FACTORING OUT SIGN LANGUAGE FROM OTHER EPHEMERAL VISUAL REPRESENTATION

THE VERY IDEA that ephemeral visual representation might constitute language has not sat well with linguists during most of their discipline’s history. In chapter 1, I quoted Bloomfield as saying that “gesture languages are merely developments of ordinary gesture” and that “any and all complicated or not immediately intelligible gestures are based on the conventions of ordinary speech” (1933:39). In a similar vein, Crystal asserts that

Crystal’s remarks raise two questions: Is there any distinction between gesture in the ordinary language sense of raising one’s eyebrows in surprise or shrugging one’s shoulders to indicate ambivalence, and using ephemeral visual representation in lieu of spoken language? And if such a distinction does exist, does the latter visual system lack anything “really like grammar,” or is the “vocabulary” necessarily small? If our minimal definition of language (chapter 3) entails both lexical and grammatical components (not to mention productivity), then any claim that a gestural system is “linguistic” entails our identification of such components.¹ Of course, the two issues go hand in hand, since the identification of discrete “lexical” elements that can be recombined systematically is one of the primary grounds for distinguishing between linguistic and nonlinguistic gesturing.

Where does gesturing end and sign language begin? The problem becomes particularly complex in that sign languages mainly grow out of “natural” human gesturing, and new signs that derive from pantomime are commonly added to sign languages. To the extent, however, that gestures are modified to conform with an identifiable class of structural conventions, the ephemeral visual representation is no longer mime but language. This process is analogous to the naturalization of a borrowed word in spoken language, whereby a word’s phonological or morphological patterning is altered to conform with the patterns of the recipient language. Witness the Anglicization of Don Quixote to [dan kwíksət], the replacement of the Latin plural aquaria with aquariums, or the malapropism Tempus fugits.

Sign language does indeed differ from gesture in a number of clearly identifiable ways. And, just as diverse speech communities develop differing structural conventions, sign languages differ in structure from one another. However, neither of these points is wholly accepted in the literature. Rather, it is often claimed that human gesturing is all of a piece, being universal across cultures. Underlying this assumption of universality is the further implication that the imposition of conventions that distinguish among gesturing groups is logically—and pragmatically—unnecessary.

THE ISSUE OF UNIVERSALITY

The purported universality of gestures has long been a favorite target of anthropologists. Weston La Barre explains, for example, that

Even the act of pointing with one’s finger is not universal:

Gordon Hewes, however, has defended the notion of universality. He has not claimed that all gestures will be universally interpreted the same way; instead, he suggests that, in an actual communicative context when it is necessary for humans to understand one another, they will gesticulate in such a way as to get their messages across. Using descriptions of voyages that refer to the use of gestures between peoples who do not speak the same language, Hewes argues that

An even stronger stance on universality has been taken by Garrick Mallery, and, derivatively, by Macdonald Critchley. They have argued that sign language itself is universal and perhaps indistinguishable from what others have seen as gestural communication:

Critchley supports his thesis with reports that members of diverse signing communities have been able to communicate with one another with little difficulty:

And again,

There is some evidence supporting Critchley’s position. Deaf signers from diverse cultural communities (e.g., at the Deaf Olympics or at the World Congress of the Deaf) have informally reported that they are able to adapt their signing to be able to understand one another (at least on a basic level) after relatively little contact.² However, such observations by no means settle the issue of universality.

As Stokoe points out, conclusions about commonalities between signing systems (e.g., between American Indian Sign and American Sign Language) are dangerous if based only on fragments of each system:

The fact that a number of members of the American and European deaf communities have felt it necessary to create an international sign language, Gestuno (see World Federation of the Deaf, 1975), should suggest that the sign languages of different deaf communities are not nearly as compatible as superficial evidence might suggest.

Furthermore, commonality in perception guarantees nothing about commonality in production. Just because members of different signing communities manage to understand one another’s signs does not mean that the signs themselves are the same. Consider the signs for “tree” in American Sign Language, Danish Sign Language, and Chinese Sign Language. As Bellugi and Klima (1976:522-524) point out, all three signs are iconically based in that they represent a property associated with trees (see figure 7.1). However, this property is different in each instance. In American Sign Language,

With the Danish sign,

Finally, in Chinese sign,

And yet, despite the differences in production, an American signer—or even a nonsigner—can understand the Danish or Chinese sign (and vice versa) with little instruction.

Perhaps the most important factor concerning the universality argument is the lack of evidence that the signs used by interlocutors not sharing a common language are the same ones that signers would employ in signing to a member of their own linguistic community. It is well-documented in spoken language that fluent speakers often speak differently to children or foreigners than they do to people known to be fluent in their language (see Ferguson, 1975). More to the point is the evidence of trade jargons and pidgin languages that arise when interlocuters do not share a common language. We know that such interface languages are not lexically and grammatically equivalent to either of the source languages (see chapter 5). Before using contact language as evidence of the universality of sign, we need careful comparisons of the contact communication with normal, intracommunity language use.

Whatever else it may be, sign language is not universal. Moreover, I have suggested that sign language differs in definable ways from gesture. What are these ways?

Fig. 7.1 Sign for Tree in Three Sign Languages © Copyright Bellugi and Klima 1976:523)

INFORMAL CRITERIA FOR DISTINGUISHING SIGN FROM PANTOMIME

Even to the untrained eye, transmission in sign language and in pantomime of the same message show strikingly different characteristics. Consider the signed and the mimed responses to the instructions, “Explain to someone how to make scrambled eggs.” The extent to which the body is involved in the transmission is much greater in mime than in sign. Signing space (at least in American Sign Language) is largely restricted to an area stretching from the head to the midriff, across the shoulders, and not more than about a foot in front of the body. The mime, on the other hand, is free to use any part of his body. Signing is typically done while standing (or sitting) in one place; the mime typically moves from place to place in formulating the message.

Another difference is in the way in which the information is conveyed. In mime, the activity itself is reenacted, while in sign the activity is described. From this, two further differences follow: the movements in mime are more iconic than those in sign (although the amount and the degree of iconicity in sign is also relative—see “Iconicity and Learn-ability”), and the quantity of contextual detail given in mime is much greater. The mime, for example, must first establish the existence of a cabinet and then open its door and remove a container in order to then indicate “add salt.” The signer needs only to convey the instruction “add salt.” Thus it generally takes longer to mime a message than to sign it.

FORMAL CRITERIA FOR DEFINING SIGN AS LANGUAGE

Distinguishing sign from gesture is not enough. We must establish not only that signing behavior has distinct properties but also that it constitutes human language.

What are signs used to represent? The traditional denial of the status of “language” to sign systems seems to have arisen, at least in part, from a confusion about what is being analyzed. Linguists have generally assumed that sign language necessarily represents the spoken language of the speech community associated with a signing community. That is, sign has been looked on as a first-order representation of speech and only a second-order representation of experience (chapter 6 discussed the same presupposition with respect to writing). In the case of American Sign Language, the associated spoken language would be English; American Plains Indian Sign Language would be associated with the spoken languages of the tribes using sign. However, sign languages of deaf communities, at least, have distinctive structures that are largely autonomous from the spoken languages of culturally related speech communities (see “Prolegomena to a Functional Typology of Signing Systems”).

More serious problems arise when we turn to nonarbitrary criteria for human language. The discussion in chapter 3 (formal criteria for defining human language) identified three properties that any language must have: a finite number of discrete lexical items, a finite number of combinatory rules, and the potential for combining these elements into a limitless number of strings. Do sign languages meet these criteria? In trying to answer this question, I shall use as reference point American Sign Language (ASL), an independent sign language par excellence. If ASL fails to meet these criteria, then other signing systems would be expected to fail as well.

In chapter 3,1 noted that, while linguistic theories specify the above criteria for languages, they do not establish a minimum or a maximum number of lexical items or syntactic rules a system must have to be considered a language. This discussion of speech and writing illustrated some of the problems inherent in quantitative criteria. In chapter 5, I argued that trade jargons are not distinct languages (that is, distinct from their contributing languages) because they lack independent syntactic characteristics. On the other hand, in chapter 6, it became evident that there was no way of arguing that a predetermined minimal amount of phoneticism was necessary before a durable visual representation scheme could be considered a language.

ASL has distinct lexical items and its own system of combination rules—but does it have enough of either to be called a language? Crystal and Craig claim that it does not. They judge the lexicon of ASL to be only a fraction of that of English:

Moreover, the authors believe that such a disbalance is inherent in the visual representational modality itself:

As further evidence for this hypothesis, Crystal and Craig suggest that

There are several problems with such an argument. Crystal and Craig offer no evidence that human visual perception is incapable of discriminating (or of remembering these discriminations) between a much larger number of signs than the presumably current 3000 in ASL. Furthermore, as the level of education and the amount of geographic mobility of the deaf community increase, the lexicon should also. A simple example is the frequent development of new signs at Gallaudet College, necessitated by the confrontation with new subject matter. It is doubtful that the signing vocabulary of a graduate of Gallaudet is limited to 3,000 items.

Another problem concerns Crystal and Craig’s covert assumption that in sign languages signs are unrelated to one another, rather than falling into semantic classes. In fact, sign languages do use semantic categorization. For example, ASL signs indicating political roles are made with similar movements at similar parts of the body; they are differentiated by handshape. The signs for “king” and “queen” are both made by moving the right hand from the left shoulder across the chest to the right-hand edge of the waist, thus indicating the sash worn by royalty. In the signing of “king,” the moving hand assumes an alphabetic “K” handshape, and for “queen,” a “Q” (see figures 7.2 and 7.3).

Crystal and Craig appear to assume that the only way to increase the size of the lexicon is to add independent, nonderived words. One of the most productive ways in which the English language increases its vocabulary is by borrowing words outright from other languages (sputnik, pajamas, assassin) or by borrowing elements from other languages and creating nonce forms (telephone, psychiatrist). However, as suggested in chapter 5, these are not the only options available to a language. Natural languages such as German and pidgin languages such as Neo-Melanesian create new words by joining together, either as words or phrases, morphemes that already exist in the language. One productive means of creating new signs in ASL is the formation of compounds. For example, the sign for “home” is made up of the signs for “eat” and “sleep” (see figure 7.4). These compounds often become abbreviated with time (see figure 7.5), a phenomenon parallel to the phonological reduction of compounds in spoken language (e.g., New + found + land is pronounced as [nufənlən] in American English).

Fig. 7.2 American Sign Language Sign for “King”

Fig. 7.3 American Sign Language Sign for “Queen”

So much for arguments about the size of the lexicon. What about syntax? Are there “enough” syntactic rules in ASL to warrant calling it a language? The arguments here are less easily identified. Perhaps because of the newness of sign language linguistics, along with a predictable degree of differences of opinion in a new undertaking, many linguists persist in their earlier impression that ASL has little or no syntax. Crystal and Craig remind us that

Furthermore, they reiterate the (by now) familiar assumption that some syntactic rules—word order in particular—are considered to be more critical than others:

Fig. 7.4 Components of the Formal American Sign Language Sign for “Home” (© Copyright Bellugi and Klima 1976:537)

Fig. 7.5 Abbreviated, Informal American Sign Language Sign for “Home” (© Copyright Bellugi and Klima, 1976: 537)

This clearly implies that if a representation scheme lacks word order constraints, it is not a language.

The issue of word order in ASL is tricky and has already generated much debate (e.g., S. Fischer, 1975, Edge and Herrmann, 1977). It still remains to be seen which ordering conventions ASL might have, and what means other than linear word order ASL might use in place of the word ordering conventions familiar in spoken languages. Until this is done, it is premature to assume that ASL lacks “sufficient” syntax for meriting the label language.

Some linguists working on ASL have confronted skeptics with the argument that ASL is not the only language which lacks the grammatical features that speakers of English presuppose other languages will have. Wilbur (1976) points out that many of the differences between ASL and English are similar to differences between contemporary English and other spoken languages. For example, unlike English, many of the worlds’ languages (ASL, Kwa languages in Africa, Thai) do not have grammatically marked passive constructions, and passives are quite uncommon in Japanese and Czech. The absence of the present tense copula in ASL is paralleled by a similar absence in Bengali and Russian; the absence of definite and indefinite articles in ASL is matched in languages like Russian, Yoruba, Hindi, and even Old English.

In addition to lexical and grammatical complexity, there is the issue of productivity. Is ASL potentially as “productive” as bona fide spoken or written languages? Since potentiality cannot be measured by the actual number of utterances produced, we need to ask whether there are structural limitations on the kinds of utterances that can be produced in ASL.

Advocates of ASL have argued that there are not. In Stokoe’s words:

Crystal and Craig find this claim to be “premature” and to “hide massive methodological problems” (1978:149). The most fundamental of these problems, in their eyes, is that ASL lacks an adequate amount of displacement (the ability to use a sign for a referent that is distant in time or space) to make indirect reference. Recall that Hockett defines displacement as one of the essential “design features” of human language (see figures 3.1 and 3.2). Crystal and Craig believe that

As evidence, the authors observe that many signs

But what relevance does this phenomenon have for the notion of displacement? Displacement is concerned with referring to something that is not visually present. As ‘long as the object I am talking about carrying is not physically present—regardless of how its identity affects the shape of my sign, we still have a perfectly good example of displacement. Moreover, we know that many languages of the world lexically distinguish verbs on the basis of other elements in the sentence. Russian has two wholly distinct series of movement verbs, one for talking about going by foot, the other for talking about going by vehicle. Either we could consider the variants in the sign for “carry” as distinct lexical items (which will help push the size of the ASL lexicon above 3000 items), or we can see these formational variants as the sign equivalent of derivational suffixes indicating a modulation of a word’s basic meaning. In either event, displacement is not at issue.

Crystal and Craig also argue that sign languages lack displacement because many signs depend on accompanying facial expression or body movement for their meaning. In Israeli Sign Language, for example, the sign for “lemon” is manually identical to the sign indicating “tomato.” The signs are differentiated by accompanying the sign for “lemon” with a facial expression indicating “sour” (Cohen, Namir, and Schlesinger, 1977:246-248). Yet what does the use of nonmanual components of signing have to do with the issue of displacement? As long as the facial expression or body movement does not restrict the range of referents to things present here and now, the fact that the interlocutor must look at the signers face and body, as well as at his hands, to comprehend the sign would seem to be irrelevant.³

Have Crystal and Craig seriously questioned the linguistic status of ASL on the basis of inadequate lexicon, syntax, or productivity? Probably not; however, until more work has been done on ASL, it will be difficult to know whether their arguments can be permanently laid to rest.

CLASSIFICATION OF SIGN LANGUAGES

I have been speaking, so far, of “sign language” almost as if there is one unified language type. Occasionally I have referred to differences between sign languages (e.g., American Plains Indian Sign vs. sign languages of deaf communities) but have not defined what these differences are.

There are at least four ways of distinguishing among sign languages: two rest upon structuralist principles, a third classification has characteristics of both structuralist and functionalist approaches to typology, the fourth is wholly functionalist in orientation.

STRUCTURAL PERSPECTIVES

Genetic Classification

For sign languages of deaf communities (see “Prolegomena”), it is possible to develop family trees comparable to those used to show relationships between spoken languages (see figure 5.1). Stokoe (1974) has suggested the outlines a genetic classification might take. In the family tree of sign languages shown in figure 7.6, we see that while French and American sign languages are directly connected, ASL and British Sign Language are not.

Such genetic reconstruction has its drawbacks. Attempts at reconstruction largely depend on written information. Given the low status of signs (and signers) in most countries, much of the necessary information is unavailable. Adequate family trees are therefore difficult to develop.

Furthermore, genetic analysis is of limited appropriateness for classifying sign languages of the world. Unlike spoken languages, which can nearly all be traced to some (or many) other spoken languages, certain sign languages have arisen de novo, unrelated to any other sign systems. Particular gestures may not be universal, but the act of gesturing is. There is no reason why systematic signing could not have evolved independently a number of times. If this is so, then an ideal genetic classification might not go far beyond a list of all known signing systems, yielding little additional information about them.

Fig. 7.6 Genetic Classification of Sign Languages: The French Branch (from Stokoe, 1974:366)

Formational Classification

Sign systems may also be classified in terms of formational properties. This approach is largely comparable to the typological classification of spoken languages. Some of the formational questions we might ask are: What are the possible hand configurations? What are the boundaries of the signing space? What constraints are there on forming new signs? A different type of formational question, however, concerns what is being represented: Is it things or ideas (i.e., reality or one’s perception of it), or is it language? In the former case, each hand configuration represents what would roughly be comparable to a single word in speech, while in the latter, speech sounds or grammatical morphemes may be represented as well. In the latter cases, finger spelling is employed (e.g., to spell out the word perplex or to indicate with a “D” handshape the past tense of a verb).

Classification Along a Sign-Speech Continuum

Another way of looking at the difference between the use of hand-signs and of finger spelling is to view sign languages along a continuum from full autonomy of the sign language to the use of signs as a near “transliteration” of speech. The idea of a speech-sign continuum can be seen as far back as the middle of the eighteenth century, when Abbé Charles de l’Épée “invented” French Sign Language, the system from which American Sign Language has largely derived.

Épée (1776) distinguished between what he called le langage des signes naturelles (“the natural language of signs”) and signes méthodiques (what Stokoe, 1978:5, takes to be a metalanguage). The signes naturelles were those which Épée observed already in use among his deaf pupils. They were, however, insufficient for teaching his students French language and culture; he needed an additional set of signs to indicate such grammatical aspects of French as tense and articles. Whenever possible, he adopted (or adapted) natural signs for his purposes. One such case is the sign méthodique for indicating tense:

In other cases, Épée invented signs himself:

Épée could, in principle, have taken his manual representation of French one step farther: rather than using signs to indicate grammatical morphemes (not to mention the objects, properties, and activities indicated by natural signs), he could have used finger spelling to represent French letter for letter. In fact, a manual alphabet had been introduced into France by Jacob Rodrigues Pereira (from Portugal) some years earlier. Épée rejected the system because he felt it was possible for deaf students to be taught great skill in encoding and decoding French with a manual alphabet without really understanding any language.

Two points are actually at issue in distinguishing between what Épée identified as le langage des signes naturelles, signes méthodiques, and finger spelling: whether manual signs are used to signify experience or spoken language and whether the signs follow the syntactic ordering of the associated spoken language or have their own syntax. The combination of these two variables gives rise to at least the following possible systems:

only natural signs

indigenous syntax

only natural signs

syntax of associated spoken language

natural signs plus grammatical signs which are based on experience

syntax of associated spoken language

natural signs plus grammatical signs which are based on associated

   spoken language

syntax of associated spoken language

finger spelling based on associated spoken language

syntax of associated spoken language

In the United States, all of these possibilities are in use.⁴ The first type we commonly refer to as ASL (although see the discussion of ASL in “Types of Signing Communities”). The terms signed English or manual English are commonly (though inexactly) used to refer to the next three types. The second type is most nearly represented by a form of signing called Siglish, which strings together ASL signs in English word order, but makes no attempt to represent English inflectional or derivational morphology. The third and fourth types are represented most nearly by Seeing Essential English, Linguistics of Visual English, Manual English, Signing Exact English, and Signed English.

Seeing Essential English (or SEE 1, as it has come to be called) was invented in the 1960s by David Anthony and was the first concerted attempt to adapt American Sign Language to English syntax. (Earlier, the Paget-Gorman Systematic Sign Language used British Sign Language in trying to accomplish a similar goal.) Anthony intended his system for deaf people of all ages. For this reason, SEE 1 currently has about 6000 words and a complex set of syntactic principles and criteria.

Several attempts have subsequently been made to provide a simpler mapping of American Sign Language onto English. Linguistics of Visual English, Manual English, Signing Exact English (commonly known as SEE 11), and Signed English are all derivatives of SEE 1 to some degree. SEE 11 was explicitly designed to be used with children, and Signed English was created for preschool deaf children. All of these systems derive over half of their signs from American Sign Language, supplementing ASL signs with additional signs for words, as well as with signs for derivational and inflectional suffixes. Some of these suffixes take the form of unitary signs, while others are composed of finger spellings (e.g., indicating the past tense with the finger-spelled letters “E” and “D”). In fact, finger spelling is used, to some degree, in all of these systems except Siglish. The Paget-Gorman system uses finger spelling for proper names; in the SEE systems, it is generally (though not entirely) avoided. Manual English makes somewhat greater use of finger spelling, for example as an alternative way of forming the past tense. In Signed English, the amount of its use varies with the user’s fluency in English and with the audience being addressed.⁵

The final type of signing system involves the exclusive use of finger spelling (paralleling the associated spoken language). In the United States this method is known as the Rochester Method, after the area in which it was first widely used. (Ironically, the Rochester Method is no longer common in Rochester.)

THE USER’S PERSPECTIVE

The discussion of how to classify spoken and written languages suggested that structural classifications by themselves often reveal little about the extent to which the formal properties of a language are actually used or why these formal properties might exist in the first place. All of the classification schemes we have looked at so far have begun by making distinctions between the languages or language symbols themselves. Another possibility would be to make initial distinctions in terms of the people who are using the language, as Rolf Kuschel (1974) does in his study of the language of a single sign language user on Rennell Island in Polynesia. Kuschel distinguishes between signs that can be understood by a member of the same cultural group as the signer, signs intelligible to someone from a related group, and signs that would be clear to anyone. Our next task, of course, is to see the possible correlations between a social and a linguistic classification. If there is an empirical basis for a social approach to sign language analysis, we would expect a high degree of predictability of structural factors from social factors, and vice versa. My earlier analysis of writing and especially of speech provided just such evidence in their respective domains.

As in the case of writing, there have been practically no attempts to compare different types of signing systems with respect to differences among their users. Therefore, these attempts must be seen as only a first step.

PROLEGOMENA TO A FUNCTIONAL TYPOLOGY OF SIGNING SYSTEMS

TYPES OF SIGNING COMMUNITIES

Stokoe distinguishes five kinds of sign systems on the basis of whether or not they are acquired as native languages, whether or not they are associated with specific spoken languages, and if so, what kinds of scripts are used in writing those spoken languages (figure 7.7). This scheme can be simplified—distinguishing only between systems that are used as the primary mode of communication in a deaf community (which normally implies that they are acquired as native languages) and systems used for special purposes by hearing populations. This second subgroup needs to be further divided into sign languages used by populations who share a single spoken language, and those who do not (figure 7.8). An example of a sign language used regularly by a deaf community is American Sign Language (ASL). American Plains Indian Sign Language (PSL) typifies a sign system used by a hearing population that, at least originally, seems not to have shared a common spoken language. And Cistercian Sign Language (CSL) exemplifies a sign system used by a hearing community that shares a common spoken language (although this language may vary, depending on the location of the monastery).

Fig. 7.7 Stokoe Classification of Sign Languages (from Stokoe, 1974:358)

American Sign Language (ASL)

ASL is the primary means of communication used by an estimated one-half of the deaf population in the United States. With nearly one million users, it is the fourth most widely used linguistic system in the United States, behind English, Spanish, and Italian (Mayberry, 1978). ASL is commonly used among deaf children and adults and by deaf adults to their children, whether deaf or hearing. Until recently, however, the deaf have commonly hesitated to use it with hearing people (Kannapell, 1977). In addition to its strictly linguistic function, ASL is a mark of social identification, somewhat analogous to the use of diphthongs on Martha’s Vineyard to identify with the island (see chapter 5). When interacting with hearing people who know some signs, the deaf tend to switch to a manual system whose structure approximates that of spoken English; this may be one of the reasons that (hearing) linguists have had so much difficulty in obtaining consistent data on ASL.

Fig. 7.8 Simplified Classification of Sign Language Types

What about the other half of the American deaf population who don’t use ASL? Moreover, what about deaf children born to hearing families who do not know sign? Households with a deaf child often work out so-called home signs to permit some rudimentary communication between the child and other family members. Like nicknames or idiosyncratic names that hearing families often develop among themselves (e.g., calling the television set the idiot box or the boob tube), these signs may be recognizable to outsiders because of the transparent relationship between the sign and its referent. However, because multiple iconic representations of the same referent are possible (recall the three signs for “tree” in figure 7.1), there is considerable variation among home signs that share the same referent. For this reason, home signs have little value for the child entering a classroom.

Some hearing parents of deaf children receive formal training in signing. Note that we have said “formal training in signing”not “instruction in ASL.” Typically, parents learn signs for objects and activities that are especially important in their children’s lives.⁶ Depending on the child’s age, signs may be selected on the basis of the child’s needs for labels relating to home life, the parents’ need to know what signs the child is learning at school, or both. It is generally assumed that these signs are strung together in the order of English syntax. Usually, if these children are to sign fluently in ASL, they must learn to do so from their peers at school.

Does attending a school for the deaf assure the hearing-impaired child a model for learning ASL? Hardly. To begin with, the use of ASL in classrooms for the hearing-impaired is far from being universally accepted. The so-called oral method, which aims at teaching the deaf to speak, read, and write English without the use of signing, has dominated deaf education for over three-quarters of a century.⁷ As early as 1904, participants in the World Congress of the Deaf proclaimed that

The call has not been heeded; many educators of the deaf (especially those teaching in oral programs) know few if any signs. Others know and use individual signs with English syntax. This does not, however, mean that they know ASL.⁸ Learning 500 words of Russian and then stringing them together with English syntax is certainly not tantamount to knowing Russian.

Until recently, regardless of how much signing educators of the deaf knew, little signing was used by teachers in the classroom. In 1955, 78.6% of the schools for the deaf in the United States used only the oral method for teaching. Another 5.1% of schools used nonoral techniques, while 14.3% combined oral and nonoral techniques (Lunde, 1956; quoted in Stokoe, 1978:22). This, despite the fact that in 1943 an estimated 78.2% of the deaf used sign language and only 1% used speech alone (ibid.). Some schools have strictly forbidden the use of manual communication to the point of slapping hands on playgrounds or making children sit on their hands in class. In recent years, these statistics (and practices) have begun to alter, particularly with the growing acceptance of total communication (simultaneous speech and signing in English syntax) in schools for the hearing-impaired across the country.

What does this suggest about the fluency in ASL of hearing-impaired children in schools for the deaf in the United States? Many deaf children do not learn ASL at all (although, even in oral schools, indigenous systems of signs develop). If a child attends a school in which signing is forbidden and then returns to a household which does not sign, there are no models available from which to learn the language. Only when the child matures and regularly interacts with other deaf people who know ASL does he begin to learn the language. This may happen when a deaf graduate of an oralist school attends an institution such as Gallaudet College or the Rochester Institute for the Deaf, at which signing is normally used. It may also happen when the individual begins to participate in local deaf clubs. The question we must ask here is, how fluent can an adult become in a new language? It is typically assumed (e.g., Lenneberg, 1967) that after puberty, normal (hearing) people cannot acquire native competence in a new language. Either we must assume that native competence can be acquired by adults in signed (as opposed to oral) language, or we must conclude that deaf signers of the sort we have characterized never attain native fluency in ASL.

A less drastic case is that of deaf children attending schools in which signing is permitted, although ASL is not the medium of instruction. Since this is the situation in the vast majority of schools for the deaf, it is an important case to consider. In these schools, the children do indeed sign among themselves. However, “signing” is not necessarily the same as using ASL. Students (and teachers) readily acknowledge that some children are “better signers” than others. (It was noted in chapter 4 that some hearing people are more fluent language users than others.) Equally important, if a signing “norm” develops in one school or geographic region, it is not necessarily the same norm as elsewhere. Lexical items and even conventions of syntax may differ from place to place.

Of course spoken languages display considerable geographic variation as well. The item that is called a bucket in one part of the country is called a pail in another. In the phonological domain, some speakers pronounce the words Mary, merry, and marry as homonyms, while others make a two-or three-way phonological distinction. Is the amount of variation in ASL significantly different from that in spoken languages? Is it considerably greater than that in spoken languages? The sociological factors dialectologists use in assessing—or predicting—amounts of variability in spoken language (e.g., geographical isolation of speakers) suggest that in contemporary American society, there would be greater variation in ASL than in English. No careful, comprehensive studies have been done, but it seems likely that this prediction will hold. The next issue, of course, will be to decide when a quantitative difference becomes a difference in quality. Is there enough variability among sign language users to deny the existence of a community of users sharing the same linguistic system? The problem is familiar: How many participants are required to make up a language community?

In discussing spoken languages, we have found no easy answer. A single speaker of Tunica (and not very proficient at that) served as the only informant for an entire grammar of the language. Moreover, there are still small tribes (such as the Tasadai in the Philippines) who regularly interact only with one another. We expect their languages to differ from those of nearby groups with whom they share a common genetic linguistic source. If, for spoken languages, we are willing to acknowledge that social circumstances may yield high degrees of variation among users of genetically common languages, we should not be surprised to find wide variability among users of ASL, whose social circumstances have prevented signers from forming a more homogeneous linguistic community.

Why have we dwelt so long on the point that ASL is not the native language of many deaf Americans? The purpose of this section is to present ASL as an example of a signing system that is used as a primary means of communication for a group of deaf people and to clear up, in advance, some of the misconceptions about this system. ASL is not the language of the deaf in America any more than English is the language of all hearing people in the United States.

American Plains Indian Sign Language (PSL)

A second type of signing community comprises signing populations whose hearing is normal but who do not share a common spoken language. Our exemplum is American Plains Indian Sign.

What was (and is) the sign language of the American Plains Indians? Where, why, and how did it develop? Who initially used it? What uses does it have now? With the exception of the last question, there is little agreement about the etiology and development of PSL. In fact, as we shall see, the structural characterizations found in the language (see “Comparison of Structural Properties”) are also fraught with contradictions. Part of the difficulty derives from the fact that the language flourished in the nineteenth century, before the development of sophisticated techniques of linguistic description in the early and middle parts of the twentieth century. This problem is further complicated by the fact that an adequate study of ephemeral visual representation necessitates the use of movie or video equipment, which has only become technically and economically feasible within the last few years. Hence, the literature on PSL generally resorts to verbal descriptions or sketches to convey the signs’ configurations.

Most of what we know about PSL comes from the writings of Garrick Mallery, an Army Colonel who worked for the United States Bureau of Ethnography during the latter years of the nineteenth century. He produced an exhaustive study of pictorial writing in native North America, Picture-Writing of the American Indians (1893, reprinted 1972b). Turning his attention to ephemeral signing behavior of the same population, he also wrote extensively of Amerindian sign (1880a; 1880b; 1881). Since Mallery’s work, a scattering of ethnologists and linguists, along with some untrained observers, have written on PSL. A substantial portion of these writings have been reprinted in Umiker-Sebeok and Sebeok (1978). Of the group, Allan Ross Taylor’s account (1975) is particularly useful, in that it draws together and compares the positions of earlier writers. Therefore, this discussion will rely heavily on Taylor. Equally important, though, is the work of La Mont West (1960), the only formally trained linguist to have conducted extensive field work on PSL. While PSL was not the only signing system used by American Indians, it was the most sophisticated (Taylor, 1975:329) and therefore the most interesting for comparison with our other paradigmatic types.

All that we know for certain about the origins of PSL is that it was commonly used within and between tribes on the American Plains by the eighteenth century (see figure 7.9). Exactly why, where, and how it developed are still issues of debate. Logically, there are at least four ways in which PSL might have developed:⁹

intertribal communication

contextual problems in using speech

origins in sign language used by deaf Indians

intratribal ritual

Intertribal communication. Our very characterization of PSL as a paradigmatic example of a sign language used by hearing people who do not share the same spoken language logically implies that the language was used for intertribal communication when a common spoken language was not shared. There is the further implication that the signing system arose because of this need for a lingua franca.

Fig. 7.9 Distribution of American Plains Indian Sign Language (from Taylor, 1975:368)

Most writers agree that intertribal communication was a central—perhaps the central—use of PSL, The relatively high degree of mobility of the Plains population may help in explaining why a visual lingua franca developed on the Plains, but not elsewhere (West, i960, vol. 2:62). Moreover, trading relationships between tribes may have been a significant factor in the development and spread of the language. Taylor writes that “trade was certainly a principal agent for the diffusion of the sign language throughout the Plains during the nineteenth century” (1975:332). Moreover, he hypothesizes that if the signing system originated in the Gulf Coast of Western Louisiana, Texas, and Northern Tamaulipas (Mexico), as some have suggested, “then trade was probably the catalyst for the spread of the gestures to the Southern Plains tribes” (ibid).

The suggestion that PSL provided a lingua franca for trade between group not sharing a common spoken language invites comparison between PSL and spoken trade jargons, such as Chinook Jargon, which were discussed in chapter 5. Besides the sociocultural context of trade between equals, there are certainly a number of other similarities. In both instances, it was largely men (not women) who learned and used the communicative system (ibid., pp.333-334). Use of the lingua franca was especially common before the adoption of English by the tribes (West, 1960, vol. 2:10). Structurally, both forms of representation display lack of stability across tribes, and both use redundancy extensively (see “Comparison of Structural Properties”). However, there are also differences between the systems—not to mention puzzles —that need explaining.

We know that the use of PSL did not preclude knowledge of spoken languages other than one’s native tongue. West (i960, vol. 2:62) dismisses the hypothesis that sign grew up because of “indisposition or inability of the plains Indian to learn a spoken language other than his own” (Dodge, 1882:46; Humfreville, 1897:153) on the grounds that 52% of his PSL informants were fluent in English and 17% knew two or more other Indian languages. One might take issue with West, arguing that the conditions that gave rise to PSL (the need for a lingua franca) cannot explain its use today. As noted before, we still put buttons on the sleeves of men’s coats, even though we are no longer trying to prevent the use of coat sleeves as handkerchiefs. The question is, how far back in time does one see the use of sign language among Indians who are bilingual? Clark (1885:218,340) attempts to correlate the presence of sign lingua francas with the absence of spoken lingua francas. West, however, argues that

What we would like at this point—but lack—is information on the uses of signed and spoken lingua francas one hundred or even two hundred years ago.

A still more important divergence between (spoken) trade jargons and PSL can be seen in the conditions under which they were used. While trade jargons seem to have been used exclusively for intertribal communication, PSL had a number of intratribal uses as well. In fact, many have argued that as early as the nineteenth century heavy use of PSL was made within individual tribes who shared a common spoken language. Umiker-Sebeok and Sebeok even state that

In the twentieth century PSL is used heavily among speakers of the same language. Half of the 111 informants West surveyed in the late 1950s reported they had learned PSL around home (the other half mentioned extratribal contacts). Moreover, today this sign language is commonly used in story telling, a situation in which the signer and the audience are likely to share a common spoken language.

Contextual problems in using speech. A second possible source for PSL is contexts in which the use of speech is impractical. In hunting and warfare it is often expedient to be able to communicate with one’s compatriots without alerting one’s prey (see Webb, 1931; reprinted in Umiker-Sebeok and Sebeok, 1978, vol. 2:96). Webb further suggested that the vast, flat open spaces of the American plains themselves facilitated the development of PSL: “On the Plains the eye far outruns the ear in its range.” This thesis is highly reasonable but certainly not conclusive. Many hunters and warriors use no sign, and inhabitants of hilly regions often do use sign systems. One of Webb’s arguments for his geographical explanation is that of how Indian signs are formed:

Taylor counters these assertions, noting that many signs are “not wide and body-clearing” (1975:331). What seems important in determining the physical scope of the signs is the users degree of formality and fluency in sign:

This reduction in physical scope with increased fluency is precisely what our theory of language use would predict.

Origins in sign language used by deaf Indians. It is always possible, of course, that PSL developed from a rudimentary signing system created by (or for) deaf Indians. Clark (1885:144) felt it likely that hearing-impaired Indians had always used this sign language, and West (i960, vol. 2:64) confirms that PSL is still used for communicating with the deaf. While authors such as West suggest that PSL is a language system that can be used in communicating with the deaf, it is also plausible that deaf Indians themselves (or Indians attempting to communicate with the deaf) may have initiated or at least contributed to the formation of PSL. A parallel process is now going on in Rhode Island, where a group of scuba divers are learning ASL at the Rhode Island School for the Deaf in order to communicate under water.

Intratribal ritual. PSL could, in principle, have arisen in rituals that require silence. These are not activities such as hunting or warfare in which a sign language might be pragmatically expedient, but rather rites in which the tribe’s conventions call for silence. There seems to be no evidence for ritual as the origin of PSL, although such origins are documented among other aboriginal tribes (see Umiker-Sebeok and Sebeok, 1978, vol. 2). More to the point, ritual is clearly the source of the third genre of signing systems we will consider.

Cistercian Sign Language (CSL)

A third category of signing systems contains those used by populations who not only speak and hear but also share a common language. There are many such contexts in which a sign language might, in principle, develop—hunting, scuba diving, working in a noisy factory, sitting in a lecture when someone else is talking. This section is concerned with those sign languages that develop when the use of speech is forbidden for ritual purposes, especially the use of sign in monasteries, and Cistercian monasteries in particular.

To what degree is Cistercian sign representative of ritual sign in general or of monastic sign in particular? Obviously, the extent to which the ritual either necessitates or allows communication between participants will strongly influence the shape of the resulting sign system. If, for example, participants are allowed to communicate only the equivalents of “hello,” “come here,” and “goodbye,” practically any sort of signs would do. On the other hand, if messages are allowed to become complex, I predict rather specific characteristics for the resulting system.

We shall therefore study Cistercian Sign Language (CSL) for two reasons: it has developed to a point where it can encode rather sophisticated messages and, of equal importance, it is well documented. Robert Barakat (1969, 1975) has produced a history, grammar, and lexicon from which the discussion of CSL is entirely drawn.

Since the early centuries of the Christian era, a number of monastic orders have imposed silence upon their members. In the Rules for Monasteries of St. Benedict, we learn that

Visual signs were, however, permitted when communication during hours of silence was necessary. According to Barakat (1975:25), it was with the founding of the monastery in Cluny (in the year 909) that a fixed system of signs first emerged. As the Cluniacs exerted more influence, their signs were adopted in other monasteries in Europe. Each monastery developed its own list of “approved” signs, which differed widely in number and formation from the signs used in other monasteries.

Monastic sign provides an especially clear example of how functional pressure to communicate creates linguistic change. Barakat relates that

The same phenomenon occurs today at the St. Joseph’s Abbey in Spencer, Massachusetts, where Barakat did his field work. New signs are created that go beyond the authorized list (Barakat cites, among others, signs for “all shook up,” “sniffles,” “electrician,” and “pumper-nickle”). He also explains that some of the authorized signs, such as those for “bookbinder” or “shoemaker” are “no longer in common use because less expensive secular sources provide the products these craftsmen would ordinarily make” (ibid., p.29). Similarly, only functional lexical distinctions are made:

A final point about function concerns the use of CSL. In recent times, the Order has recognized the need to strike a balance between traditional practices and more modern ways, which include the use of some verbal communication. Barakat observes that

Even when signs are used, CSL seems to have limited usefulness:

COMPARISON OF STRUCTURAL PROPERTIES

Comparisons of signing systems are not new. West (1960, vol. 2) makes an extensive, albeit rambling comparison of different sorts of sign languages. Stokoe (1974) offers a number of comparative comments, while Crystal and Craig even compare a number of signing systems with respect to a revised version of Hockett’s design features for human language (1978:156-159). Lacking is a systematic comparison of sign language types with respect to a consistent set of syntactic and semantic properties of the sort linguists use for comparing and analyzing spoken language.

The following discussion is a first attempt at such a systematic comparison. It is organized to parallel, as much as possible, the structural characterization of trade jargons and pidgins presented in chapter 5. The categories were selected not only for their appropriateness in discussing signing systems but also because they provide a natural bridge for comparing signing systems on the one hand and spoken systems on the other.

Let us consider each of the following structural points in turn:

Stability.

A sine qua non for any scheme of representation is a body of elements and combination rules whose use is common to a community of language users. Even if users disagree on the meaning of a particular element or combination rule, there must still be a reasonable consensus as to which phenomena belong to the language.

Of the signing systems considered above, ASL most closely approximates the user stability of spoken and written languages. Such stability should increase as the deaf community grows in its level of mobility and, to a lesser degree, education. PSL seems to vary widely from tribe to tribe. Critchley distinguished between two major dialect groups:

West, who did not investigate sign language in the southern plains, finds a major split between what he recognizes as “Plains Standard” and local dialects. While the Plains Standard Signing is somewhat restrained, the local dialects—especially in the far northern area—are more free-wheeling (i960, vol. 2:75).

West has also observed in the use of PSL what Charles Ferguson (1959) has called a diglossie situation, in which a user displays different language behaviors with different audiences:¹⁰

Ferguson (1959:330) describes precisely the same phenomenon in the use of local varieties of Arabic as opposed to Classical Arabic or of Haitian Creole as opposed to French: speakers who use local forms of Arabic or Haitian Creole deny even knowing these local languages, which lack social prestige. Moreover, in PSL, the Plains Standard itself does not seem to be known and used by a significant proportion of the population. There is

West indicates that currently “local dialects can diverge to the point of mutual unintelligibility and lexical sharing in the range of 20 to 30 percent within the space of one or two generations” (ibid., p. 19).

Comparing West’s observations about PSL with the previous analysis of (spoken) trade jargons, both appear to display considerable variation from one group of users to another. Lack of regular contact between users from different tribes probably helps undermine the establishment of a standard that is known and understood across tribes.

The case of CSL is different from that of ASL or PSL. Within a single monastery, CSL would appear to be highly stable across monks because monks form a close-knit community (Barakat, 1975:46). Divergences between varieties of CSL appear when one compares other monasteries in the United States or in foreign countries (ibid., p.45). In America, we find divergences between the signs needed to handle local conditions. Abroad, the spoken language of the signing community is also a factor. Because signs in CSL are often intimately tied to the spoken language (see below), such speech-based signs lose their significance when transported to a monastery located in a country in which a different spoken language is used.

Comparing the stability of ASL, PSL, and CSL, similar social factors appear to work in each case. The broader the social base, the greater the homogeneity across users. However, ASL is paradigmatically the most stable of the three because, for the majority of its users, there is no alternative linguistic modality.

Sources of vocabulary and syntax.

ASL: largely indigenous

PSL: unclear

CSL: heavily dependent upon English

The previous discussion of ASL identified a spectrum of signing systems used by the American deaf population that varied with respect to the amount of influence English had on their signs and syntax. At the end of the spectrum which shows the least influence from English (in signs and syntax) is ASL. Having its own lexicon, ASL directly represents experience rather than a spoken language. As Robbin Batti-son (1978) has shown, ASL does borrow items from English, such as the finger spellings for English words whose referents are not already named by indigenous ASL signs. However, these signs tend to become “naturalized,” losing their distinct finger spelling elements and becoming incorporated into a single sign with an arbitrary link between signifier and signified.

The question of sources for ASL syntax has generated a great deal of debate in the linguistic literature. Of particularly current interest is how much influence French Sign Language has had upon ASL and how much of ASL is indigenous to American signs. Whatever their signing source, though, it is now generally accepted that ASL has its own rules of syntax, which, a priori, have no greater similarity to English syntax than to the syntax of any other spoken language (see Wilbur, 1976; Friedman, 1977; Siple, 1978).

PSL is more difficult to describe because of disagreement over the extent to which PSL is an independent language or a representation of speech. John Harrington takes the extreme position that PSL is merely a visual representation of speech:

Harrington illustrates his case with an example of word-order parallels between sign and speech:

Similarly, Kroeber (1958) argues that PSL is a speech surrogate.

West assumes the opposite stance, asserting that there is “no evidence that Amerindian sign language is in any sense derivative of spoken language.” He does admit, though, that

Extending the argument for the autonomy of sign, Taylor points out that the possibility of lexical borrowing must be considered not only from spoken Indian languages to sign, but in the other direction as well:

Our current state of understanding of the relationship between PSL and spoken language is summarized by Umiker-Sebeok and Sebeok. Speaking for all aboriginal sign languages, of which PSL is but one example, they write:

However, the fact that PSL is more closely tied to spoken language than are sign languages of the deaf is the result of social circumstances that have held the autonomous development of PSL in check:

The fact that social circumstances limiting the use of PSL have helped to maintain its ties with spoken language is reminiscent of our earlier observation that trade jargons do not develop an independent, stable syntax because use of the jargons is socially restricted.

The source of signs and syntax in CSL is much clearer. The large majority—although not all—of the CSL signing system derives from English.

Lexically, there are three groups of signs that do not depend on speech but represent either ideas or experience directly. These include pantomimic signs, such as that for “sleep,” for which the “palm of hand or hands are placed on one side of the head and then the head is tilted to one side as though resting on a pillow” (Barakat, 1975:36—see figure 7.10). Pure signs show arbitrary relationships between signifier and signified. Thus, the sign for “abbot” is made by “placing the tips of the right forefinger and middle finger on the forehead so the tips of the fingers are pointing up” (ibid., p.39—see Figure 7.11); “green” is signified by using the tip of the right forefinger to draw a line from the right ear to the tip of the nose. Barakat’s qualitative signs are more metaphoric in character, since they define their referents “in terms of what qualities are associated with them” (ibid., p.40). Thus “Irish” is indicated by either the sign for “green” or the sign for “potato eater” (see figure 7.12), and the city of Dallas is named by the sequence of signs for “secular” + “courtyard” + “president” + “K” + “shoot.”

Fig. 7.10 The (Pantomimic) Sign for “Sleep” in Cistercian Sign Language. Reprinted by permission of Cistercian Sign Language (Cistercian Studies, No. 11, p.121) by Robert Barakat. Kalamazoo, Michigan: Cistercian Publications, 1975-

Fig. 7.11 The (Pure) Sign for “Abbot” in Cistercian Sign Language (ibid., p.93)

In CSL, there are, however, many signs that clearly depend on speech. Thus, “baloney” is denoted by the signs for “bull” + “O” + “knee,” “cookie” being signed as “cook” + “key” (see figure 7.13), and the city of Cincinnati being indicated by a reduplication of the sign for “sin” (hitting the chest with the right hand) plus the signs for “A” and “T.”

Like the lexicon, the syntax of CSL is heavily, but not exclusively, derived from English. Where deviations exist, they often result from syntactic circumlocutions that arise from forming derived nouns of the sort we have already seen. Moreover, since some elements of English, such as the verb be or the expletive there, are not represented in CSL, some syntactic rearrangements are used to convey the desired message. Finally, the lack of lexical distinctions comparable to the vocabulary of English may make for syntactic divergences between CSL and English. Here are some examples of the syntax of CSL:

Fig. 7.12 The (Qualitative) Sign for “Irish” (= “Potato” + “Eat”) in Cistercian Sign Language (ibid., pp.103, 115)

Fig. 7.13 The (Derived) Sign for “Cookie” (= “Cook” + “Key”) in Cistercian Sign Language (ibid., pp.109, 144)

In sum, the extent to which a signing system draws on spoken language for its vocabulary and syntax depends heavily on the availability of a spoken language and the extent to which the community of users makes productive use of the system. These two factors are also accurate predictors of the other structural characteristics of signing systems considered below.

Character of syntax.

Word order

ASL: may not be as important as in spoken language

PSL: may not be as important as in spoken language

CSL: generally approximates English word order

The existence of grammatically describable word-order constraints has long been considered a necessary property of human language (Saussure, 1959). While recognizing some flexibility in ordering, especially in highly inflected languages, linguists have nevertheless assumed an underlying or unmarked word order in every language (see Greenberg, 1963).¹¹ In fact, researchers studying chimpanzees who have been taught to sign have considered word order a sine qua non for language. Roger Brown (1970) initially denied that Washoe was using language because of her lack of consistent signing order, while Herbert Terrace (Terrace et al, 1976) has gone to great pains to prove that Nim Chimsky actually does have some notion of sequencing constraints.

It is still not clear whether ASL has such an underlying word order and if so, what that order is. This is also true of PSL. There do appear to be some ordering conventions, such as the placement of negative markers and question markers (West, 1960, vol. 1:90). However, according to West, word order is less important in sign language than in speech. In sign,

What grammar exists is due “almost entirely to internal sign morphology” (ibid.).

It appears that in visual languages the importance of word-order constraints is directly proportional to the dependence of the visual system on a spoken model. As mentioned above, CSL, which draws heavily upon English, does follow comparatively strict ordering constraints.

Circumlocution

ASL: low

PSL: varies

CSL: reasonably high

The amount of circumlocution in sign is a function of the degree to which members of the signing system are in regular contact with one another and also depends on external constraints upon the size of the vocabulary. We would expect stable circumlocutions to be relatively rare in ASL because it is the primary means of communication for a sizable population. In PSL there appears to be a greater amount of circumlocution, roughly comparable to that found in spoken trade jargons and pidgins. Taylor reports that Indians use circumlocutions when they are not sure of a sign. Thus, although PSL has a sign for “wolf,”

Other documented examples are use of the signs “white” + “soldiers” + “walk” to indicate “infantry,” or “female” + “chief” + “big” to indicate “queen.” Finally, Webb has commented that “the Indians repeated signs more often than do the deaf’ (1931, reprinted in Umiker-Sebeok and Sebeok, 1978, vol. 2:97).

CSL also uses a great deal of circumlocution. We have already seen that the city of Dallas is denoted by “secular” 4- “president” + “K” + “shoot.” Other examples are:

raw = “fruit” + “green”

forest = “all” + “wood” + “courtyard”

freeze = “arrange” + “hard” + “water”

The very presence of such circumlocutions would seem to follow from the fact that the “Cistercian sign language was never intended to expand communication among the brothers but to restrict it to some extent.” For this reason, “the administration of the Order has rarely seen fit to increase the sign inventory” (Barakat, 1975:44). Thus, the circumlocution among two radically different groups, the Plains Indians and the Cistercian monks, seems to have arisen for quite different reasons. The Indians’ lack of frequent contact made circumlocution necessary to ensure that one was understood. Among the monks, circumlocution was motivated by artificial constraints on the size of the lexicon.

Redundancy

In addition to the use of circumlocution as a form of redundancy, let us consider the extent to which redundancy occurs in manual systems through the concomitant use of facial expression or speech. Discussions of ASL have often noted that the amount of facial expression used directly correlates with the degree to which “native” ASL is used. Wilbur (1976) observes that use of the face is more extensive when using ASL than when using some other form of manual or signed English. In the same vein, Stokoe finds that

Recalling our discussion of word order in ASL, Stokoe’s comment implies that in ASL, facial expressions helps to mark grammatical relations that, in spoken languages, are typically marked by ordering constraints.

Many writers have observed that users of PSL use far less facial gesture to accompany their signs than do deaf signers (see Stokoe, 1972: 108; Taylor, 1975:338; Critchley, 1975:72). On first consideration, the observation seems curious, given Horatio Hale’s observation that, when Indians switch from their own languages to Chinook Jargon, they gesticulate actively. Several hypotheses might explain this discrepancy. First (and least interesting theoretically), we may be dealing with two different populations of Indians, one of which gesticulates freely and the other of which does not. Second, the degree of accompanying gesticulation may be a function of ones fluency in signing. Recall West’s observation that “sign talkers unfamiliar with the [Plains] standard tend towards larger, freer movements, exuberant use of facial expression and posture” (1960:vol. 2:75). Finally, there is West’s hypothesis (ibid., p.77) that the Indians whose signs he observed use little facial gesture because they believe that whites expect them to appear stoical—an interesting case of life imitating art.

What about CSL? While Barakat does not raise the point, we would hardly expect active gesticulation in a religious order that has developed signing as a way of restricting communication.

A second type of redundancy that can occur in signing systems is the use of speech to accompany signs. Such redundancy is used in “simultaneous” or “total” communication, which strings together mostly ASL signs in English word order, accompanying each signed item with its verbal equivalent. ASL proper, however, is entirely silent. So is CSL, for obvious reasons.

PSL, on the other hand, does make frequent use of speech to accompany sign. Use of both modalities is especially common in storytelling. The use of speech alongside signing does not necessarily imply, however, that both modalities convey identical messages. While this is sometimes true, West observed cases in which the speech “involved comments about what was being simultaneously signed,” and other cases in which the speech constituted “an independent lecture or conversation intended for the writer [presumably West] or some by stander” (1960, vol. 2:76).

Sign formation

Iconicity

ASL: decreases with time

PSL: currently relatively high

CSL: well-represented

How much do the hand configurations in a signing system resemble the items of experience which they represent? What factors are important in determining the degree of iconicity? In the last part of this chapter, we shall consider this question in some detail, especially with respect to ASL. However, because the issue bears upon our comparison of ASL, PSL, and CSL, I introduce the subject here.

It is generally acknowledged that, while ASL has a number of signs that might be judged iconic, it also has a sizable number of arbitrary pairings between signifier and signified. Moreover, when new iconic signs are introduced, they tend to lose their iconicity over time. In his Dictionary of American Sign Language, Stokoe (1965) lists about 25 percent of all entries as currently iconic. (Wescott, 1971:418, hypothesizes that about two-thirds of the remaining 75 percent were originally iconic.)

Signs in PSL are usually considered more highly iconic than those in ASL. The interesting issue here is whether this higher level of iconicity is a necessary property of a signing system used by hearing people or whether it is an accident of history. There seems to be some evidence for the latter position. Dodge (1882, reprinted in Umiker-Sebeok and Sebeok, 1978, vol. 2:4-5) suggests that sign languages of hearing populations historically become increasingly arbitrary. Ljung goes one step farther, attributing the high degree of iconicity found in PSL to its “relatively recent origin” compared to naturally spoken languages. Had PSL “been allowed to develop freely, it would no doubt have become more arbitrary as time went by” (1965, ibid., p. 221). However, because PSL became increasingly a language of ritual rather than one of open-ended communication, such increased arbitrariness in signs did not have a chance to emerge.

There is little to say about CSL on the issue of iconicity; although it has iconic signs (e.g., “sleep” in figure 7.10), it is not clear what proportion of the signing vocabulary they constitute. However, given the heavy dependence of CSL on speech for forming new signs, iconicity is unlikely to be nearly as important in CSL as in ASL or PSL.

Signing space

One of the criteria students of sign languages use to distinguish between pantomime and language is the restriction on the physical area in which the movements are performed. The more restricted the signing space, the more “language like” the sign system is assumed to be. The same economy of effort exercised with respect to signing space is responsible, in spoken language, for motivating such abbreviations as TV and SPCA. However, the restrictions on signing space have a theoretically important aspect as well. The very fact that some signs are spatially “grammatical” which others are not further indicates that the system in question is a language governed by rules.

In normal conversation (e.g., when not being poetic or dramatic), the ASL signer generally restricts sign formation to an area between the head and the waist (vertically), and between the shoulders (horizontally). The signing space for both PSL and CSL is less restricted. PSL also utilizes the head, chest, and arm areas but also includes the legs and occasionally the back (Taylor, 1975:38). Signs in PSL are sometimes described as large and expansive, although size may be a function of one’s degree of fluency (cf. West’s comment that those less familiar with the Plains standard use larger movements). Similarly, in CSL we find signs made by touching the heel (e.g., the sign for “socks”) or back (the sign for “back”). Such spatial diversity is possible, at least in part, because the number of signs to be learned is limited in both instances.

Semantics

Range

ASL: potentially unlimited

PSL: unclear

CSL: limited by design

Are there restrictions on the range of topics signing systems can be used for? One answer may lie in the size of the lexicon. Another concerns the exclusion of whole ranges of topics from discussion.

A problem with comparing vocabulary size is the lack of information. While Clark (1885) estimated there were 1,100 signs in PSL, West (1960) puts the number at closer to 3,500 (see Taylor, 1975:343). Furthermore, we do not always know whether root words or compounds are being counted. (Recall here the cases of German and Neo-Melanesian, in which a large proportion of the vocabularies are formed by combining other elements.) Moreover, since many signs may be unstable through time, especially in PSL and CSL (see Barakat, 1975: 46), a tally of only the basic, stable signs may sharply underestimate the signing options actually available at a given time.

The alternative approach is to ask whether there are any limitations on the topics that can be discussed in the signing system. In ASL the answer, in principle, is no; the issue is less clear for PSL. Some linguists have questioned the semantic flexibility of sign (e.g., Voeglin—see Stokoe, 1974:355). West even writes that some topics were excluded from the sign language texts he collected (i960, vol. 2:76), although the absence of profanity and obscenity and the low frequency of emotive and evaluational terms may be a function of the elicitation situation. On the other hand, Umiker-Sebeok and Sebeok (i978:xx) conclude that the range of topics that could be discussed was “seemingly limitless.” It appears likely that such discrepancies result, at least in part, from differing data sources.

In the case of CSL, we would expect, as a matter of principle, that the semantic range of signing would be restricted. It is unlikely that a system that was devised to restrict communication would be as semantically rich as the spoken language it replaces.

Dependence on context

ASL: low

PSL: unclear

CSL: higher

One of the claims that Crystal and Craig (1978) had leveled against signing systems was their lack of displacement (their inability to talk about items of experience not physically present). In the previous discussion of formal criteria for sign languages, I argued that Crystal and Craig failed to make their point for ASL. However, this is not to say that sign languages of deaf communities do not use context when contextual information is available.

One common way of creating proper names in ASL is to denote the person through a characteristic property of that person (using the sign for “mustache” as the name for an individual who wears a mustache). In this sense, the origins of such proper names may be seen as contextually bound. However, the sign becomes acontextual (and the language capable of displacement) if I can make the name sign for my mustachioed friend when he is absent or if I am able to keep using the name sign even after my friend shaves.

The facts about PSL are unclear. We might hypothesize that because PSL is commonly used in storytelling—indirect reference par excellence—it would not be particularly context bound. However, we need more data before drawing any conclusions. In the case of CSL, we do have evidence indicating contextual dependence. Barakat admits that one of the major defects in the system is “the need for context; that is, many messages could not be understood if the persons whom the message is about were not present” (1969:116).

CONCLUSIONS

We have looked in some detail at three signing systems choosen as representatives of three distinct social contexts. As we have seen, none of the examples, perhaps with the exception of CSL, precisely exemplified its type. We have also seen that the structural properties of each of the three signing systems did not neatly cluster so as to yield categorical statements about the linguistic character of each sociolinguistic type. Nonetheless, the following conclusions can be drawn from this comparison:

ICONICITY AND THE LEARNING OF SIGN LANGUAGE

The comparison of ASL, PSL, and CSL touched on iconicity, that is, resemblance between a sign and that which it signifies. Yet the problem of resemblance has repeatedly arisen in previous discussion of representation. For reasons that will become clear, iconicity is an especially important issue in sign language studies.

ICONICITY

The arbitrariness of the linguistic sign, a principle established by Saussure, has dominated linguists’ thinking about representation for most of the twentieth century. A word such as dog no more looks or sounds like the four legged creature to which it refers than does chien or Hund; so the idea of arbitrariness seemed reasonable enough.

In recent years, this Saussurian dictum has been questioned, although the challenge has not come on entirely the same grounds from which Saussure took his position. Saussur es concern was with spoken language, while the arguments in favor of iconicity as a significant linguistic parameter have assumed sign languages and written languages to be part of the broader domain of linguistic representation. One proponent of this position, Roger Wescott, has suggested that “as regards iconism . . . the only realistic question we can ask about a given form is not Is it iconic?’ but rather ‘How iconic is it?’” (1971:426).

The importance of “relative” iconism has emerged most fully in studies of ASL. Sign language research, only recently accepted within the fold of legitimate linguistic inquiry, has taken great care to establish that criteria used for judging and analyzing spoken language can be applied to sign data as well. This quest for legitimation seems, in part, to underline attempts to establish that sign languages are not overwhelmingly iconic after all, and therefore, by the principle of the arbitrariness of the linguistic sign, qualify as languages.

Yet it is clear that visual signs are different from signs in spoken languages. Admittedly, mature signers might not typically exploit or even be aware of the iconic roots of a sign any more than a speaker of German, in using the word Handschuh (“glove”), is conscious of the literal etymology—“hand” + “shoe.” Nonetheless, the average level of iconicity in sign languages is higher than in speech.

The place of iconicity in sign language has typically been considered in light of what we might dub the “progressive loss of iconicity” hypothesis, which suggests that the longer either a sign or a sign language user is linguistically “active,” the less important iconicity becomes. Nancy Frishberg’s work on historical change in ASL (e.g., 1975) has established this principle diachronically. Bellugi and Klima (1976) make a similar point about the changing shape of signs in the course of a conversation: if a novel iconic sign is introduced, the sign loses its iconicity with continued use in the conversation.

A third dimension to the “progession” hypothesis, as seen in Roger Brown’s work (1977), suggests that, ontogenetically, iconicity is initially an important aid in learning signs; as the signer becomes more fluent, iconicity becomes less valuable. As we shall see, this generalization is far too sweeping. The assumption that all beginners in sign language acquisition—children of deaf parents, adult learners acquiring sign as a second language, autistic children, retarded people, even chimpanzees—make use of iconicity is not verifiable. This is true because the types of learners that Brown groups together are sharply distinguishable along functional lines that he overlooks.

DIMENSIONS OF DISCOURSE

This book has shown that linguistic structures are significantly (though clearly not entirely) derivable from the social and communicative needs of the language users. The remainder of this chapter demonstrates the relevance of this thesis to the issue of iconicity in language.

Four variables are important in predicting the role of iconicity in visual language:

the character of the language learner

the character of the message sender and receiver

the character of the message itself

the character of the community in which the linguistic transaction takes place

Most of what I have to say about iconicity concerns characterization of the language learner. (My comments on the other three parameters are meant only to be suggestive.) Therefore, I shall reserve discussion of the language learner till last.

The Character of the Message Sender and Receiver

The degree of iconicity contained in an actual message may be affected by the relationship between the interlocutors. Is the interlocutor deaf or hearing? Is the interlocutor conversant either in some form of manual English or in ASL? We have a good deal of evidence that deaf people tend to use ASL with members of the deaf community but a form of manual English with hearing people who sign—whether or not the hearing interlocutor knows ASL. The amount of iconicity should be greater in using ASL than in using manual English, although we lack appropriate studies. The amount of iconicity should also increase (regardless of the base sign system) if the interlocutor is perceived as knowing little or no sign. We have anecdotal evidence here but lack carefully controlled studies.

The Character of the Message

It is with respect to the nature of the linguistic message that we probably know the most about the iconic properties of sign. The work of Klima and Bellugi (1979) has established not only that “poetry and wit” are quite possible in ASL but also that it is in the use of figurative language that the iconic properties of signs tend to be most fully explored. Again, however, we need careful studies comparing how the same signer communicates with the same interlocutor in different linguistic genres before we can hope to understand the effects of genre on iconicity.

The Character of the Community in which the Linguistic Transaction Takes Place

Our third variable is the linguistic community—especially the number of active signers. How many individuals are needed to constitute a linguistic community? Obviously, at least a single message producer and a single receiver. But is that enough? Even if it is, does the language that develops in such a restricted milieu differ in any significant way from language used by multiple producers and receivers?

We are fortunate to have data from which to make comparative studies of levels of iconicity in signing communities of differing sizes. Kuschel’s study of Kagobai, the only deaf person on Rennell, a small Polynesian island, uniquely documents the communication system that has evolved between a single deaf message sender and a sizable hearing population. Kuschel reports that many of Kagobai’s signs are highly iconic (at least with respect to his cultural milieu); he hypothesizes that this may be a function of Kagobai’s unique position and that iconicity might decrease as the number of signers increased (1974:37).

A strikingly similar finding has been reported by Heidi Feldman (1975) for the deaf children of hearing parents that she and Susan Goldin-Meadow (1975) studied in Philadelphia. While the parents were raising the children in oralist schools, the children independently created their own signing systems. Feldman predicted that the highly restrictive size of the knowledgeable signing community (the children themselves) would generate a highly iconic set of signs:

Feldman’s study bore out the hypothesis. In both Kuschel’s work (on the one hand) and Feldman and Goldin-Meadow’s (on the other), it would be interesting to see under what conditions such signing begins to lose its strong iconicity. Is it, in principle, necessary to increase the number of deaf signers in a community before iconicity diminishes, or is an increase in the number of comprehending message receivers sufficient?

THE LANGUAGE LEARNER

As stated before, one goal of studying linguistic representation is the solution of problems in communication. Perhaps the most serious problem we can envision is never learning a system of linguistic representation. It has recently become fashionable to teach elements of ASL to populations who either will not or cannot learn spoken English. As we shall see, one of the major reasons for the choice of ASL is its high degree of iconicity, which is believed to facilitate learning. Our beliefs about the correctness of this hypothesis will, understandably, influence our pedagogical techniques; therefore, it becomes a pragmatic imperative to examine the hypothesis in some detail.

In his paper “Why are Signed Languages Easier to Learn than Spoken Languages?” (1977), Roger Brown hypothesizes not only that the underlying premise in his title is true but also that the reason for the greater ease of learning sign, at least in its early stages, is its greater degree of iconicity.¹² Well aware of the strenuous efforts of sign language researchers to establish arbitrariness as a significant factor in sign language use, especially among fluent signers, Brown suggests that the significance of iconicity should nevertheless not be over-shadowed, a conclusion he reaches through the collection of several kinds of data, some of which he has gathered himself, and some of which he has drawn from other researchers. I shall be questioning this collection of sources, suggesting that an understanding of the role of iconicity in learning language requires a much more fine-grained analysis of what is being learned and by whom than Brown’s single generalization would lead us to suspect.

To summarize his paper briefly: Brown surveys the literature on autistic children, mentally retarded children, and chimpanzees being taught sign language (actually in most cases, lexical items from ASL). He finds the lexicon of all three groups to be (at least impressionistically) highly iconic, and hypothesizes that such iconicity might have been perceived by the learner and actually utilized in acquiring the sign. His own data derive from an entirely different population: four-and five-year-old normal children learning signs as a game. In his first experiment, six children were taught eight signs for concrete objects (“chair,” “ball”), the experimenter pairing the sign in ASL with the English gloss. Another six children were taught eight other ASL signs but given inappropriate glosses. Brown’s second experiment replicated this design, but used more abstract (and generally less iconic) signs. In both instances, the results clearly showed that when a sign was paired with its appropriate gloss (i.e., when the child had a reasonable chance of deriving an iconic interpretation of the sign), comprehension far exceeded that of inappropriate pairings (in which iconic hypotheses would be hard to make).

Brown relates his findings to Rosch’s work concerning the Basic Object Level (e.g., Rosch et al, 1976). In normal first language acquisition, according to Rosch, the earliest words learned are usually on the Basic Object Level (the level of generality at which language users most naturally form class terms—chair as opposed to the superordinate furniture or subordinate kitchen chair). Generally speaking, there are more characteristic attributes and movements for Basic Object Level terms than for superordinate or subordinate terms. Moreover, attributes and movements are the principle sources of iconicity in ASL. Therefore, Brown concludes, “there is a greater potential at the Basic Object Level for the creation of iconic signs than at any other level” (1977:21). Since the words in the initial experiment with normal children were largely at the Basic Object Level, he hypothesizes that iconicity is useful in the early stages of sign language acquisition.

This hypothesis is interesting, particularly because it is stated in sufficiently strong form to be dissected and tested. That dissection will be undertaken here by stepping back from the issue of iconicity and considering the particular kinds of language learning we are talking about. Only then shall we be in a position to understand the role of iconicity in different types of language and to determine if it is the same in each instance, as Brown’s hypothesis would seem to suggest.

Who are we talking about when we speak of sign language learners? Is there any reason to believe that the spectrum of individuals or classes of individuals involved will all learn language—of any sort—in the same way? Research on the acquisition of first and second spoken languages has made us aware that the process by which a person learns additional linguistic systems is not necessarily a recapitulation of native (first) language acquisition (Dulay and Burt, 1974; Politzer, 1974). Similarly, experience with language-deficient children makes it clear that language therapy programs cannot merely attempt to make children pass through “the same” stages of language acquisition as normal children (assuming such an idealized sequence of stages exists) (e.g., Menyuk, 1964; Leonard, 1972).

Therefore, in considering whether iconicity is an aid (at least potentially) to early lexicon building in sign, the first question we need to ask is, Lexicon building by whom? There are at least four different groups of learners to be examined separately:

children learning sign as a first language (normally from deaf signing parents)

oral language users learning sign as auxiliary information

autistic children

mentally retarded children

There may be other significant subgroups (including nonhuman primates), but these four give us a place to begin.

Children Learning Sign as a First Language

It is commonly observed (e.g., Schlesinger and Meadow, 1972) that children of deaf signing parents start acquiring first words earlier than their counterparts who are acquiring speech. First signs have been reported as early as four and one-half months.¹³ This observation clearly supports Brown’s hypothesis that “the first steps in language learning must be easier with signs, than with words” (1977:1), but does it necessarily imply that iconicity is the key to learnability?

In answering this question, it may help to distinguish three facets of any language acquisition: the structure of the sign to be learned, the linguistic and cognitive abilities of the learner at the time of acquisition, and the process by which the sign is taught or modeled by a mature language user. Empirically, many, if not most, of the early signs children learn when acquiring sign as a native language may be highly iconic—but that does not imply that either the child or the pedagogue is conscious of that iconicity. Donald Moores (1977) reports that extensive longitudinal observations of deaf parents interacting with their young children have yielded no examples of adults using iconicity to explain the meanings of signs, even though the signs were often objectively iconic. We know that signing children do invent signs, often from a clearly gestural base. Bellugi and Klima mention a three-year-old deaf child who “invented a nonce form for ‘cinnamon roll’ which she made with a cupped hand representing the roll, and an active pointing hand indicating the swirls of cinnamon sugar on top of the roll” (1976:515). Carl Kirchner at the Kendall School in Washington, D.C., reports that his son, at age nineteen months, invented a sign for “doorbell” by pressing in on his nose with one finger. A use of the iconic properties of signs? Obviously so. Yet we have no reason to assume that because iconicity is used by the child in invention it will necessarily be perceived independently in an existing word, or understood as a teaching device in the hands of a pedagogue.

Studies of children’s unfolding metalinguistic abilities in spoken language have shown that unconscious linguistic behavior typically precedes its conscious counterpart (Gleitman, Gleitman, and Shipley, 1972; Cazden, 1975). Children can, for instance, correct their errors before they can respond to corrections from others, or comment on the grammaticality of a model which is presented. There is no reason to question that the awareness of iconicity (either through one’s own perceptions or through the instruction of another) may indeed be important in somewhat more mature children learning sign, but this does not constitute evidence for Brown’s hypothesis about iconicity in the early stages of the acquisition of sign as a native language.

Oral Language Users Learning Sign as Auxiliary Information

There are two types of learners we can include in the category of language users who have some command of a spoken language: young children of the sort Brown studied and adults who take part in similar studies with isolated signs—n.b., we can also include in this latter category the “early stages” of sign acquisition by mature (spoken) language users who set about learning sign as a foreign language. In these cases of auxiliary language acquisition iconicity does prove an important asset to learning. Brown’s experiments clearly indicated that iconicity was a significant factor in children’s memory for signs. Mark Mandel (1977), using a different experimental design with adults, reached a similar conclusion. In neither case can we really be sure whether is was the presence of a prior spoken language which allowed the subject to conceptualize the iconic relationship between the object and its referent or whether general cognitive maturity was the key to making productive use of iconicity. Perhaps there was some altogether different feature, inasmuch as iconicity may also prove significant for certain socially normal but mentally deficient children (see below).

Autistic Children

A population recently highlighted in sign language research is autistic children. A growing number of studies (e.g., Creedon, 1973; Webster et al. 1973; Bonvillian and Nelson, 1976; Fulwiler and Fouts, 1976; Baron and Isensee, in press) has confirmed the viability of simultaneous communication (signing and speaking) as an entrance into language for many autistic children for whom speech therapy has failed. While most of these programs were conceived and carried out independently, their derivative findings have been surprisingly similar. Many autistic children who learn signing begin using language (in this case sign language) spontaneously for the first time (e.g., Webster et al. 1973); Bonvillian and Nelson, 1976). Some children who become good signers start to speak spontaneously—i.e., with no special oral language therapy (e.g., Creedon, 1973; Oxman et al. 1976), even to the point of dropping the manual communication and only speaking (e.g., Miller and Miller, 1973; Oxman et al. 1976). Finally, in many cases, a decline in destructive activity and growth in social behavior has been observed as the ability to use language (in this case, sign language) develops (e.g., Bonvillian and Nelson, 1976; Fulwiler and Fouts, 1976).

Why has sign language (or at least teaching lexical items from ASL) been so successful with many children who have been unable—or unwilling—to learn to speak? John Bonvillian and Keith Nelson (1978) have conjectured that the comparative ease with which these children learn manual signs (as opposed to oral speech) may be attributable to one of three factors: ease of molding the child’s hand and clarity of visual feedback, general superiority of visual and motor skill or auditory-vocal skills in autistic children, or iconicity of the signs themselves. Existing studies using sign with autistic children do little to help us distinguish between these hypotheses.

A study at Brown University (Baron and Isensee, in press) bears on the possible role of iconicity in the learning of signs by autistic children. The study itself was designed to assess the effectiveness of sign language versus spoken language as a means of testing the lexical and syntactic abilities of an autistic child. We worked extensively with a twelve-year-old hearing but mute girl who was clinically diagnosed as autistic.

One of our pretests was designed to determine the subject’s ability to apply a class term to different instantiations of the same referent. For example, if we taught her the sign for “mailbox” in the context of a picture of a mailbox, we wanted to know if she would be able to produce the sign for “mailbox” when shown a toy mailbox or an actual mailbox on the street corner. The testing procedure involved a total of nine items (the signs for “shovel,” “umbrella,” “kite,” “mailbox,” “fence,” “saw,” “guitar,” “scissors,” and “iron”). The subject was first taught the sign in one context (with a criterion of three consecutively correct trials for each item) and then exposed to each item in the two experimental contexts in which it had not been presented initially. The subject was required to produce the sign when the stimulus item was shown (which, as Mandel, 1977, and others have demonstrated, is more difficult than providing a gloss for a sign that is shown or selecting an appropriate picture for a sign).

The results were somewhat unexpected, especially since the subject had earlier been diagnosed as having practically no productive or receptive linguistic abilities. In the testing session that immediately followed training, she scored 100 percent correct. She again scored 100 percent correct when retested after two days, and once again after another six days, with no additional training being given. Although we did not collect comparable data on normal children or adults learning sign for the first time, both Brown’s and Mandel’s data seem to suggest that it is doubtful that normal subjects would match this performance.

Are these results surprising? Yes and no. They were unexpected in light of previous diagnoses of our subject’s linguistic and cognitive abilities but consonant with the behavior patterns typical of autistic children: fixation on objects, insistence on sameness, and eidetic imagery (e.g., Kanner, 1943).

But where does iconicity enter the picture? We have no evidence that the subject did not use the iconic properties of the signs, to the extent they existed. Moreover, we did not explicitly control for iconicity in selecting the signs. However, there was no evidence whatsoever that she was aware of iconic properties that might have been transparent to the experimenters.

One study explicitly examined the role of iconicity in the learning of ASL signs by autistic children. The authors hypothesized that iconic signs would be learned more easily by the subjects than noniconic signs. In fact, the authors conjecture that

The subjects were five autistic and autistic-like children with a mean chronological age of 9 years, 1 month. Unfortunately, since four out of five children were also diagnosed as retarded, it is impossible to determine whether autism or retardation was the characteristic being tested.

The children were taught a total of sixty signs, half of which were judged to be iconic and half noniconic by an independent group of normal children. One-third of the test items were grammatically labeled as nouns in English, one-third verbs, and one-third adjectives:

Subjects were tested by pointing to appropriate pictures, imitating signs which the experimenter made, or producing signs when shown pictures.

The results of the experiment are ambiguous. While iconicity played no significant role for items grammatically labeled as nouns, iconic signs for verbal and adjectival items were learned significantly better than their noniconic counterparts. However, since the overall learning of nouns (iconic and noniconic) was inferior to learning of verbs or adjectives (a result which the experimenters had not predicted), there is reason to question the choice of the ASL signs and illustrative pictures used in the experiment. Before any conclusions can be drawn about the possible use of iconicity in the learning of ASL signs by autistic children, considerably more research needs to be done.

Mentally Retarded Children

Mentally retarded children have successfully learned at least some lexical elements from ASL when spoken language therapy has failed (e.g., Wilson, 1974; Bricker, 1972). Studies summarizing the results of sign language programs with language-deficient children have tended to group together children with social, mental, and, we might add, physical handicaps. It is often difficult to determine that a given child is autistic but not retarded, or vice versa. However, to the extent that we can distinguish, at least in principle, between syndromes—and learning strategies and behavior patterns characteristic of these syndromes—it is helpful to do so.

Until recently, there has been little attempt to determine whether iconicity helps mentally retarded children to learn sign. In practice, as Brown observed, the early signs that both autistic and retarded children are taught are predominantly iconic (again, in the eyes of the instructor); but this guarantees nothing about the actual use either set of children might make of iconicity in learning signs. Because of this lack of data, we decided to collect our own data.¹⁴

The population we studied was composed of six largely nonverbal handicapped children, ranging in chronological age from 6 years, 3 months to 14 years, 11 months. Their mental ages ranged from 2 years, 7 months to 10 years, 1 month. Two of the children were diagnosed as having Down’s Syndrome and were in a classroom for the trainable mentally retarded. Two others were in classrooms for the educable mentally retarded. The fifth child was in a classroom for the learning-disabled, and the sixth was diagnosed as being neurologically and orthopedically impaired. (Despite concerted efforts, it was not possible to locate a homogeneous subject population.)

The children were taught a total of thirty-six signs. The signs were taken from Signing Exact English (as presented in Gustason, Pfetzing, and Zawolkow, 1975), rather than from ASL because a clear set of standardized pictured signs were readily available for Signing Exact English. As in the Konstantareas study with autistic children, one-third of the signs were for referents labeled by English nouns, one-third for verbs, and one-third for adjectives. The signs were also chosen to represent highly iconic, partially iconic, and noniconic signing relationships, as determined by extensive testing with a large group of college undergraduates. In sum, the thirty-six signs were distributed as follows:

Examples of the signs used appear in figure 7.14.

Signs were taught in association with pictures illustrating the object, action, or property to which the sign referred. Examples of the pictures used appear in figure 7.15. Testing was done by having the experimenter show a sign and ask the child to select the correct picture (out of four) which went with the sign, and showing the child a picture and asking him to produce the manual sign associated with it.

Test results revealed that iconicity had no overall significant effect on the ease with which the mentally retarded children learned the manual signs. (Iconic signs for referents grammatically labeled by adjectives were learned slightly better than their noniconic counterparts, but the results were not highly statistically significant.) Can we therefore conclude that iconicity plays no role in the learning of signs by mentally retarded children? Unfortunately, not yet.

All subjects did well in the experiment, averaging nearly 80 percent correct or better in most categories of test items. The test may have been too simple. A more challenging test might have revealed a more significant role for iconicity. Moreover, the subject population was far from ideal. In addition to being heterogenous, its members all had some oral language skills, both in production and comprehension. Ideally, one would like to study sign acquisition by retarded children who neither produced nor showed evidence of comprehending spoken language. Realistically, it is quite difficult to find, much less conduct, controlled experiments with such a group.

Meanwhile, some highly suggestive evidence that retardates do not need iconicity to learn visual language has been coming from the Georgia Retardation Center, which has been experimenting with symbols from Yerkish (the language Duane Rumbaugh devised for use with chimpanzees) with nonverbal retarded children.¹⁵ As shown in chapter 6 (see figure 6.15), the symbols are explicitly designed to be noniconic. The fact that nonverbal retarded children have been able to use arbitrary durable visual symbols to participate in their first rudimentary form of communication suggests that iconicity may also be unnecessary for retardates learning an ephemeral visual set of signs.

Fig. 7.14 Sample Signs Used in Iconicity Experiment with Mentally Retarded Children (Medeiros, 1979)

Fig. 7.15 Sample Pictures Used in Iconicity Experiment with Mentally Retarded Children (Medeiros, 1979)

CONCLUSIONS

It is difficult to draw conclusions about the role of iconicity in language learning within a heterogeneous population about which we still know relatively little. Even at this stage of investigation, though, it is clear that iconicity is not equally important for all sectors of the population. Those who already have reasonable command of one linguistic system (e.g., normal five-year-olds or normal adults) can perceive and make use of iconicity in learning signs. At the other end of the spectrum, iconicity has little or no value in the normal acquisition of signs as a native language. What remains to be seen is which role iconicity may play among people being taught signs as a delayed first language.