Review of Neil Smith and Ianthi-Maria Tsimpli, The Mind of a Savant
In Brazil, Ziad Fazah, a Lebanese immigrant, is fluent in 56 languages, says the Brazilian version of the Guinness World Book of Records; he aims by the end of seven years to have learnt the rest of the world's 3000 languages. In Canada, Naiman, Frohlich, Stern and Todesco (1978) found an anthropologist who was fluent in 32 languages. In this book, Neil Smith and Ianthi Tsimpli have now described Christopher, a 33-year-old speaker of sixteen languages or more who lives in a sheltered community in England because he cannot look after himself. At one level, this book forms one of those unique case studies that appear in the linguistics literature from time to time, ranging from the Wild Boy of Aveyron to Genie to Laura and so provides a fascinating account of a unique case. Unfortunately, for a wider public the book demands a high level of knowledge of linguistics of the current principles and parameters type. It is not something that can be recommended to a student or a lay person as a readable account of a unique individual. Doubtless the popular version remains to be written, if Oliver Sachs is interested.
The central issue is the idea of 'double dissociation' - whether language and cognitive level are out of step. Most commonly, this has been investigated through SLI (Specific Language Impairment), where otherwise normal children fail to acquire language, most famously now in the KE family brought to media attention by Myra Gopnik. But even here it is unclear whether cognitive faculties are truly unimpaired in people with SLI. According to Vargha-Khadem, Watkins, Alcock, Fletcher, and Passing-ham (1995), among other differences the average IQ of the SLI members of the KE family was 86 compared to 104 for the rest.
So Christopher can provide a test-case for separation of language from other cognitive faculties in the reverse direction someone who is linguistically normal but otherwise cognitively and socially deficient. It is necessary, then, first to show that Christopher is cognitively deficient, secondly that his first language is normal, and thirdly that he has acquired competence in second languages. Chapter One opens with an account of Christopher's mental capacity - IQ between 42 and 76 according to different tests, and failure similar to autistic children on some tests of point of view, but not on others. Other cognitive tests were an improvised test of Conservation of Number, which he failed, and a Witkin's Embedded Figures test, used here as a test of intelligence not of cognitive style, which he also 'failed'; he can't play noughts and crosses. But there is no detailed comparison of Christopher with other people with language or cognitive disabilities, say, in Yule and Rutter (1987). The last part of the chapter introduces M. Anderson's theory of intelligence, which the authors then revise. We are not told how this theory has been operationally tested with normal children, disabled children, or cognitively disabled adults. In other words, while Christopher would indeed seem to be deficient in non-language aspects of life, there are no precise measurements of his Working Memory capacity, or his performance on scales used to decide whether individuals are capable of functioning outside an institution, to take two examples.
The second chapter sets out to show that his first language competence is perfectly normal. The evidence for this is mostly grammaticality judgements that Christopher gave of sentences such as My shoes is dirty; the sentences are related to current issues in syntactic theory. Since the elicitation was informal, there is no statistical treatment of the results, nor is there an overall balanced scheme of exposure to the same sentence with lexical variation; as each sentence is treated as a one-off. It is a sentence-token test rather than a grammaticality judgements test. The results show that Christopher appears to be a normal speaker of English, scoring at maximum level (age 16:10) on a reading test and 121 on a Peabody Picture Vocabulary test (norm 100). He did, however, consistently reject sentences with topics that are fronted or repeated at the end such as Susan, I met her yesterday and He stayed at Mary's house, Steven, and with sequence of tenses in sentences such as It is time that we left for Newcastle. The authors conclude that his 'linguistic competence in his first language is as rich and as sophisticated as that of any native speaker' with patchiness due to interference from other processes in using sentences. Since there is no comparison with other native speakers' judgements, it is hard to know whether one should agree. For example I am doubtful that Christopher is actually stepping out of line in rejecting sentences such as Susan, I met her yesterday. I suspect that a fair number of naive native speakers would reject these in these circumstances for prescriptive reasons of verbal hygiene. But a full picture of Christopher's language ability cannot be derived from these grammaticality judgements, since there is again a lack of standardised tests or profiling techniques such as vocabulary size, Mean Length of Utterance (the authors at one stage remark that his conversation 'can tend to be monosyllabic', which might suggest a low MLU), or checklists of conversational abilities such as those in Lesser and Milroy (1993).
The third plank in the argument is his ability to speak over sixteen languages. Chapter One attests to this ability via 1 to 3-line examples of his translations. Chapter Three provides evidence in the form of translation tests on English/French cognates with a comparison to other L2 learners and one native speaker. Here Christopher scored fourth out of the group of 15 people, some with A-levels. Of the three people above him, one is French and one a professor of linguistics. Though this is quite an exhaustive test, the results are presented as percentages with no statistical treatment attempted. The bulk of Chapter Three consists of grammaticality judgements, forced choice tests on sentences and sentence completion tasks, concentrating on the pro-drop parameter and word order of VS elements. Results show that he accepted null subject sentences in the appropriate languages; he tended to reject that trace sentences that are in fact grammatical in pro-drop languages; and he tended to reject VS sentences in languages where they are in fact grammatical (Greek, Spanish). The conclusion is his L2 knowledge chiefly consists of lexical and morphological properties, his syntax being largely transferred from English, the result of differences in the 'processing load' imposed by lexicon and syntax. No information is offered on his phonology and virtually no comparison is made with any other research into L2 learning. Since the research techniques are individual it is hard to compare Christopher with other learners. Again the test sentences appear to have been one-offs rather than a balanced design with several specimens of each type. The results may well be correct but it is not clear that one can accept them as firm evidence. The recent SLA literature abounds with tests that could have been used with Christopher.
The main gap from an SLA perspective is that, putting to one side whether the picture of Christopher's second language linguistic competence is adequate, there are only a few hints on the linguistic input and the nature of the process through which he acquired the second languages; he uses grammar books for Dutch; he learnt French from his sister's French books at the age of six; he learnt Greek from Hugo's Greek in Three Months; he has visited some other countries such as The Netherlands. But this is effectively all we are told. Learning a language from grammar books and written forms in this way may in itself account for the lexical bias in Christopher's acquisition. Indeed it may be a classic case of the inefficiency of 'learning' in Krashen's sense: you cannot indeed learn proper syntax by this route, particularly if the grammar books you are learning from do not even mention things like pro-drop, which seem likely to be the case. Christopher's language exposure must have been deprived in many ways. His attempts can be compared with the lack of success of students of the traditional grammar-translation method of language teaching, now banned from English state schools by the 'communicative' National Curriculum. Christopher's constant use of translation may betray that this was indeed the mould that his textbooks had forced him into. Without further evidence of clear differences from other L2 learners who had indeed followed this route, there is no reason to see his case as telling us anything general about second language acquisition apart from the well-known deficiencies of the grammar-translation method.
This criticism is partly countered by the authors' attempts to teach Christopher new languages. The method they used, however, was of course grammar translation: 'sentences, accompanied by a word-by-word gloss and a free translation ... a brief paradigm of the "subject markers"'. One language was Berber, where he did succeed in accepting VSO order, but this tendency, argue the authors, was 'probably a result of his conscious awareness ... an example of meta-linguistic knowledge ... not ... linguistic competence in the technical sense'. He also deduced that Berber was pro-drop without having had any examples. The other language was a made-up language called Epun (Nupe in reverse?) which had supposedly impossible constructions such as structure-independent movement and morphological agreement resolution rules; Epun was taught to four first-year linguistics undergraduates, too. All of them found the structure-independent rules impossible, but Christopher could solve the morphological problems while the others could not. This suggested that Christopher's talent is indeed with morphology. Again no reference is made to the vast literature on the learning of Micro Artificial Languages, and results are sentence by sentence with no detailed comparison between the four other subjects and no statistics. And of course the reliance on morphology may be the result of years of Hugo's rather than of a built-in bias.
The final chapter puts the argument together by revising Anderson's model. Christopher's problems are seen as 'slowness in encoding ideas into the language of thought'. The mind divides into an executive and specific processors. It is Christopher's central executive shared with other cognitive processes that is deficient, particularly the control function. Yet there is no reference anywhere in this chapter to SLA theories such as those proposed by Krashen, Pienemann and Bialystok, with their reliance on control and processing.
Tracing the argument in some detail magnifies the faults from a second language acquisition perspective and lowers the perception of what is in fact an extremely interesting book, clearly argued from a firm linguistics perspective. It will be useful in the questions that it poses and the possible solutions it has to offer, if not in the precise details of the research. The snags about the book are also its strengths: its concerns and methodology are within a particular branch of linguistics that is currently flourishing but pays little heed to the methodology and claims of surrounding areas such as first language acquisition, language disability and second language acquisition research. And it leaves open the question of whether the deficiencies in Christopher's remarkable L2 abilities are simply the product of his necessarily limited contact with the L2. As a contribution to empirical study of second language acquisition it does not rank highly, since its links to the field are sketchy both in terms of methodology and of theory; as an account of a fascinating single case it does not fare particularly well, since its descriptions are so theory-dependent that they are not accessible to the non- specialist and not re-interpretable in terms of the theory to come in five years. But as a provocative account of interesting ideas on the relationship of language and other mental faculties, it is extremely good value.
Lesser, R., and Milroy, L. (1993). Linguistics and Aphasia, Longman, London
Naiman, N., Frohlich, M., Stern, H.H., and Todesco, A. (1978). The Good Language Learner, OISE, Toronto
Vargha-Khadem, F., Watkins, K., Alcock, K., Fletcher, P., and Passingham, R. (1995). Praxic and nonverbal cognitive deficits in a large family with a genetically transmitted speech and language disorder. Proc. Natl. Acad. Sci. USA, 92, 930-933
Yule, W., and Rutter, M. (eds.). Language Development and Disorders. MacKeith Press, London, 248-261.