The ability to walk upright across an uneven terrain involves astonishingly complex co-ordination of motor and proprioceptive systems. No other primate shares this skill, yet all normal humans master it without specific instruction within the first few years of life, regardless of whether their local environment is a desert plain, a rainforest, a snowy mountainside or a Western city. Attempts to build bipedal robots have met with some success, but the end results still fall far short of a typical 3-year-old's competence. However, a young child who is unfortunate enough to suffer from muscular dystrophy will have difficulty in standing and walking, while retaining intellectual, social and linguistic skills. Although the disease affects muscles throughout the body, difficulty in walking is the most noticeable symptom, and the child will usually retain good manipulative skills and control of facial expression long after being forced into a wheelchair. In many forms of muscular dystrophy, a single defective gene is responsible. Do we therefore conclude that there is a "gene for walking"? Of course we do not. It is apparent that walking depends on integrity of a wide range of underlying systems, involving muscles, nerves, and central control processes that regulate balance, proprioception and motor planning. Muscular dystrophy has a specific effect on just one of these systems - the muscles - but this is sufficient to make walking difficult or impossible.
There are many points of similarity between walking and grammatical competence. Like walking, grammar is a species universal, i.e. common to all normal humans but not seen in other primates, which develops without overt instruction in a wide range of environmental circumstances. Some unfortunate individuals, however, have a specific developmental disorder that selectively interferes with language development, causing marked problems in learning grammatical morphology. There is mounting evidence, from both pedigree and twin studies, that this specific language impairment (SLI) has a genetic basis, and in some cases may be caused by a single defective gene (Hurst et al, 1990; Gopnik & Crago, 1992; Tomblin & Buckwalter, 1994; Bishop, North & Donlan, in press). However, unlike experts in motor control, who have shown remarkably little interest in using muscular dystrophy as evidence for an innate module that controls walking, psycholinguists have used the existence of genetic forms of SLI as evidence for an innate, biologically based module for grammar processing, which has evolved through natural selection and is under the control of one or more genes.
The work that drew attention to this issue is a series of reports by Gopnik and Crago (Gopnik, 1990, Gopnik & Crago, 1991; Crago & Gopnik, 1994) describing a three-generational family, where 16 out of 30 members are affected by a severe form of SLI that is associated with major impairments in the ability to produce and understand regular inflectional endings marking such features as number, tense and aspect. The notion of a grammar gene was introduced tentatively by Gopnik and Crago (1991), who proposed an "interim hypothesis" that a single gene might control mechanisms responsible for learning morphological paradigms. However, those citing this work have often been much more forthright in their claims, and, in some cases, the description of the data, as well as its interpretation, is more extreme than Gopnik and Crago's account. Thus, Gazzaniga (1992) states that affected family members are "unable to form plurals of words, apparently as a result of a particular genetic disorder. The grandfather, father, and sons all have this abnormality in the presence of an otherwise normal language system. ...Mutations to one gene can cause a deficit in one grammatical dimension of a language without affecting others." (p. 81). Jackendoff (1993) gives a more accurate account of the characteristics of affected family members, but comes to a similar conclusion, arguing that: "these results strongly suggest that the impairment is genetic, and that it specifically affects the ability to construct a mental grammar, leaving other cognitive abilities intact. In order for this to be possible, there must be at least one gene that is responsible for a special-purpose mental endowment for language acquisition. The part of Universal Grammar having to do with acquiring inflectional endings must not be a general-purpose learning strategy." (p. 116).
My aim here is not to argue against modularity of grammar, but rather to question whether SLI really does provide evidence for or against this question. I believe the case to have been overstated because three facts about disorders of language development are often overlooked.1. Disruption of basic-level processes early in development can cause selective impairments:
Learning of grammar must involve co-ordinated operation of basic processes of auditory perception, memory, learning, and attention. Why assume that a defective gene causes maldevelopment of a specialised grammar module, when it could have its effect by causing impairment in one of these basic operations?
This line of argument is typically rejected on the grounds that the impairments seen in individuals with developmental language disorders are highly selective. The assumption is made that we can predict the types of language problems that general cognitive or perceptual limitations might cause, and these are not what are seen in SLI. For instance, Gopnik and Crago (1991) explicitly consider the popular notion that SLI might be a consequence of an impairment of auditory perception, but discount this on the grounds that this would mean that the auditory characteristics of language, rather than grammatical functions, should determine which aspects are impaired, and problems should be less apparent in written as opposed to spoken language.
However, neither assumption is safe. In cases of neuropsychological disorders acquired in adulthood, it is reasonable to use the specificity of a language impairment as evidence for damage to a distinct linguistic module. This logic does not apply so clearly, however, in the case of developmental disorders, where low-level general impairments of perceptual or cognitive systems can lead to unexpected patterns of deficit, both because of compensatory mechanisms on the one hand, and because of interactions between different components of a developing system on the other. If we take the extreme case of children with profound congenital hearing loss, we find that they typically have severe difficulty in processing grammatical contrasts in an oral language. Bishop (1982, 1983) showed that on tests of grammaticalcomprehension of English, there were many deaf children who had good understanding of the vocabulary used in the test sentences but who had great difficulty in understanding the significance of contrasts signalled by morphological endings, function words and word order, regardless of whether or not these were perceptually salient. Furthermore, presentation of sentences in a written form did very little to overcome these problems. Indeed, it was noted that there were some deaf children who could read aloud every word in a sentence such as "the boy is not running" or "the man is pushed by the elephant" but still selected the picture of a boy running and the man pushing the elephant. When confronted with complex syntactic constructions, deaf children did not just guess at random, but rather adopted systematic strategies which led to below-chance performance. Even those children with facility in a native sign language had major difficulties in decoding the syntax of English - presumably because the surface forms used in oral languages to mark grammatical functions (e.g. word order and morphological endings) are not well-suited to processing in the visual modality, where simultaneous rather than sequential grammatical processes are the rule. Thus hearing loss did not simply lead to slowing of language acquisition, nor did it lead to a predictable pattern of grammatical impairment affecting only non-salient morphemes. In fact, the distinctive response patterns seen in hearing impaired children were remarkably similar to those seen in hearing children who had been diagnosed as having receptive language disorders. The crucial points emphasized by these studies of deaf children are that a general nonlinguistic impairment in auditory perception early in development can (i) have disproportionately severe impact on ability to produce and comprehend certain grammatical morphology and vocabulary) and (ii) affect written as well as spoken forms of an oral language. Note that this pattern of findings is totally different from what one would see if a hearing loss was acquired in adulthood, after proficiency in oral language had developed. Hearing loss does not just disrupt specific parts of the language system; if acquired early in life, it alters the entire course of oral language development.
Of course, I am not suggesting that the gene that causes developmental language disorders does so by inducing hearing loss. rather, the point is that it is dangerous to imagine that we can predict on a priori grounds what pattern of grammatical impairment would result if the child had a nonspecific cognitive or perceptual deficit that affected language learning, e.g. a disorder of auditory perception such as has been proposed by Tallal et al (1985) or a limitation of phonological short-term memory as mooted by Gathercole and Baddeley (1990).2. Knowledge of grammatical morphology is poor rather than absent in children with developmental language disorders.
Gopnik (1990) made the strong claim that children with SLI lacked a grammatical module that made it possible to learn regular paradigms for inflectional morphological endings such as plural -s or past tense -ed correctly, but she argued that these uses did not reflect rule-based knowledge but arose for one of two reasons. Either (i) the speaker treated inflected and uninflected forms as phonological variants, without appreciating that the inflection conveyed any meaning; or (ii) specific inflected forms were learned as whole items, without appreciating that theywere composed of more than one morpheme. Thus the child might understand that "books" referred to a group of objects, without recognising any morphemic overlap with "book".
Bishop (in press) studied use of grammatical morphemes in the expressive language of 12 children with SLI aged 8 t 12 years who had severe grammatical problems. As Gopnik had noted, it was found that these children did not simply omit inflectional endings such as plural and past tense; rather they produced a mixture of correctly inflected words and words with obligatory inflections missing. in fact, for most constructions, these children more often applied a correct inflection than they omitted one. however, their performance could not be explained in terms of their treating inflections as meaningless phonological variants, because they virtually never made errors of commission, i.e. adding an inflection in an inappropriate context. Thus, when they did use an inflection, they used it correctly. Nor did it seem plausible to account for their correct usage in terms of word-specific learning, because in obligatory contexts the same word was sometimes inflected and sometimes not. In fact, the pattern of performance seen in these children was not so very different from what one observes in normal young children, who also go through a phase of continuing to omit obligatory inflections, even though they often use them correctly. This pattern is hard to account for in terms of a missing component of the grammar. The child seems to have the knowledge of where and when to apply inflectional endings, but sometimes fails to apply that knowledge when producing utterances under real-time constraints. On this view, the difference between children with SLI and young normal children with SLI and young normal children is more a matter of degree than one of qualitative difference, and the difficulties experienced by language-impaired children may have more to do with limited processing capacity in a system that needs to integrate lexical, phonological and syntactic information in on-line production and comprehension, rather than lack of a grammatical module. Notions of limited processing capacity may be justly criticised for being vague, but they do make testable predictions that children's ability to produce and comprehend language will depend not just on the nature and complexity of specific grammatical forms, but also on the overall processing demands placed on the child. They steer us away from looking only at the grammatical aspects of language, and draw attention to the trade-offs that may exist between different levels of processing (Masterson & Kamhi, 1992); and to the effects of variables such as speech rate on linguistic processing (Ellis Weismer & Hesketh, 1993).3. Specific language impairment is less specific than it appears at first sight.
One has only to scan the pages of the Journal of Speech and Hearing Research to realize that it is hard to find a task that involves language on which children with SLI are not impaired. Impairments in phonology, semantics, lexical learning and pragmatics have been widely documented. Of course, one might expect that a primary deficit in one component of language, such as grammar, would have knock-on effects on other language functions, so one might expect to see other impairments as secondary consequences. For instance, if grammar is weak, then vocabulary learning may be affected because the child is poor at using syntacticcontext as a cue to word meaning. however, this line of reasoning is a two-edged sword; if one accepts that two language deficits co-occur, it becomes difficult to know which is primary. Thus, as Watkins (1994) has indeed argued, it could equally well be claimed that a fundamental problem in developing a lexicon of verbs could retard the learning of morphological paradigms for tense and aspect. The challenge for those claiming selective deficits in one domain is to demonstrate that in terms of severity and in qualitative characteristics, this deficit is distinctive and disproportionate.
Another aspect of SLI which is less well known is that the diagnostic term is something of a misnomer; there are several areas of nonverbal functioning in which deficits are reliably found. Although some of these impairments are plausibly explained as secondary consequences of language limitations that impair verbal encoding of nonverbal material, others are not easily accounted for this way. Johnston (1994) and Bishop (1992) have suggested that the apparently specific language problems that we see in these children might just be the most obvious consequences of a general limitation of processing capacity that affects development of any skill that involves retention and manipulation of transient mental representations. Adequate performance would be seen on many nonverbal IQ tests, because these present the subject with concrete stimuli that can be directly manipulated without holding material in memory. This is a speculative hypothesis that needs considerably more testing before it can be accepted; it is not my purpose here to argue for its adequacy, but rather to note that there are alternative explanations of SLI in terms of deficits in systems that are not language-specific but which do play a particularly important role in the normal development of language.
The view of SLI as a testing ground for the notion of the modularity of grammar has undoubtedly stimulated much excellent research in this field; however, the more research is done, the clearer it becomes that SLI cannot be regarded as involving a selective impairment in a single component of the grammar, with all other linguistic and cognitive abilities intact. Perhaps the final word should go to Crago and Gopnik (1994), whose earlier work was used as a basis for some of the more extreme claims in this field: "It is... plausible to assume that, rather than a 'gene for grammar', as certain media reported, there might be genetic alterations to an ability or a variety of abilities that include or underlie grammatical abilities ... Claiming that language is impaired and that there is an impaired gene is not equivalent to describing the specifics of how these two factors fit together... It may be that the impaired gene causes some form of cognitive deficit that is at the base of a language deficit, or it may be that it causes only a language deficit". (pp 45-6).References:
Dorothy Bishop took a first degree in Experimental Psychology at Oxford University in 1973, and went on to do a Masters training in Clinical Psychology at the Institute of Psychiatry in London, before returning to study for a doctorate at the Neuropsychology Unit in Oxford. The subject of her doctoral thesis was children'slanguage disorders, and her principal research interest since that time has been the causes and nature of "specific language impairmentÓ, defined when a child persists in using immature or abnormal language for no apparent reason. In the past it was assumed that these disorders resulted from brain damage, or from inadequate language input from parents, but recent work finds strong evidence of a genetic contribution to this kind of disorder. She is currently senior editor of the Journal of Child Psychology and Psychiatry, and holds a Senior Research Fellowship at Churchill College, Cambridge. She is employed as a Senior Research Scientist at the Medical Research Council's Applied Psychology Unit in Cambridge, UK. As well as numerous publications in scientific journals, she has written a book on Handedness and Developmental Disorder (1990) and has co-edited, with Kay Mogford Language in Exceptional Circumstances (1988). She is currently completing a new book, Uncommon Understanding: comprehension processes in specific language impairment, which will be published by Psychology Press/Erlbaum in 1997.