal IQ increased as the study publication dateincreased. This improvement in IQ test scores maybe due to better societal awareness of the specialneeds of deaf children in recent years. In addition,Braden (1992) noted that methods of IQ test admin-istration have changed radically over the years. Stud-ies that reported using a strictly oral means of testadministration reported lower mean nonverbal IQscores than those that used a combination of speechand signs. The finding is logical; children performbetter on tests when they understand the directions.

On the performance scale of the WISC-R, deafboys tend to perform at somewhat higher levels thandeaf girls on subtests that are highly visual andspatial in nature (specifically, block design, objectassembly, and picture completion). Deaf girls tend tooutperform deaf boys on the coding subtest. Somepsychologists think that the coding subtest, althoughnonverbal in nature, is related to language ability.Both the direction and magnitude of the reportedsex differences among deaf students are identical tothose reported for the general population, accountingfor less than 1% of the variance (Phelps and Ensor,1987).

Hearing preschoolers are often tested with TheMcCarthy Scales. Deaf preschoolers (25 years withno additional handicaps) perform similarly to hear-ing preschoolers on the McCarthy subtests that re-quire no language, namely, block building, puzzlesolving, and draw-a-design. The performance of deafpreschoolers on these nonverbal subtests is unrelatedto the severity of their hearing loss, within a range of65110 dB (Bond, 1987).

86



Although the mean nonverbal IQ of the deaf pop-ulation approximates that of the hearing population,some research suggests that the strengths and weak-ness of performance IQ in the deaf population are notidentical to those of the normally hearing population.Such a finding would not be surprising, given the factthat deaf children must rely on vision to figure outthe world to a much greater extent than do hearingchildren. As expected, then, analyses of performanceIQ test items reveal that deaf children do not showthe same hierarchy of item difficulty across varioussubtests as compared with hearing children (Maller,1997; Zwiebel and Mertens, 1985).

Common IQ tests

The most frequently used test in North America isthe Wechsler Intelligence Scale for Children (WISC).Typically the performance scale is administered; theverbal scale is infrequently administered (McQuaidand Alovisetti, 1981). In a review of 285 studies re-porting measures of deaf childrens IQ, Braden (1992)found that 14 different nonverbal tests have beenused. Of these nonverbal IQ tests, five tests havenorms for deaf children. Watson and Goldgar (1985)compared deaf childrens performance on two of thetests, the WISC-R performance scale and the HiskeyNebraska Test of Learning Aptitude (H-NTLA). Theyfound performance on the two tests to be highly cor-related (r = +0.85). However, they also found thatthe H-NTLA showed greater kurtosis, that is, a distri-bution of scores in comparison to the WISC-R yield-ing many extreme scores, namely, scores < 70 and> 119, where 100 represents average performance.Braden (1992) found no significant differences for re-ported nonverbal IQ scores across studies that usednorms for deaf children compared with studies thatused norms developed for hearing children. This im-portant comparison demonstrates, again, that the non-verbal IQ of deaf children does not differ from that ofhearing children, as Table 1 shows. Indeed, this is arobust finding, as we discuss below.

Unique performance patterns

Verbal IQVerbal (i.e., language) tests or subtests are infre-quently used to estimate the IQ of deaf children due

to their pervasive language delay, as described in de-tail above. For example, Geers and Moog (1988) ad- ?#1ministered both the verbal and performance scales ofthe WISC-R to a group of profoundly deaf studentswho were educated orally. Despite their excellentspoken language skills however, the deaf studentsscored 22 points lower on the WISC-R verbal scale(with a mean of 89) than on the performance scale(with a mean of 111).

The verbal scale of the WISC-R has been admin-istered experimentally in sign language (in severaldialects) to deaf students. Mean verbal score in signlanguage was 96.3 when the digit span subtest wasincluded in the overall score and 97.8 when it wasexcluded (Miller, 1985). Deaf students often show areduced sequential short-term-memory span in com-parison to hearing students for word lists (includingdigits) even when the stimuli are signs rather thanwritten or spoken words, which we discuss in detailbelow (Hanson, 1982; Mayberry and Waters, 1991).Another study failed to replicate the finding of a nor-mal verbal scale score and reported lower than aver-age verbal scale scores on the WISC-R for a groupof 27 deaf children when test administration was invarious kinds of sign language (Chovan, James andBenfield, 1994). These conflicting results probablyreflect real differences in the sign language skills ofthe deaf students tested in the two studies. Level ofsign language development, and hence level of signlanguage comprehension and production, is hetero-geneous in the deaf population and thus needs to becontrolled in such studies, as explained above. Maller(2001) administered both the performance and verbalscales of the WISC-III to 110 deaf children rangingin age from 816 using Pidgin Sign English thatwas mostly ASL (Maller, 2001: 5). The deaf chil-dren with deaf parents significantly outperformed thedeaf children with hearing parents on the verbal scaleof the WISC-III with mean scores of 86.48 and 77.19respectively. Note that the verbal scale performanceof these deaf signing children was comparable to thatof the orally educated deaf children discussed abovereported by Geers and Moog (1988). ?#2

For the hearing population, verbal and perfor-mance scale scores on the WISC-R are correlated(r = +0.50). Verbal and performance IQ are alsocorrelated for deaf students (r = +0.45, Craig andGordon, 1988; Watson, Sullivan, Moeller and Jensen,


87

Ch. 4 R.I. Mayberry

TABLE 1

Intelligence tests used to assess deaf and hard-of-hearing people a

Name of test N studies Mean IQ

Performance testsChicago Non-Verbal Examination 5 97.25GraceArthur Performance Scale 16 96.02HiskeyNebraska Test of Learning Aptitude 17 97.53KaufmanAssessment Battery for Children 6 96.86Leiter International Performance Scale 12 87.19Ontario School Ability Examination 6 98.44SnijersOomen Nonverbal Tests 5 96.71WAIS-R Performance Scale 9 102.84WechslerBellevue Performance Scale 11 107.32WISC Performance Scale 38 101.22WISC-R Performance Scale 44 100.81

Motor-free nonverbal testsDraw-a-Man/Person 13 91.72Pintner Non-Language Test 13 91.87Ravens Progressive Matrices 17 97.56

Verbal testsWAIS-R Verbal Scale 5 8436

a Published reports of IQ tests used during the past 50 years, table 1 from Braden, 1992.

1982). Performance on both the WISC-R and WAIS-R performance scales are also correlated with aca-demic performance (Paal, Skinner and Reddig, 1988;Watson et al., 1982). Indeed, the correlation is im-pressive given the depressed language skills of mostdeaf students.

Uneducated deaf childrenThe assumption underlying the large-scale studies ofIQ discussed above is that the deaf children understudy are, or have been, enrolled in an educationalprogram appropriate for deaf children at from age5 or 6 onwards as would be the case for normallydeveloping hearing children. The expectation of nor-mal, nonverbal IQ test performance clearly cannotbe applied to deaf children who are educationallydeprived. This is shown by case studies of deaf chil-dren who immigrated with their families to NorthAmerica in late childhood or adolescence after hav-ing attended no school in their home country or aschool for hearing children with no accommodationsor interventions for their deafness. For example,Grimshaw and her colleagues (Grimshaw, Adelstein,Bryden and MacKinnon, 1998) investigated the de-velopment of a deaf boy raised in Mexico where he

did never attended school. His WISC-R and WAIS-Rperformance levels were 85 with above average per-formance on the Object Assembly subtest and verylow performance on the Coding and Picture Arrange-ment subtests (specific scores were not given). An-other case study of a 16-year-old deaf girl raised inGuatemala where she never attended school reporteda nonverbal IQ level of 8 years based on perfor-mance on the Leiter and unspecified subtests of theWAIS (Emmorey, Grant and Ewan, 1994). Morford(2001) reported the performance by two additionaldeaf students. One 13-year-old girl performed atthe 5th percentile on the Ravens after having at-tended school for only 3 months. After 32 monthsof educational experience, she performed at the 50thpercentile on the Ravens. Finally, a 12-year-old boy,who had attended a school for hearing children in hishome country (but without any special considerationof his deafness), received a nonverbal IQ of 71 onthe WISC-R after 19 months of being enrolled in aNorth American school for deaf children. After 31months of special schooling he received a nonverbalIQ of 85 on the same test.

Together these case studies show that the av-erage nonverbal IQ performance reported for the

88



deaf school-aged population is a consequence of ap-propriate educational experiences at the appropriateyoung ages despite often severe language delay. Deafchildren who suffer from educational deprivation donot show normal nonverbal IQ performance.

Higher than average nonverbal intelligenceSeveral investigators have observed that a subset ofthe deaf population shows higher than average per-formance on nonverbal IQ tests. In a study of morethan 1000 deaf students, Sisco and Anderson (1980)observed that deaf students who had deaf parentsoutperformed their deaf peers who had hearing par-ents on every subtest of the WISC-R performancescale (with an overall mean performance IQ of 107as compared to 96). Moreover, the deaf studentsraised in deaf families performed at higher meanlevels compared to the hearing population meanon every subtest of the WISC-R performance scale(with a mean score of 107 as compared to 100).

Several explanations have been proposed to ac-count for the higher than average nonverbal IQshown by deaf children raised in deaf families.One hypothesis is that the performance differencereflects the fact that intelligence is, in part, inherited.This account, which we call the genetic hypothesis,argues that, because at least half of intelligence is in-herited, deaf children who have deaf parents performhigher than average on nonverbal IQ tests becausethey have inherited genes for superior intelligence(for example, see Kusche, Greenberg and Garfield,1983). Underlying this proposal is the idea that thedeaf individuals who have been the most successfulat reproducing over multiple generations are also themost intelligent.

The second explanation, which we call the earlylearning hypothesis, emphasizes the impact of thechilds early environment on cognitive development.Some researchers speculate that early language ex-posure facilitates intellectual development in general(e.g., Vernon, 1968). Another version of this hy-pothesis is that deaf parents are better prepared thanhearing parents to meet the early learning needs ofthe deaf child (e.g., Schlesinger and Meadow, 1972;Sisco and Anderson, 1980). Underlying both expla-nations is the premise that deaf children with deafparents perform higher than average on nonverbal IQtests because they have acquired language on sched-

ule (in early childhood). In a third version of thehypothesis, other researchers propose that the effectrelates to the physical form of the language beinglearned in early childhood. They posit that learninga visuospatial, or three-dimensional grammar, suchas in ASL, boosts the childs visual and spatial abil-ities to higher than average levels (Bellugi et al.,1990).

The available data do not clearly adjudicate be-tween the genetic and early experience hypothe-ses, although recent research has shed light on thequestion. As previously noted, Sisco and Anderson(1980) found that genetically deaf children (thosewith deaf parents) performed at higher mean levelson the WISC-R performance scale than the generalpopulation. However, Conrad and Weiskrantz (1981)only partially replicated the finding. They comparedthe performance of two groups of British deaf chil-dren whose deafness was genetic (those with deafparents and those with hearing parents but deaf sib-lings) using norms from the British Ability Scales.Both deaf groups performed at higher than averagelevels than British hearing children. This finding issimilar to the American results. Next they comparedthe performance of the two groups of deaf childrenwith an age-matched group of hearing children froma private school. The performance of two deaf groupswas indistinguishable from that of the hearing groupon the Ravens Progressive Matrices (a test of vi-sual reasoning), Block Design, and a test of visualdesign recall (Conrad and Weiskrantz, 1981). How-ever, Paquin (1992) argued that the hearing controlgroup used by Conrad and Weiskrantz (1981) wasnot representative of the hearing British school-agedpopulation but rather was a select sample from asmall private school that probably selected studentsfor above-average intelligence.

In a subsequent study, Kusche et al. (1983) repli-cated the original American findings (Sisco and An-derson, 1980). They compared the WISC-R perfor-mance scores of two groups of genetically deaf chil-dren (those with deaf parents and those with hearingparents but deaf siblings) to two matched groups ofnongenetically deaf children (those with hearing par-ents and hearing siblings). The groups were matchedon a number of factors including sex, age, hearinglevel, grade, and mothers educational level. The twogroups of genetically deaf children had higher mean


89

Ch. 4 R.I. Mayberry

performance scores on the WISC-R than the hearingpopulation mean (112 as compared to 100). The twogenetically deaf groups also showed higher meanperformance compared with the two nongeneticallydeaf groups (112 as compared to 101). These find-ings were partially replicated in another study (Craigand Gordon, 1988).

The higher than average performance IQ of thegenetically deaf groups studied by Kusche et al.(1983) cannot be attributed to knowledge of sign lan-guage, because all the deaf groups, genetically deafor not, knew and used sign language. This suggeststhat some portion of the higher than average nonver-bal IQ of genetically deaf children may be inheritedor that nongentic deafness can be associated withcognitive problems of some sort related to the causeof deafness. One way to test this hypothesis wouldbe to test the performance IQ of the deaf childrensparents and siblings.

Paquin (Paquin, 1992) investigated the genetichypothesis by testing the nonverbal IQ of 31 familiesin which both the parents and the children were deaf.He obtained nonverbal IQ scores for both parents in18 families, one parent in 13 families and 50 deafchildren of the 31 families. He used either the WISC-R or the WAIS-R nonverbal performance scales.Performance on the two scales has been found toproduce the same score (Braden and Paquin, 1985).Paquin found mean nonverbal performance to be114.5 for the deaf parents and 114.0 for the deafchildren. Replicating the previous American studies,the performance of the deaf parents and their deafchildren was significantly above the mean of thehearing population (Paquin, 1992).

As Paquin (1992) notes, there are two possi-ble mechanisms of genetic transmission of genesfor deafness and nonverbal IQ; these mechanismsare assortative mating and pleiotropism. Assortativemating would produce higher nonverbal IQ amonggenetically deaf individuals if brighter deaf peoplehave tended to marry other bright deaf people acrossgenerations. Pleiotropism would produce higher IQamong genetically deaf people if genes for bothdeafness and intelligence actually became linkedover time and were expressed together. Such geneticlinkage would predict that genetically deaf childrenwould show higher than average nonverbal IQ butthat their hearing siblings would not because the two

genes are fused in expression (Paquin, 1992); thislatter prediction has not been tested.

A major problem for the genetic explanation ofthe higher than average nonverbal IQ of deaf indi-viduals with deaf parents, however, is that there over100 different kinds of genetic deafness, both domi-nant and recessive (Boughman and Shaver, 1982). Itis improbable that all genetic deafness is associatedwith higher than average performance IQ. However,this does not rule out the possibility that some typesof genetic deafness may be linked with some typesof intelligence.

Another possible explanation for the higher thanaverage nonverbal IQ of genetically deaf individualswas investigated by Braden (1987). He tested boththe manual reaction time and manual movement timeof three groups of adolescents, genetically deaf, non-genetically deaf, and normally hearing; he also testedthe groups performance on the Ravens ProgressiveMatrices. The reaction time measure required partic-ipants to lift a finger in response to a given light; themovement time measure required participants to pusha button below a given light. The deaf adolescentswith deaf parents outperformed the other two groupson the reaction time measure. In addition, both deafgroups outperformed the normally hearing studentson the movement time measure. The performance ofthe genetically deaf and hearing groups did not differon the Ravens Progressive Matrices but the nongenet-ically deaf group performed at a lower level.

Factors related to language and early environ-ment rather than to genetic background can explainthese findings. First, it is important to know that theRavens Progressive Matrices is an untimed reason-ing test that has been found to be related to someaspects of language skill. In a study of 468 deaf stu-dents between the ages of 15 and 16, Conrad (1979)found that performance on the Ravens was highlycorrelated to measures of phonological decoding inmemory for written words, i.e., to mental repre-sentations for vocabulary. Other research has foundthat performance on the Ravens is highly related toworking memory (Carpenter, Just and Shell, 1990).Together these findings explain why the deaf chil-dren from deaf families performed similarly to thehearing group on the Ravens test. The deaf childrenfrom deaf families likely had a native sign languageand hence developed mental language representa-

90



tions in addition to working memory (as we discussin detail below). This would also explain why thedeaf children from hearing families performed morepoorly than the other two groups on the Ravens; theywere likely to have experienced late and/or incom-plete language exposure in sign or spoken languageand thus have underdeveloped mental language rep-resentations and working memory. The finding thatboth the deaf groups outperformed the hearing groupon the manual reaction time could be due to the factthat both groups used sign language as a primary lan-guage. They were highly practiced at moving theirhands quickly. Finally, the finding that the deaf stu-dents from deaf families outperformed those fromhearing families on reaction time to a visual stim-ulus could be due to the effects of exposure to avisual language in early childhood, something thedeaf group from hearing families lacked. Researchhas found that exposure to sign in early life enhancessome aspects of visual processing, as we discussbelow.

Visuospatial skills and memory

Supernormal visual skills

Research investigating the visuospatial skills of deafand hearing individuals has uncovered numerouspositive effects of using sign language. For exam-ple, Bettger, Emmorey and Bellugi (1997) found thatindividuals who used ASL, native deaf signers, na-tive hearing signers, and non-native deaf signers, allperformed more accurately on the Benton Test of Fa-cial Recognition than did hearing non-signers. Theadvantage was especially apparent on shadowedfaces, that is, stimuli that were harder to perceive.Other research has found that deaf children whoknow no sign language do not show performanceadvantages on the Benton Test of Facial Recogni-tion (Parasnis, Samar, Bettger and Sathe, 1996). Thepositive effects of using sign language on face recog-nition are not limited to native signers. Non-nativedeaf signers and non-native hearing signers (teach-ers) were observed to require fewer trials to com-plete a face matching task compared with hearingnon-signers (Arnold and Murray, 1998). As few as 2years of sign language learning produced increasedaccuracy of facial expression identification in hear-

ing individuals, especially of facial expressions thatsignal emotions of sadness and disgust. These fa-cial expressions are harder for general population toidentify (Goldstein and Feldman, 1995).

Learning and using sign language has also beenfound to sharpen the visuospatial abilities of rec-ognizing movement patterns and generating and ro-tating mental images. For example, Chinese deafchildren who use sign language were found to per-form at higher mean levels than age-matched Chi-nese hearing children on memory for the movementsof Chinese characters drawn in the air (Fok, Bellugi,Van Hoek and Klima, 1988). Both deaf and hearingnative signers were found to generate and rotate vi-sual images more quickly than hearing non-signers(Emmorey, Kosslyn and Bellugi, 1993). Moreover,deaf individuals who do not know sign language donot show these effects (Chamberlain and Mayberry,1994). The degree to which these effects are contin-gent upon early sign exposure has not yet been fullyexplored (Emmorey, 1998). Also, it is important tonote that the degree to which these effects interactwith the well-documented, but small, differences inspatial processing between males and females hasnot been adequately investigated. In some studies,the differences between the experimental groups ofhearing and deaf participants are more apparent formales than for females. Thus, future research needsto control for both age of sign language acquisitionand sex. We discuss the brain correlates of thesefindings below after turning to memory skills.

Short-term memory

As early as 1917, Pintner and his associates found aconsiderable discrepancy between the digit spans ofdeaf and hearing children. The oldest deaf subjectsin their studies had a mean digit span three digits lessthan the youngest hearing subjects. (For a review seeChamberlain and Mayberry, 2000). He concludedthat deafness causes a delay in development whichhe called mental retardation (Pintner, 1928). Hoe-mann (1991) noted that more than four decades ofresearch was required to refute this erroneous claim.

Recent investigations into the memory processesof deaf individuals explain Pintners early findings,which have been consistently replicated. The digitspan of hearing subjects typically exceeds the digit


91

Ch. 4 R.I. Mayberry

span of deaf subjects, even when the digits aregiven in a written or signed presentation. Researchhas found that auditory experience does influenceSTM span but indirectly, not directly. Audition af-fects STM span because it plays a major role inspoken language acquisition. Deaf children varywidely in their level of language development (spo-ken or signed), familiarity with written words, andin whether and how they use mental rehearsal as astrategy to facilitate recall.

Language development (or more simply, famil-iarity) determines how well words from a givenlanguage can be recalled. Individuals can recall morefrequently occurring words than infrequent ones; in-dividuals who speak two or more languages canrecall more words from their native language thantheir second, or weaker, languages. Deaf childrenoften show reduced STM spans for written words incomparison to age-matched hearing children simplybecause they are less familiar with English words.The STM spans of congenitally deaf children forwritten words are thus predicted by the dual fac-tors of degree of hearing loss and degree of lan-guage development (Conrad, 1970, 1979; Hamiltonand Holtzman, 1989; Novack and Bonvillian, 1996).When the linguistic element is removed from theSTM task, and deaf children are asked to recallnonsense figures, their performance is equivalent toage-matched hearing children (Furth, 1966). Whendeaf childrens visual strengths are tapped by askingthem to recall the spatial array of a set of items,they often show equivalent (Logan, Mayberry andFletcher, 1996) or superior performance in compar-ison to age-matched hearing children (Belmont andKarchmer, 1978). On spatial memory tasks, such asCorsi blocks, deaf children raised in deaf familieshave been found to outperform those from hearingfamilies (Wilson, Bettger, Niculae and Klima, 1997).

Whether deaf children use rehearsal (mental rep-etition of the items to be recalled) as a strategy inmemory processing is another important factor thatexplains observed STM span differences betweendeaf and hearing children. Bebko (1984) found thatboth orally and TC trained deaf children laggedseveral years behind hearing children in their devel-opment and use of spontaneous rehearsal strategiesin serial STM tasks. Rehearsal development, in turn,was predicted by the childs language history. The

probability that a deaf signing child would rehearseitems during a STM task was thus related to thenumber of years he or she had used sign language(Bebko and McKinnon, 1990).

Even at the post-secondary level, STM discrepan-cies persist (Bellugi, Klima and Siple, 1975; Bonvil-lian, Rea, Orlansky and Slade, 1987) but the exactcause of these discrepancies is unclear. For example,deaf and hearing college students show similar STMspans when recall is free, or unordered (Hanson,1982). When recall is serial, or ordered, the spa

Cognitive development in deaf children: the interface of...

Documents

Transcript of Cognitive development in deaf children: the interface of...