Language Accessibility in a Transliterated Education: English Signing Systems

Kelly Stack

The key question to be asked about the use of English signing systems with deaf children is Do they work? Do children exposed to English signing systems actually acquire competence in English? To answer this question, we must examine how children acquire natural human languages. Are the same processes available to children acquiring English signing systems? What have we learned from the study of English competence in children exposed to English signing systems?

In this chapter, I will argue that English signing systems do not offer children the input needed to trigger development of English grammar. First, after comparing SEE II (Signing Exact English¹) with English and other natural human languages, we will see that English signing systems intrinsically lack critical characteristics of human language, particularly at the phonological level, which is the level responsible for mapping physical signals received by the ears or eyes to the mental constructs, or phonemes, that constitute the “basic building blocks” of language. Next, the chapter will examine the outcomes of numerous studies of children attempting to learn and use SEE II, and we will see that many of the difficulties children experience learning SEE II can be attributed to “dysharmony” at the phonological level. Finally, we will look at the case of one child who, despite almost perfect circumstances for acquiring SEE II, actually rejects those aspects of SEE II that are least natural, leading her to develop an idiolect that, while linguistically sound, is neither complete nor conventional.

THE ROLE OF PHONOLOGY IN LANGUAGE ACQUISITION

Phonology imposes psychological order on the chaotic physical world of speech sounds. Although acoustical measurements may show that the physical intervals between a series of sounds that change slowly from [d] to [t] are identical, human beings consistently perceive those intervals quite differently depending on which language they grew up with. Thus, how we perceive the smallest units of language depends on which language we use natively.

In addition to organizing the smallest units of language into rule-governed classes, phonology also regulates the production and perception of entire utterances, grouping words into phrases based on prosody. In general, prosody may be thought of as the underlying rhythm of a language, and different languages use different prosodic indicators; in English, stress plays the prominent role; in French, it is syllable structure; other languages use pitch accent or tone.

For example, compare the syntactic constituent structure of the sentence “He kept it in a large jar,” with the prosodic structure (Hayes 1989) in figure 1. The content words (also known as open class or lexical items) are the heads of the three prosodic words (also called clitic groups), and the grammatical words (also known as closed class or functional items) group with the heads within syntactic phrases. So, although, syntactically, the words in a large jar form a single phrase, in speech, we place stress on both large and jar, creating two prosodic phrases, and the entire sentence consists of just three prosodic words: [_che kept it] [_cin a large] [_cjar].

It is widely accepted that phonology treats open class morphemes differently from closed class morphemes. In general, closed class morphemes, both free (such as in, he, a) and bound (such as -ing, -ly, -s), are unstressed, undergo phonological reduction (I am going to becomes I’m gonna), are underlyingly underspecified and subminimal (English past tense is pronounced [t], [d] or [әd] depending on the neighboring sounds), and are subject to cliticization within phonological phrases, as shown in figure 1 (Selkirk 1978, 1980, 1981; Hayes 1989; Nespor and Vogel 1982; Abney 1987).

Considerable evidence indicates that the comprehension of speech depends critically on prosodic structure and on its role in marking lexical and grammatical boundaries in the speech stream. Prosody maps onto and provides cues to linguistic segmentation, and it is essential for comprehension of the message.

Prosody and rhythmic structure also appear to be crucial to language acquisition. There is evidence that children acquiring English use stress patterns in continuous speech to work out whether to use the grammatical classification of open or closed classes of words and morphemes (Gleitman et al. 1988). In particular, closed class functors serve “as a kind of frame for perceiving and producing content words” (Gerken, Landau, and Remez 1990, 213), even when they do not regularly appear in small children’s speech.

FIGURE 1.

Source: B. Hayes, “The Prosodic Hierarchy in Meter,” Phonetics and Phonology 1 (1989): 201–60.

THE STRUCTURE OF ENGLISH SIGNING SYSTEMS

To understand that SEE II’s awkwardness is not merely aesthetically displeasing but also fundamentally flawed in comparison with ASL and other signed languages, we need to look at the relationship between the phonetic and phonological domains in language. Considerable evidence shows that English signing systems are incapable of reproducing the prosody of spoken English, particularly its stress patterns. This incapacity appears to be the primary reason why they are unsuitable language models.

In SEE II, the citation-form vocabulary of ASL was used as the basis for developing signs. English word order was followed. Signs were invented to represent inflectional and derivational morphemes such as -ING, -S, -NESS, etc. Initialization was used to create new signs to represent the individual vocabulary items within “families” of English words that “belong together conceptually.” For example, the system maintains “the basic ASL sign for BEAUTY, using the P handshape in the same movement for PRETTY, and the L handshape for LOVELY” (Gustason, Pfetzing, and Zawolkow 1980).

The bound and functional (closed class) morphology of SEE II consists of mainly irreducible forms. A reducible form is one in which the underlying structure is not fully specified. In English, for example, we commonly say nouns are pluralized by adding an -s. In fact, the s sound sounds like an s only when the word ends in a non-voiced, nonsibilant segment, like the p in top. If a word ends in a voiced, nonsibilant segment, like the g in dog, then the plural form ends in -z. If a word ends in a sibilant, an entire syllable is added, such as -ez as in judges. The explanation for all this variation is that the English regular plural form is neither -s nor -z; it is a segment that is not underlyingly specified for voicing. It “borrows” its voicing feature from the neighboring segment in the word to which it is attached.

Like English and other spoken languages, ASL and other natural signed languages contain morphemes with underspecified segments. For example, many verbs in ASL have subject and object agreement morphology; they are signed beginning in one location and ending in another depending on the subject and object. GIVE is a commonly cited example, signed with the beginning location closer to the body in I GIVE-TO YOU but with the ending location closer to the body in YOU GIVE-TO ME. The subject or object agreement morphemes are specified for location but are not specified for hand configuration or movement; those elements are “filled in” by the verb.²

Although Gustason and Zawolkow (1993) provide suggestions for incorporating “ASL principles” into English signs, efforts such as these are destined to become mired in the underlyingly fully specified structure of SEE II. For example, in figure 2, the idea of dancing and the grammatical notions of person, existence, negation, mood, and aspect are represented by seven signs in SEE II; two signs in ASL; and two prosodic words in spoken English ([wi ’arnt] [gәnә ’dæns]). Although speed and fluidity can be enhanced if the SEE II signer follows Gustason and Zawolkow’s suggestions and pronounces the signs in closer proximity to one another, no real phonological reduction occurs.

FIGURE 2.

Source: G. Gustason, D. Pfetzing, and E. Zawolkow. Signing Exact English (Los Alamitos, Calif.: Modern Signs Press, 1990) and T. Humphries, C. Padden, and T. J. O’Rourke, A Basic Course in American Sign Language (Silver Spring, Md.: T.J. Publishers, 1981).

In natural languages, the physical phonetic space (acoustic or visual) works in harmony with the psychological domains of language such as phonology, morphology, and syntax. In signed languages, phonologically dependent elements are ordinarily realized as simultaneous morphemes rather than as morphemes occurring sequentially. For example, the marking of aspect in ASL involves the phonologically dependent element, movement. Different aspects are marked by modifying the movement parameter, which is realized simultaneously with the handshape and location morphemes. Put another way, there are no natural languages in which morphological and grammatical processes are not accompanied by phonological processes that are reflected in the physical phonetic signal. SEE II, however, exhibits very little relation between its phonetics and the rest of the language. SEE II grammatical processes are not reflected in the pronunciation of SEE II signs.

PREVIOUS STUDIES

The literature is clear that children do not acquire English signing systems as first languages. One of the well-known facts about normal language acquisition in children is that it occurs consistently under all but the harshest of conditions; as long as there is access to the ambient language, children who are neglected or even abused attain grammatical competence equal to peers who have been bombarded with nurturing attention. The fact that children do not acquire the ambient language when it is SEE II must be explained.

Children exposed to SEE II are reported to sprinkle signs for inflectional morphemes ad hoc throughout their utterances (Maxwell 1987; Suty and Friel-Patti 1982). These children tend to sign affixes as separate words, even on those occasions when they produce them in the correct order with a root. One child “developed a strategy of signing -ING any time his father indicated that a correction was necessary, whether or not -ING was appropriate” (Maxwell 1987, 332).

Children exposed to SEE II from age three to age six did as well on the GAEL (Grammatical Analysis of Elicited English) Simple Sentence Level Test³ as deaf children at the same age who had been exposed to no signed language at all (Suty and Friel-Patti 1982). In other words, previous exposure to SEE II did not enhance their ability to understand English.

Another group of children, ages seven to fourteen, from programs very committed to SEE II, “produced roughly half their sentences perfectly (M = 55%, SD = 20)” (Schick and Moeller 1992). Their errors mainly had to do with the use of bound morphemes, articles, copulas, and auxiliary verbs (i.e., closed class items). To place these results in context, grammatical competence for native users of a language is normally defined as use of obligatory morphemes in 90 percent of the cases. Children natively acquiring a language simply do not make the quantity and quality of mistakes evidenced by children exposed only to SEE II.

Supalla (1991) found that deaf children who grew up in a strictly SEE II-signing environment failed to acquire the SEE II (and thus the English) pronoun system. They pronounced SEE II verbs correctly only 20 percent of the time, leaving off tense markers and innovating subject-object verb agreement systems. Most disturbing, Supalla found that, although these children were in school together, each child had innovated his or her own verb inflection strategy, which was internally consistent for the child but not shared with peers.

These studies show that children exposed to English signing systems do not make the kinds of errors one expects from a child who is acquiring a natural spoken or signed language. They omit or misuse grammatical morphemes and function words. They treat affixes like whole words. The pattern of errors they make diverges considerably from the error patterns of children who are acquiring natural languages.

JAMIE’S STORY

Unfortunately, most children using English signing systems are not exposed to the systems very early in life. Therefore, the errors they make may be attributable to lack of access to any language during the critical period for language acquisition (see chapter 2 by Monikowski in this volume). The child in this study, however, was exposed to SEE II at a relatively early age, so any difficulty she might have experienced mastering the system was likely to have been caused by deficiencies within SEE II rather than by declining ability to acquire language because of critical period effects.

“Jamie” is a deaf child of hearing parents, who was exposed to SEE II from the age of eleven months.⁴ Both Jamie’s parents are avid and skilled SEE II signers, and at the time the study began, when Jamie was four years and three months, she had not been exposed to American Sign Language. At about age five, Jamie began to have contact with deaf children of deaf adults (and therefore began to have contact with ASL) in the context of the SEE II preschool and kindergarten she attended. The study ended when Jamie was five years and five months. Initial testing showed that Jamie possessed an excellent open class vocabulary (i.e., nouns, verbs, adjectives, adverbs) equivalent to that of hearing children her age. Jamie’s open class phrase structure was appropriate; moreover, she used open class vocabulary in a wide variety of standard grammatical functions. Noun phrases surfaced in her utterances as subjects, objects of verbs, objects of prepositions, appositives, and complements of determiners. Jamie’s use of SEE II verb phrases (except for the copula) in spontaneous conversation was, for the most part, normal. She used intransitive and transitive verbs, including ditransitives and transitives with sentential complements and adjuncts. Analysis of spontaneously occurring conversation revealed strong evidence that Jamie had standard English (i.e., SEE II) prepositional phrase structure; prepositions always appeared with complements, and they served standard adverbial, predicative, complemental, or adjectival functions. Finally, Jamie’s use of adjectives and adverbs was, by and large, congruent with standard English.

However, as will be discussed below, when her functional vocabulary (i.e., determiners, complementizers, inflectional elements such as -ing, -s, etc.) was evaluated, a different picture emerged, one of significant developmental delay. Results of testing using the CYCLE⁵ (Curtiss and Yamada 1987) revealed that Jamie at age four years, three months had acquired the closed class vocabulary that one would expect of a normal hearing child age two years, six months.

Jamie’s closed class phrase structure deviated remarkably from what would be expected from a child Jamie’s age having the same mastery of open class vocabulary and phrase structure. Jamie not only routinely omitted grammatical morphemes in obligatory contexts but also invented her own grammatical innovations. This performance is remarkable given that the literature of normal children acquiring natural spoken or signed languages describes no cases of grammatical innovations. Jamie’s acquisitional gaps and innovations in this area indicate a profound breakdown in the ability of SEE II to provide children with the tools they need for normal language acquisition.

In spontaneous conversation, Jamie produced the following SEE II pronouns: I, ME, YOU, SOME+ONE, ONE, and IT. When she used them, she used them in the correct context, but Jamie did not always produce a pronoun when there was an obligatory context for one. In addition to the SEE II pronouns, Jamie used an innovated form, a pointing gesture, for third-person references, which I glossed as PROPOINT. Jamie sometimes pluralized PROPOINT by reduplicating it or by tracing an arc in the air. She also pluralized YOU in two instances by reduplicating it, even though the English and SEE II forms of YOU are the same for singular and plural. The SEE II pronominal system is shown in figure 3.

FIGURE 3.

Source: G. Gustason, D. Pfetzing, and E. Zawolkow. Signing Exact English (Los Alamitos, Calif.: Modern Signs Press, 1990).

Jamie produced a first person singular pronoun in 74 percent of obligatory contexts. Most of the time, this pronoun was ME. In subject position, she used I only 17 percent of the time, but in object position, ME was always used. An important point to note is that Jamie never used I in object position, which would have been an error unattested in the literature on English child language acquisition. In contrast, her misuse of ME as subject is attested in spoken English acquisition (e.g., “Me do it.”), so in this respect, Jamie was not unusual, just immature. All in all, Jamie’s performance with first and second person pronouns was comparable with that of a child’s early English.

Although children acquiring spoken English tend to master third person singular pronouns by approximately the age of two, Jamie did not master SEE II third person pronouns at any point during the period of study according to her scores on the CYCLE-R subtests involving comprehension of pronouns.

In spontaneous conversation, out of 572 contexts that were obligatory for a third person pronoun, Jamie supplied a SEE II third person pronoun only eleven times (2 percent). In conversational contexts, she used first person pronouns in 69 percent of obligatory contexts and second person pronouns in 93 percent of obligatory contexts, otherwise omitting them. These results are summarized in figure 4.

Why did Jamie do so much better with first person and second person SEE II pronouns than with SEE II third person pronouns? Such a dramatic disparity involving mastery of different types of pronouns has never been reported in normal children acquiring natural spoken or signed languages. The answer lies not in grammatical person but in the morphophonemic structure of the pronouns. Jamie was able to acquire SEE II pronouns that use phonological location morphemically. Most known signed languages use phonological location for pronominal reference, establishing a point in space as a referent for a noun. But instead of the SEE II third person pronouns that do not make use of space, Jamie innovated a third person pronoun (similar to ASL’s third person pronoun): POINT, which she used consistently throughout the period of study. Of those 572 contexts that were obligatory for a third person pronoun, Jamie used the POINT pronoun 40 percent of the time.

FIGURE 4.

In the original version of the CYCLE-R Third Person Object Pronouns subtest, the subject is shown a page with three pictures on it and is told, “Point to them,” or “Point to her,” etc. I created a revised version in which I briefly described each picture (e.g., “THE GIRL IS SIT+ING ON A CHAIR.”) and established a pronominal location in space for it, using SEE II verbs (in this case, SIT+ING) as locational predicates. To eliminate the possibility of copying the locations from the pictures, the pronominal locations I signed were not identical to the position of the pictures on the page. When I asked Jamie to identify the correct picture, I used the POINT pronoun to refer to the pronominal locations. Jamie passed with 100 percent correct responses on the revised version. These results are shown in figure 5.

Jamie’s perfect performance on this subtest is particularly remarkable because it shows mastery of the use of arbitrary locations in space for grammatical purposes. Ignoring the case and gender inflection found in SEE II and English, Jamie’s developing pronominal system instead used location to mark person, making systematic use of a morphosyntactic feature of natural signed languages that is almost ignored by her target language, SEE II. This pattern also extended to possessive pronouns.

Although exciting from the point of view of a generative linguist interested in child language acquisition, Jamie’s innovations were disheartening in terms of acquiring English or SEE II. Instead of learning that pronouns in English are marked morphologically for gender, Jamie had developed a system in which pronouns agree with their referents without regard to gender.

Jamie’s innovations did not stop with pronouns. In English, number on nouns is expressed in most cases through the bound plural morpheme, -s. In SEE II, the regular plural is formed by concatenating an -S to the end of the noun; irregular plurals (such as man or men) are formed by reduplicating the noun (except in the case of 0-morpheme plurals such as fish, which are not marked for number).

FIGURE 5.

Jamie used an overt SEE II plural marker sixteen times in 248 obligatory contexts, two times using –S and fourteen times using reduplication. None of the reduplicated forms were irregulars and, thus, should strictly count as SEE II errors. In testing, Jamie was unable to reliably comprehend or produce the plural morpheme –S. A pretest demonstrated that Jamie understood the difference between singular and plural, but in the absence of the modifiers ONE and MANY, she was unable to distinguish SEE II singular and plural morphology, performing at about the same level as one who is responding by chance.

However, Jamie performed with 90 percent accuracy on an altered version of the CYCLE-R Noun Singular/Plural subtest. This time, instead of using the SEE II –S morpheme, I presented the items using reduplication for plurals. The result suggests that Jamie was not learning to pluralize nouns as SEE II does but was, instead, developing a grammar in which all nouns, not just irregulars, are marked for plurality through reduplication.

Out of a corpus containing 1,122 obligatory contexts, Jamie produced just fifty-six utterances with SEE II inflectional forms in them. Almost all of these inflectional forms were free morphemes such as modals and negatives rather than bound morphemes such as suffixes and prefixes.

Jamie tended to lexicalize grammatical morphemes where possible, omitting bound morphemes. For example, she distinguished between plural and singular nouns by using lexical items such as ONE or MANY without marking the plurality on the noun itself. Jamie never used either of the two SEE II devices for marking past tense (shown in figure 6). She did occasionally use the lexical completive ALREADY instead of a past-tense bound morpheme.

In spoken and signed languages with rich verb agreement systems, overt lexical subjects, objects, or both are frequently optional. ASL is such a language, as illustrated in figure 7, which shows an example where the inflecting verb TEACH may grammatically occur with or without lexical arguments (a) and (b), but the plain verb LIKE may occur only when accompanied by lexical arguments; that is, (d) is ungrammatical, as indicated by the “*”.⁶

FIGURE 6.

Source: G. Gustason and E. Zawolkow. Signing Exact English (Los Alamitos, Calif.: Modern Signs Press, 1993).

FIGURE 7.

Jamie produced ten verbs that appear to inflect for person, number agreement, or both and that could be modified in terms of their beginning or ending phonological locations. One might be tempted to view these modifications as ad hoc mimetic extensions, but Jamie’s systematic treatment of these verbs with respect to argument structure strongly suggests they were indeed grammatical innovations.

Jamie provided required lexical arguments for noninflecting verbs such as LIKE 83 percent of the time. However, she provided lexical arguments for inflecting verbs such as TEACH only 52 percent of the time (see table 1).

Why would Jamie suddenly “lose track” of subject-object requirements for just the inflecting verbs? The mystery is solved when we include agreement morphology as an additional way to satisfy the requirement for a subject or object, as it is in highly inflected spoken languages. In this case, we find a combined total of lexical and nonlexical arguments appearing 92 percent of the time in obligatory positions, as summarized in table 2.

Jamie’s differentiation between inflecting and noninflecting verbs by supplying bound inflectional morphemes as arguments for the former (e.g., ₁TEACH₂; “I taught you”) and free lexical items for the latter (e.g., ₁INDEX LIKE ₂INDEX; “I like you”) is strong evidence that her innovation is principled, not ad hoc.

TABLE 1

TABLE 2

In contrast, Jamie used SEE II functional elements in unprincipled ways that are unattested in children’s normal English language development. All of Jamie’s SEE II inflectional structures were fragile, occurring in a very small proportion of obligatory contexts.

In many respects, her acquisition and use of SEE II resembled those of children exposed to language after the critical period of language acquisition, who characteristically acquire relatively large open class vocabularies but little or no grammar. Unlike those children, however, Jamie, who was still within the critical period when exposed to SEE II, additionally innovated a rich but idiolectic grammatical system.

Children who are normally acquiring natural spoken and signed languages never reject entire grammatical systems in the target language and instead innovate their own systems, even though they may play creatively with limited aspects of some systems. With SEE II, Jamie did what normally is not done. Jamie herself is normal, and her acquisitional circumstances were nearly normal (early exposure to language from both parents). We can conclude only that the use of SEE II produced a deleterious effect on Jamie’s ability to acquire English and that this use has the same effect on other deaf children.

NOTES

1. In the late 1960s and early 1970s, three varieties of English signing systems sprouted up in as many years (Gustason 1990). The most widely adopted of these has turned out to be SEE II (Signing Exact English).

2. This discussion of underspecification in signed and spoken languages is intended to introduce the concepts for the nonlinguist, not to make claims based on any particular phonological theory.

3. According to Suty and Friel-Patti (1982), “This test assesses the English language competence of deaf children for basic grammatical categories in simple sentences and for the use of grammatical function words and inflectional affixes” (p. 156).

4. Eleven months is still rather late in terms of language acquisition, but it is well before the age of four, which has been identified by Newport (1990) as being too late for normal language acquisition to occur. Jamie’s parents describe a burst of language acquisition by Jamie as soon as she was exposed to SEE II, suggesting that she attained the one-word stage at a relatively normal age, despite her late start.

5. The Curtiss-Yamada Comprehensive Language Evaluation (CYCLE) was developed to provide information about the comprehension and production of specific English grammatical structures and features in semantics, syntax, morphology, and phonology. The CYCLE consists of many subtests that are divided into CYCLE-R (receptive language) and CYCLE-E (expressive language).

6. Lexical arguments are noun phrases that must co-occur with verbs, usually as subjects or objects; an example would be it in it is raining. It would be ungrammatical to say is raining because the verb rain requires a subject.

REFERENCES

Abney, S. P. 1987. The English noun phrase in its sentential aspect. Ph.D. dissertation, Massachusetts Institute of Technology, Boston.

Curtiss, S., and J. Yamada. 1987. Curtiss Yamada Comprehensive Language Evaluation (CYCLE). Los Angeles, Calif.: University of California at Los Angeles (UCLA) Linguistics Department.

Gerken, L., B. Landau, and R. E. Remez. 1990. Function morphemes in young children’s speech perception and production. Developmental Psychology 26: 204–16.

Gleitman, L. R., H. Gleitman, B. Landau, and E. Wanner. 1988. Where learning begins: Initial representations for language learning. In Linguistics: The Cambridge Survey, vol. 3, ed. F. Newmeyer. Cambridge: Cambridge University Press.

Gustason, G. 1990. Signing Exact English. In Manual communication: Implications for education, ed. H. Bornstein. Washington, D.C.: Gallaudet University Press.

Gustason, G., D. Pfetzing, and E. Zawolkow. 1980. Signing Exact English. Los Alamitos, Calif.: Modern Signs Press.

Gustason, G., and E. Zawolkow. 1993. Signing Exact English. Los Alamitos, Calif.: Modern Signs Press.

Hayes, B. 1989. The prosodic hierarchy in meter. Phonetics and Phonology 1: 201–60.

Humphries, T., C. Padden, and T. J. O’Rourke. 1981. A Basic Course in American Sign Language. Silver Spring, Md.: T.J. Publishers.

Maxwell, M. 1987. The acquisition of English bound morphemes in sign language form. Sign Language Studies 57: 323–52.

Nespor, M., and I. Vogel. 1982. Prosodic domains of external sandhi rules. In The structure of phonological representations, part I, ed. H. van der Hulst and U.N. Smith. Dordrecht: Foris.

Newport, E. 1990. Maturational constraints on language learning. Cognitive Science 14: 11–28.

Schick, B., and M. P. Moeller. 1992. What is learnable in manually coded English sign systems? Applied Psycholinguistics 13: 313–40.

Selkirk, E. O. 1978. On prosodic structure and its relation to syntactic structure. In Nordic prosody II, ed. T. Fretheim. Trondheim, Norway: TAPIR.

———. 1980. Prosodic domains in phonology: Sanskrit revisited. In Juncture, ed. M. Aronoff and M. L. Kean. Saratoga, Calif.: Anma Libri.

———. 1981. On the nature of phonological representation. In The cognitive representation of speech, ed. J. Anderson, J. Laver, and T. Myers. Amsterdam: North-Holland Publishing.

Supalla, S. 1991. Manually Coded English: The modality question in signed language development. In Theoretical issues in sign language research, vol. 2, eds. P. Siple and S. Fischer. Chicago: University of Chicago Press.

Suty, K., and S. Friel-Patti. 1982. Looking beyond signed English to describe the language of two deaf children. Sign Language Studies, 35: 153–68.