Grammatical Analysis of Language Disability Crystal, D., Fletcher, P. and Garman, M. (1989), Grammatical Analysis of Language Disability - 2nd Edition. London: Cole and Whurr. Reproduced with permission from the publisher.. This out-of-print volume has been digitized through the joint efforts of: • The authors of the publication: David Crystal, Paul Fletcher and Michael Garman, and • The Child Language Centre (http://www.cmds.canterbury.ac.nz/clc/), Department of Communication Disorders, University of Canterbury, New Zealand The author grants the right to use and distribute this work for scholarly and/or professional use only. Copyright is retained by the authors, and commercial publication and/or distribution is expressly forbidden. Typography and visual imagery © Cole and Whurr. Digital copy created 23 Sep 2011 / Source: http://hdl.handle.net/10092/5483 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. The grammatical analysis of language disability The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. To the memory of Timmy The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. -- ----_ ............... - ~ ............... ~-.- CRYSTAL FLETCHER G GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY STlTDIES IN DISORDERS OF COMMUNICATION SECOND EDITION COLE AND WHURR LONDON The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. --.- @ Cole and Whurr Limited 1989 First published 1976 by Edward Arnold (Publishers) Ltd Second edition 1989 published by Cole and Whurr Limited 19b Compton Terrace, London NI 2UN British Ubrary Cataloguing in Publication Data Crystal, David, 1941- Grammatical analysis of language disability.-2nd ed. I. Language disorders. Analysis. methods. Language assessment. and Screening Procedure Syntactic Remediation I. Title II. Fletcher, Paul III. Garman. Michael IV. Series 616.85'50754 ISBN: I 871381 037 A11 rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means. electronic, mechanical, photocopying. recording or otherwise. without the prior permission of Cole and Whurr Limited. This publication is sold subject to the condition that it shall not, by way of trade or otherwise. be lent resold, hired out. or otherwise circulated without the publisher's prior consent in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed upon any subsequent purchaser. Printed and bound in Great Britain by Antony Rowe Ltd. Chippenham. Wiltshire The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Contents General preface Preface VI Note to the second edition 4 1 The study of syntax 5 2 Some critical factors in devising a syntactic remediation procedure 20 3 The syntactic framework of description 33 4 The development of syntax in children 59 5 Language assessment, remediation and screening procedure (LARSP) 86 6 The linguistic analysis and treatment of language disorders 109 7 Case study of a child 128 8 Case study of an adult 161 9 LARSP 1975-1990 ]97 Appendices A LARSP child data collection instructions B Child language assessment sampling procedure C Communicative history form D Sample of syntactic analysis of adult speech Revised (1981) LARSP Chart: Summary of changes Glossary of symbols References Index (names) Index (subjects) 213 214 217 220 223 227 229 239 242 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. General preface This series is the first to approach the problem of language disability as a single field. It attempts to bring together areas of study which have traditionally been treated under separate headings, and to focus on the common problems of analysis, assessment and treatment which characterize them. Its scope therefore includes the specifically linguistic aspects of the work of such areas as speech therapy, remedial teaching, teaching of the deaf and educational psychology, as well as those aspects of mother-tongue and foreign-language teaching which pose similar problems. The research findings and practical techniques from each of these fields can inform the others, and we hope one of the main functions of this series will be to put people from one profession into contact with the analogous situations found in others. It is therefore not a series about specific syndromes or educationally narrow problems. While the orientation of a volume is naturally towards a single main area, and reflects an author's background, it is editorial policy to ask authors to consider the implications of what they say for the fields with which they have not been primarily concerned. Nor is this a series about disability in general. The medical, social, educational and other factors which enter into a comprehensive evaluation of any problems will not be studied as ends in themselves, but only in so far as they bear directly on the understanding of the nature of the language behaviour involved. The aim is to provide a much needed emphasis on the descrip tion and analysis of language as such, and on the provision of specific techniques of therapy or remediation. In this way, we hope to bridge the gap between the theoretical discussion of 'causes' and the practical tasks of treatment-two sides of language disability which it is uncommon to see systematically related. Despite restricting the area of disability to specifically linguistic matters-and in particular emphasizing problems of the production and comprehension of spoken language-it should be clear that the series' scope goes considerably beyond this. For the first books, we have selected topics which have been particularly neglected in recent years, and which seem most able to benefit from contemporary research in linguistics and its related disciplines, English studies, psychology, sociology and education. Each volume will put its subject matter in perspective, and will provide an introductory slant to its presentation. In this way, we hope to provide specialized studies which can be used as texts for components of teaching courses, as well as material that is directly applicable to the needs of professional workers. It is also hoped that this orientation will place the series within the reach of the interested layman-in particular, the parents or family of the linguistically disabled. David Crystal Jean Cooper The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Preface This book is essentially an attempt to introduce, describe and justify a grammatical assessment and remediation procedure that can be used with children or adults displaying some kind of language disability. In writing it, however, we have also found it necessary to incorporate two other kinds of information: for readers who have had little previous contact with studies of this general type, we have added background infonnation about syntax, and about some of the main studies in syntactic abnormality that have been made in recent years; and for those well versed in the literature on syntactic assessment, we have added points of comparison with our own procedure, and some degree of critique. The first two chapters, accordingly, present information and discussion which is at times introductory, at times technical. We have tried to maintain a reasonably smooth level of exposition by using footnotes for some of the more specialized comments and by a fairly frequent use of bibliographical references for elaboration of issues that we felt to be less central to our purposes. We have nonetheless been very aware of the difficulty of writing for two audiences at once, and we hope that the bringing together of these distinct expository modes will not prove unduly disconcerting. In our view, such problems are minor compared with the need to develop a critical and flexible attitude towards the use of syntactic procedures in remediation; and it is in the interests of developing this awareness that we have written our opening chapters in this way. The book is the outcome of some ten years of work which progressed at various rates and intensities, and involved various stages of formulation and revision. It may therefore be helpful-especially to those teachers and therapists who have had some contact with this approach in courses and conferences in Great Britain to be given a brief historical account of its development. The Department of Linguistic Science was estabJished at the University of Reading in 1965, and regular contact with the Audiology Unit of the Royal Berkshire Hospital began shortly afterwards. The period 1965-9 involved the first author observing and participating in speech therapy and audiology sessions at the RBH, out of which emerged various case studies and partial analyses. Simul taneously, the Department developed its teaching activities in this area, specifically in relation to the Diploma for the Teaching of Speech Therapy (Reading School of Education) and in its MA in Linguistics, on which speech therapists and mother tongue teachers were beginning to be accepted. During this time also, in-service courses were taught by the first author to speech therapists and teachers in the Reading area on the relationship between linguistics and the remedial field, the general tone of which is summarized in Crystal 1972a. The demand for in-service lectures and courses became more insistent between 1969 and 1972, and culminated in a series of courses specifically on linguistics, languge acquisition and applied The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 2 PREFACE remedial work, held at Castle Priory College, Wallingford, Berkshire, throughout 1973 and 1974. For these courses, the decision was made to concentrate on syntax (for reasons explained in chapter I), and a more intensive working-up of the syntactic approach followed. The third author had joined the Department as Lecturer in 1971, and the second author as a Canada MRC Research Fellow in 1973, thus allowing a more systematic application of the approach to a wider range of patients. A number of ex-Reading speech therapy students also began to use aspects of the approach in their work (e.g. Hutchison 1972). Adults were first systematically studied using the procedure in J 972, and between 1972 and 1974, the language assessment, remediation, and screening procedure (LARSP), as it came to be called, was introduced and discussed in relation to fields other than speech therapy-in particular, remedial teaching, the teaching of reading (Crystal I 973b, 1974a), special education, the teaching of the deaf (Crystal I 972b), and educational psychology. It is the widely found support for the approach, and interest in its application in these areas, which has led us to publish the present introduction, rather than wait a few more years for standardized data to become available. We are well aware of our limited experience. LARSP has been used systematically in detailed studies of some 30 children and IO adults only, though a further 200 children and 50 adults have been studied in a more partial way. We have therefore been much reassured by our contact with clinicians, teachers, and others, that despite the acknowledged weaknesses, there seems to be a point in publishing an outline of the approach now. We welcome discussion from readers about any aspect of our procedure, or its application in individual cases. Correspondence about the procedure as a whole should be addressed to the first author; further information about the case studies may be obtained from Dr Fletcher, who was primarily involved with chapter 7, and Dr Garman, for chapter 8. Further copies of the LARSP Profile Chart are also obtainable from the authors, University of Reading, Whiteknights, Reading RG62AA. Our thanks are due to many people who have supported our interests over the past decade. In particular, without the willing cooperation and enthusiasm of Jennifer Schmit and Joan Telfer, the therapists involved in our main case studies, we would never have been able to progress so far in such a short time. The speech therapy staff at the Audiology Unit of the Royal Berkshire Hospital have displayed patience beyond measure in discussing their work with us, especially Ann Owlett, Ann Rundle and Pat Touche. Space unfortunately does not permit us to name all those who have helped us in various ways-especially by sending us material for analysis-from our local hospitals, the Reading and Berkshire Health Authorities, and schools and clinics up and down the country. But we should be at fault if we did not add a special work of thanks to Mr R. Hunt Williams and Dr Kevin Murphy, for introducing us to the work of the Audiology Unit, and allowing us to use their facilities, for the sympathy and assistance of Ray Johnson and Joyce Knowles of Castle Priory College, and to the participants on the courses they organized for counterbalancing our theories with their experience. We must also acknowledge our thanks to those colleagues in the Department of Linguistic The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. PREFACE 3 Science who, while not being directly involved in this work, have nonetheless given generously of their time in criticizing earlier drafts of this book: Frank Palmer, Peter Matthews, Ron Brasington, David Wilkins, Peter Trudgill, Arthur Hughes, and Bill Hardcastle. Naturally, ours is the sole responsibility for the orientation and content of the final version. Above all, to Mr J, and to the parents of Hugh, who gave permission for our intensive case studies to proceed, we are most grateful. Note to accompany the 1981 printing David Crystal University of Reading August 1974 In the five years since this book was published, LARSP has come to be used routinely in many clinical and teaching settings, and several suggestions have been made for its improvement. We have not felt it necessary to make any radical changes in the approach, but we have introduced several minor modifications. In particular, we have resolved some ambiguities in layout and categorization which we failed to anticipate when the Chart was first designed; and we have added a few distinctions which we have found make the approach more discriminating in its clinical application. The modified Chart follows the design of the original Chart very closely, however, so that those who wish to continue with the original version. because it is more familiar, may do so. Also, because the changes are minor, we have not felt it necessary to re-set the charts throughout the book: a summary of the changes, along with a copy of the modified chart, is given on pp. 209-14. A fuller exposition of the modified procedure is given in D. Crystal, Profiling linguistic disability (Edward Arnold, 1982), Ch. 3. Apart from a few previously unnoticed typographical corrections, we have made only the following additional changes: at appropriate places, we have introduced references to the further discussion of the procedure presented in the companion volume, Working with LARSP(Edward Arnold, I 979)-abbreviated as WWL. It should be stressed that the present book gives only an outline of the procedure, and 1acks the fuller i1tustration and workbook material which Working with LARSP provides. David Crystal March 1981 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Note to accompany the second edition We have taken the opportunity of a new edition of this book to review the main developments in the use of LARSP in the past 12 years, and these observations appear in an additional Chapter, which replaces the 'Postscript' of the first edition. At the same time, we have updated our references, mainly to take account of the various books and journal papers we have seen which have used the procedure. We would welcome further information about applications which make use of LARSP, or any of its cross-linguistic adaptations. Copies of the profile chart may now be obtained from the authors, c/o PO Box 5, Holyhead, Gwynedd LL65 1 RG, UK June 1988 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 1 The study of syntax The neglect of syntax Of all the points of contact between linguistics and those who work in the field of language abnormality, none has been more neglected until recently than the study of syntax. To the linguist, this is somewhat paradoxical. Work in general linguistics since the 1950s has clearly shown the indispensability of syntax in the analysis of human language. The two most influential books of the period reflect this concern in their titles: Syntactic structures, and Aspects of the theory of syntax (both by Noam Chomsky). And syntax has come to be seen as the network of organizational principles underlying linguistic expression, without which language would become an incoherent jumble of vocabulary and sound. It would accordingly be surprising if language disorders did not need to be related to syntax in some fundamental way; and it is surprising when one discovers how little attention has in fact been focused on this point. 1 The indispensability of syntax, is evident in the outline of most linguistic theories, which generally recognize three distinct aspects or levels in the study of language. These levels can be represented in the following way: language structure ____ I _____ pronunciation grammar2 meaning (phonetics and (morphology (semantics) phonology) and syntax) 1 The neglect has been particularly apparent in Great Britain, where the work initiated by Menyuk, Lee and others in the 1960s seems to have had no influence on clinical training and practice. See further ch. 2, and for a convenient collection of relevant papers, Longhurst 1974. • In this book the term grammar will be used to refer to all matters of structural organization exclusive of pronunciation and semantics. This is therefore a different use from that common in generative linguistics, where 'a grammar' of a language subsumes all these notions. Grammar for us comprises two subfields: morphology and syntax. The concept of the 'word' is at the centre of this distinction: morphology studies word structure (e.g. prefixes, suffixes, compounds, word endings, or accidence); syntax studies the way in which sequences of words constitute larger patterns-phrases, clauses, and in particular, sentences and sentence-sequences. Our focus in this book is on syntax, but we shall be referring to morphology at various places, and whenever we do not wish to focus on the notion of syntax as such, the term grammar will be used. The common phrase. '~rammar and syntax', often used in the literature on disorders, for us has no meaning, therefore, and we recommend it should be avoided. Terminological caution is crucial: for example, Myklebust (1965) uses a 'syntax quotient' and finds it to be an unsatisfactory measure-but in fact he is dealing more with morphological errors under this heading. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 6 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY There is currently a great deal of controversy about the relative importance of these levels, and how their interrelationships might best be seen. In particular, the relationship between syntax and semantics is undergoing much discussion (see, for example, McCawley 1968, Chomsky 1970). But despite this controversy, the usefulness of having the three above-named levels as focal points for theoretical and descriptive enquiry is generally recognized, and the central role of syntax, suggested by the model, is widely accepted. It is of course likely that, during the 1970s, arguments for the importance of semantics will be developed that will make the role of syntax seem less central, a more peripheral concern for the theoretical linguist. Such arguments tend to be of the following general kind: 'The com munication of meaning is the primary purpose of language, and a linguistic theory ought to reflect this priority in its analytical models. Semantic analysis is at the heart of the matter; and the choice of syntactic expression in order to communicate our meanings is of secondary significance.' This approach may well be correct, but in practice it proves impossible to take account of it, at the present time. No semantic theory has been worked out sufficiently for descriptive studies of any general validity to have taken place; and the way in which semantic analyses are related to syntactic patterns is still highly controversial, As a result, while the study of language disorders can benefit from the occasional insights of the semantic approaches currently being developed, we are of the opinion that there is no chance of a theoretical or descriptive framework capable of application in a therapeutic or remedial context being evolved in the foreseeable future. We have consequently chosen to concentrate our efforts on the application of syntactic studies, where a great deal of theoretical and descriptive agreement is apparent. Our view of the centrality of syntax may then seem conservative to some from the point of view of linguistic theory; but from the point of view of clinical application, we claim it is realistic and necessary. Our sense of paradox, then, results from the neglect of syntax at the expense of studies of pronunciation and vocabulary, especially the former. In speech therapy, the traditional focus of training in language has been in phonetics (the study of human soundmaking, in terms of articulation, acoustic transmission, and auditory reception) and, more recently, in phonology (the study of the sound systems of particular languages).3 Training in syntax-or in other areas of linguistics, for that matter-was not given, and it was in fact only in 1974 that a syllabus in linguistics, paying introductory attention to syntax, was introduced by the College of Speech Therapists in Great Britain. In remedial education, the tradi tional focus has been on vocabulary enrichment, and semantic considerations generally; and the same applies to primary education as a whole, where studies of the reading process, for example, have concentrated on the two sides of the diagram above, ignoring the centre. The classical debate in the teaching of reading, for instance, is between the respective merits of 'phonic' and 'look-and-say' approaches, which is exclusively a matter of the left-hand side of the diagram- 3 See O'Connor 1973 for a general account of these areas; for a discussion and application specifically to English, see Gimson 1980. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 7 specifically, the way in which the phonological system can be related to its visual analogue: speech writing pronunciation ----- ------ phonetics phonology t ~ ~ I ~ visual shape I 'graph>!tics' segments (vowel and consonant patterns) t I ~ spelling ~ prosody (intonation, rhythm, etc.) t I I ~ 'graphology' In a different connection, a great deal of time has been spent debating the question of vocabulary appropriateness, especially the selection of the proper vocabulary for children of specific socioeconomic backgrounds. But the whole field of grammar in the teaching of reading has, until very recently, been ignored. 4 We have also looked widely in the fields of mental subnormality and deaf education, to see whether the role of syntax in assessing the nature of a disability has been given proper emphasis and treatment; but again we have found little mention of it, and certainly only minimal attempts to think in terms of schemes of assessment and remediation based upon the findings of syntactic analysis. (The exceptions to this generalization will be reviewed in chapter 2.) In all this, we are not saying that syntax has been totally ignored; but we do feel that its systematic importance has been quite underestimated, and that peri pheral or superficial grammatical matters tend to be referred to at the expense of the more important underlying processes of sentence formation. The remainder of this chapter deals with four examples of what we consider to be a misplaced emphasis. (i) The focus on 'parts of speech'. The classical tradition of grammatical analysis handed down by the Greeks viewed language in terms of parts of speech, e.g. noun, l'erb, conjunction. In describing a language, the aim was to identify the part of speech a word belonged to, and then to present the various sequences permitted • Schemes such as Breakthrough to literacy and Link-up (see MacKay, Thompson and Schaub 1970, Reid and Low 1973 respectively) have begun to take grammar into systematic account; but from the perspective of the grammatical information outlined later in this book, there is still a great deal yet to be considered, Compare, however, the aims of Skylarks (Bevington and Crystal 1975). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 8 THE GRAMMATICAL ANALYSlS OF LANGUAGE DISABILITY in the formation of sentence patterns. This emphasis permeated the whole of the traditional grammatical study of English, and it is reflected throughout the field of language development and language disorders. Most of the works on grammar referred to by McCarthy (1954), in her major review of child language studies, deal with the range and frequency of parts of speech in children (e.g. Davis 1938, Ellsworth 1951), there is a similar emphasis in work on adult aphasia (e.g. Jones, Goodman and Wepman 1963), and one often finds in contemporary therapy the (often implicit) principle of structuring a series of sessions in terms of parts of speech, e.g. working on prepositions one week, pronouns the next, and so on. The point that has to be made strongly is that this orientation, by itself, is inadequate as an account of language ability or disability. Grammar is far more than an ordered collection of parts of speech, and parts of speech constitute a very minor aspect of grammatical analysis and description when compared with the various processes of sentence construction. The reason is simple: the very definition of parts of speech depends on these processes. For example, if the question is asked 'What is a noun 1', a generalized answer in grammatical terms5 presupposes our knowledge of sentence-patterns, e.g. in English 'A noun is a word which may occur as the subject of a verb, as in Dogs bark', or 'A noun is a word which may occur "governed by" a preposition, as in With luck . . :, or 'A noun is a word which may be preceded by an article, as in The boy . . .'. We build up our definition of a noun by examining all the grammatical contexts in which words like dog, boy, luck are used; and the same procedure applies to all the other parts of speech. In other words, to focus on parts of speech without paying proper attention to the language's syntactic patterns is to put the cart before the horse. And if one is not aware of this, it is easy to confuse grammatical study by asking artificial questions and setting up irrelevant problems. For example, 'What part of speech is round really preposition, adverb, noun ... l' The answer is: 'It depends on the context in which it occurs. If it occurs after an article, it is a noun; if it occurs before a noun, it is functioning as an adjective; and so on. There is no "basic" function to this word, such as is implied by the word "really".' Perhaps the most frustrated questions come from people who try to analyse the speech of very young children into 'parts', e.g. 'What part of speech is aI/gone, ta, peep-bo, gimme ... ?' The enquirer is doomed to permanent frustration if he persists in putting the question in this form, for two reasons. Firstly, it is premature to ask such questions of young children's utterances, months before they have begun to develop word-uses capable of description in adult terms. Even with 'clearer' cases, like want, shoe, these words are at best only distant approximations to adult usage (see further, p. 64). And secondly, it is fallacious to assume that all words are capable of being assigned unambiguously to some traditional part of speech. Words like hello, please, sorry seem to defy c1assification in traditional terms, and need to be classified under fresh headings, if they are to be classified at all. • Not in semantic terms, note. 'The noun is a name of a person, place, or thing' is a traditional answer, which tells us something about what a noun means (albeit rather vaguely), but gives us no information about what a noun does in the grammar of the language-where it goes in a sentence, whether it has case-endings etc. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 9 For such reasons, attempts to count the number of parts of speech in a speech sample, or to ascertain the ratio of nouns to verbs, or nouns to adjectives, are going to produce at best partial, and at worst positively misleading results, if the analyst has been at all casual about his method of allocating words to the traditional labels (this tends to happen particularly frequently over the use of the label 'adverb'). Without careful attention to the bases of definition, one can never be sure that two analysts are in fact using a label in the same way-for example, whether 'proper names' (such as London, John) are being seen as nouns or as a separate part of speech-and comparing their linguistic descriptions (of two patients, for example) becomes at once an uncertain and ambiguous undertaking, unless one manages to find out exactly how they are using their descriptive labels. But as soon as they do give their formal criteria (which is not commonly done, in the traditional literature), this places the focus of attention fairly and squarely on the field of syntax. 6 (ii) The focus on length. The length of a sentence or utterance is another measure often referred to in the literature on child development and disorders of language. It is a notion which, in various terminoLogies, has been in use for over half a century. MLR (mean length of response) is used by Nice (1925) and many others between 1930 and 1950 (see the review by McCarthy (1954», the number of words per response being averaged over a given number of sentences, usually 50 in all. More recently, Roger Brown and his colleagues have developed a concept of MLU (mean length of utterance), based on the number of morphemes per utterance (see the specific instructions for calculation in Brown 1973, 54ff.). The trouble with. measures of length is that they readily motivate the making of superficial judge ments (though. this is not a failing of Brown and his colleagues). It is true that sentence-length increases with age to some extent, but any single normative scale is very much an approximation, when one considers all the influencing factors. There is no neat linear development. especially between ages 3 and 5. Minifie et al. (1963) in fact argue that an index of length can be up to two years in error in either direction, as length is so readily affected by such variables as socioeconomic background, sex, IQ, and birth-order, as well as 'temporary' states such as health, emotion, the nature of the addressee, and so on. Cowan et al. (1967) showed how easy it is to establish bias in any length measure by varying the stimulus materials and the experimenter's role. Shriner (1967) finds MLR of little value oyer 5 because of increased response variability, and Brown (1973) does not use MLU after his stage V because by that time the child 'is able to make constructions of such great variety that what he happens to say and the MLU of a sample begin to depend more on the character of the interaction than on what the child knows, • For a further discussion of word-classification, see Robins 1971, 216ft'. and Crystal 1966. The need for criteria becomes more pressing as descriptive phrases become Jess specific and more impressionistic, e.g. a comment that a child's speech is getting 'less concrete' and 'more abstract' is intolerably vague without reference to formal criteria-and if concreteness reduces to (say) number of nouns and adjectives of a particular type, only reference to such criteria will permit consistency in making comparative assessments. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. ]0 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY and so the index loses its value as an indicator of grammatical knowledge' (54). Length by itself tells us very little. Two sentences may be exactly the same length, but internally be poles apart as regards their complexity. For example, most people would agree that the sentence The man and the woman saw the cat and the dog is less complex than The man that the woman was speaking to saw the dog, though both have the same number of words. The complexity would seem to be the important thing to try to pin down, and the notion of length therefore needs to be supplemented by a great deal more syntactic information. Apart from this, length is not as straightforward a thing to estabiish as may be thought. To begin with, there may be problems over identifying the unit which is being measured, e.g. if SENTENCES are being measured. does the analyst count all of the following as sentences? Yes. In the morning (said as response to the question When will you arrire?). So much/or that. And how many sentences are there in utterances such as the following (taken from a 3-year-old's monologue while playing with his toys): yes and it goes on here and it-and this one goes in the lorry and it goes up up up the hill and it goes up the hill-and it takes this all up the hill ... The difficulty is even more serious when it is UTTERANCES that are being measured, or RESPONSES, for it is far more difficult to define these notions than that of the sentence (and they rarely are defined in the papers we have read). This is one reason for caution, therefore. A further difficulty with length is that, even assuming that the units to be measured have been agreed, one still has to decide which units of measurement to use, and attempt to apply this measuring-rod consistently. There are many possible contenders for units of measurement-words, morphemes, intonation-uni ts, syllables, stressed syllables, phonemes-and results will vary depending on which unit you choose. The boys are running quickly is longer than The boy isn't running now in terms of morphemes (the - boy - s - are - run - ing - quick-Iy; the - boy - is - n't - run - ing - now), shorter in terms of the number of stressed syllables, and the same in terms of words. Moreover, each unit poses its own problems of applicability, e.g. in counting words. does one take contracted forms (it's, isn't etc.) as one word or two? Are idioms to be counted as separate words (e.g. is spick and span three words, even though the usage is fixed in this form)? There is also a whole history of problems in trying to make the notion of the morpheme work consistently in English. A morpheme is the smallest meaningful unit of grammatical form, and broadly corresponds to the notions of root and affix in traditional grammar. Most English words present little problem, when analysing them into morphemes: it is obvious that the word blackbird has two constituent morphemes, black- and -bird, and that boys has two morphemes boy- and -s (signalling the plural 'meaning'). But what does one do with raspberry (the constituent -berry is clearly identifiable, but what about rasp- ?), mice (where the root mouse and the plural have somehow been fused), sheep (where the singular and plural forms are identical), and the many others? There are various possible answers to these questions, and in the 19405 the linguistics literature was full of debate about the alternatives. 7 Given such complexity, it is extremely difficult to ? See Joos 1957, or, for a recent review of the problems, Matthews 1974. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX II work consistently in establishing MLU in terms of morphemes, and results have to be viewed with caution.s Length by itself is an inadequate indicator of grammatical ability, as Rees concludes (1971, 291): 'this procedure tends to obscure the nature of the child's grammatical skill as well as his efficiency in using language for expression and communication.' It has some methodological value, as a way of imposing a preliminary developmental ordering on samples of data, before carrying out a grammatical analysis, and this is the way in which it is used by Brown and his colleagues (see Brown 1973, Morehead and Ingram 1973, and other references there): 'two children matched for ML U are much more likely to have speech that is, on internal grounds, at the same level of constructional complexity than are two children of the same chronological age' (Brown 1973, 55). [n this way, of course, it might be used as a screening test in a clinical context. Nor is there any reason why it should not be used along with a measure of complexity to produce a combined evaluation, e.g. Shriner's (1967) LeI (length-complexity index), which is a combination of MLR and various indices of structural complexity (after Templin 1957). But on its own, the data any length measure provides can be highly misleading. All there is in its favour is ease of applicability-a factor which is of no small value for the clinician or teacher pressed for time-and this has often led to its being used instead of sentence complexity (e.g. by Renfrew (n.d.). But the question of optimum use of time makes sense only in the context of the long term, as we shall argue below (p. 24), and it is clear that in the long term, isolated measures of length can lead to faulty assessment and inappropriate choice of remediation techniques. To take two obvious problems, length will not differentiate between echolalic and novel speech; nor will it say anything about someone's ability to interact with his interlocutor in dialogue. Nor can length measures, by themselves, generate suggestions about specific therapeutic procedures. Getting the pupil or patient's (P)9 sentences longer is a desirable aim, but the critical question is, How? (iii) Linguistic realism. It is essential to develop an accurate conception of the contemporary language, in any discussion of syntactic norms. Any procedure must allow direct comparison of the abnormal utterance with adult norms or the norms of children comparable in other respects, and T9 must therefore be well aware of the nature of normal utterance, and use procedures capable of being applied with equal cogency to both normal and remedial speech situations. Most of the available syntactic assessment procedures seem to lack a normative dimension, in fact something bemoaned by Longhurst and Schrandt in their review (1973, 248) and we have ourselves often found Ts unsure of how a normal P would react when placed in a test situation of the kind T regularly uses. Perhaps the most , The concept of MLR is reviewed by Shriner (1969), MLU by Crystal (1974b). Various measures are compared by Sharf (1972). Ingram (l972c) uses a combined length measure based on both words and morphemes. 9 For convenience, we conflate the two main categories of remedial person under one heading. We shall do the same for teacher and therapist-To The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 12 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY far-reaching misconception, in this respect, is for T to think of syntax purely or predominantly in terms of the written language, with speech being felt to have 'little' or 'no' or 'debased' grammar. This has been a particular hindrance in work with the hearing-impaired, where the problem of oral unintelligibility has led to a playing-down of the importance of syntax or judging its correctness solely in terms of written norms (see, for example, Heider and Heider 1940, Cooper 1967. and the review of the field by Pressnell 1973). But worries about allowing colloquial speech too major a role are widespread, in all areas of disability, though they sometimes appear in print in a rather indirect way, e.g. Muma 1971, 436, 'inasmuch as the data were obtained from oral samples, the contraction transformation is undoubtedly spuriously high' (our italics), which we read as a complaint about the frequency of such items as don't, can't, isn't etc.; or Lee and Canter (1971, 326), who say, in relation to wanna, gonna, gotta etc., that the child 'should not be penalized for this articulatory error.' These attitudes, whether consciously or unconsciously held, are not restricted to the field of language abnormality, of course. The entire Western grammatical tradition has only in this century begun to free itself from the rigid view that the written language enshrines the laws of correct grammatical expression, and that the spoken language should be made to conform to this (see Crystal 1971a for further discussion). But even these days there is still a reluctance to recognize the very considerable differences between the spoken and written forms of language, and syntactic remediation programmes, if they are not aware of these differences, inevitably suffer-sentence-structures become artificial, and misleading strategies are advocated. An example of artificiality in sentence structure may be found in the tendency to make sentences longer by symmetrical expansion of the elements of structure, e.g. Noun + Verb (NY) becomes NNV, then NNVV, then NNNVV, then NNNVVV, and so on, which rapidly produces absurd structures. Conn (1971) gets into difficulties over this, for instance. One of his sentences develops into Jack and Jill are washing Jim and Jane, to which a corresponding question stimulus is proposed: Whom and whom are Jim and Jane washing? An example of a mistake in strategy would be the attempt to eliminate the use of and as a means of linking sentences. This is something which is widely accepted as an appropriate stylistic correction in the written language of children, and on the basis of this it has been said to be an appropriate correction for speech too, and an improper feature to reinforce in therapy. Lee and Canter, for example, say (1971, 329): 'Since there is no grammatical constraint on the endless use of conjunctions, special rules had to be created to avoid deceptively long, high-scoring sentences.' Only one and per sentence at clause level is accordingly allowed, and 'This treatment may be given to any other overused conjunctions' (our italics). But at the appropriate linguistic age (around 3 years, see p. 76, reinforcement of and, and other sentence-linking features of this kind, we find to be one of the most important strategies to advocate (see also Hutchison 1972). It is important, too, to be familiar with the extent of 'normal nonfluency' in colloquial speech (see Crystal and Davy 1969, 104), and the range of variations in colloquial usage which affect the definition of syntactic rules. Deciding on the acceptability of a sentence is not a hard-and-fast matter (see Quirk and Svartvik The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 13 1966). In this respect, therefore, we find the 'all~or-none' scoring technique-as used by Lee and Canter (1971), for example-too inflexible, as it does not allow for the varying degrees of grammatical completeness which we find to be a regular feature of the speech of children at certain ages. Lee and Canter say, for instance: 'A structure is not given a score unless all the required syntactic and morphological rules have been observed. No intermediate steps are credited' (317). For example, pronouns need all of the features of person, number, gender and case before a score is allowed-mine car, for instance, would score O. They do allow some colloquial reductions, but only if they conform to the structural patterns assumed as a general model; for example, usages like I can 'qualify as sentences since they contain both subject and predicate. While the verbs are incomplete, they are not exactly incorrect in a conversational, spontaneous speech sample' (325). But the majority of the elliptical patterns of spontaneous speech are not allowed, and there is accordingly a marked bias against colloquial speech in the scoring. 10 Intonational questions-to take a further example-are 'scored as incorrect questions'. We therefore doubt the conclusion that their technique 'allows a clinician to estimate the child's ability to formulate and produce grammatically "loaded" sentences in the kind of conversational setting which he encounters daily with his parents, his teachers, and his peers' (337, our itaJics).l1 In short, any syntactic procedure must be firmly grounded in the facts of both written and spoken usage, and of the various styles (formal, informal etc.) and dialects subsumed under these two headings. But above all, the approach must pay proper regard to the facts of colloquial speech~the most frequently used variety of adult language, and the one children hear most consistently during their linguistically formative years. This principle is worth emphasizing as it is frequently transgressed. Bizarre syntax is common in both test materials and reading schemes. In the Northwestern, for example, there is the distinction between The man brings 10 A reluctance to accept elliptical patterns is quite common in traditional grammar, and turns up frequently in applied linguistic studies. Another example is Sharf (1972, 66), who excludes from his samples any specific replies to questions: 'fill-in answers' 'were excluded on the basis that they were not expressive of spontaneous grammatical construction.' But knowledge of what to omit is an important aspect of linguistic creativity (see further below). 11 On more general grounds, we see important limitations in approaches (such as in the Northwestern Syntax Screening Test) which permit tbe occurrence of error always to outrank the presence of positive structural ability. The instructions to that test say: 'Any change of the examiner's spoken sentence which affects the test item is considered a failure, even though the child's response is grammatically and semantically correct' (Lee 1969, 5), and 'Any response which contains a grammatical error, even though it is not the test item, is considered a failure on the grounds that the test item, though correct, may have introduced enough complexity to cause other structures to be dropped' (6). But how can this view conceivably be validated? On what linguistic grounds should the correct or incorrect use of -$ in the verb, for example, affect the use of a determiner in the subject noun-phrase? Lee 1969 is only a screening test, not a general assessment, but a screening test which does not bring out areas of strength as well as areas of weakness is likely to produce a distorted picture. Contrast Dever (l972a, 1973), where phrase level error 'cannot affect clause level classification' (Dever and Bauman 1971, 25) and vice versa.. (We would not go as far as Dever and Bauman, however, in saying that 'when a child gives an incorrect response we know nothing' (6).) The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 14 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY the boy the girl and The man brings the girl the boy. There too most of the sentences make use of the present tense (e.g. The boy jumps), which is a relatively uncommon usage. Overuse of this tense-form is a dominant feature of reading schemes also. Some examples of odd syntax from such schemes: What have you in the shop? (an archaic usage, where these days we would say What have you got etc.), One kitten runs to the basket (strange use of an emphatic determiner in an unemphatic context, and a 'sports commentary' present tense again), and One little, two little, three little kittens. Some of these sentences would be acceptable with appropriate intonation, of course (e.g. nursery-rhyme intonation for the last example), but we have not noticed such intonations to be a regular feature of T's or P's performance in the teaching situation. (iv) Selective commentary. We suggested above that reference to syntax has not been adequately systematic in the literature on disorders. What we note now is a readiness to list information about syntactic features, without relating this inventory to some general framework. Almost any case study or longitudinal comparison of Ps will demonstrate this. The following extract is taken at random from a recent Journal (it is in fact a study which is well above average in its syntactic awareness, and where the author makes it quite clear that there is a great deal left unsaid): The same patient, D.Ed, retained his severe agrammatism, to use the traditional term, long after his active lexicon had multiplied many times. His early difficulty in using verbal forms-or anything other than substantives-has been mentioned. This was fairly resistant to efforts on the therapist's part to prevent the development of telegraphic speech: language habilitation centered on helping the patient to produce various simple syntactic patterns such as imperative forms ... descriptions like 'It's heavy' ... in answer to questions of the type of 'What's the hammer like?' ... He could be helped to retrieve or acquire adjectives fairly soon after the nouns. but there long remained difficulty in combining the two .... The inclusion of a preposition to effect a linguistic relationship was difficult .... He frequently omitted articles, and, which interfered more with communication, the auxiliary 'is', or 'was' in verbs.. . (Hatfield 1972, 76) This quotation does not do justice to the sensitivity, balance and illustrative detail of this paper, but it does show the limitations of its method. This kind of selective commentary is easy enough to do, if one has had any kind of exposure to gram matical tradition, and it can spark off ideas. But it is of dubious value as an instruc tional or remedial method, because it fails to relate its observations to some general perspective, without which any linguistic profile is inevitably arbitrary. Comparison with other Ps is difficult because of the idiosyncratic selection and ordering of the data (the movement from verbs to adjectives to prepositions to articles etc. in the above quotation is unlikely to have been motivated by any theoretical con siderations). Assessment on the basis of such an inventory is unreliable, because there is no guarantee that one has spotted the most important syntactic processes, and there would be an inevitable tendency to pick on those features of syntax which are the most readily describable, e.g. pronouns, omission of grammatical words. such as lite, of Nor is remediation assisted. as here one needs to be systema- The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 15 tically aware of what P has not got in syntax, before choosing an area on which to work; but given an inventory of what is present in a sample of P's speech-even if this is an exhaustive account of the sample-it is left up to the analyst (or reader interested in using the results) to work out for himself what is not, and this is an extremely difficult and irritating task. The same criticism applies to syntactic assessment procedures which use only a small selection of features of syntax: the question has to be asked, Why these and not others? Has the selection been made on a principled basis, or is it impressionistic? For example, when Reynell (1969) isolates pronouns, prepositions, conjunctions, and questions (other than intona tional) as four categories to be scored (item lIon the Expressive Developmental Language Scale), one wants to know on what grounds this particular selection of features was made (see further, p. 18). Or again, is the concentration on morpho logical data and 'function words' in the well-known tests, and the exclusion of more general tests of syntactic ability, a decision of principle on the part of the authors? (See, for example, the Grammatic Closure Sub-Test of the Illinois Test of Psycholinguistic Abilities (Kirk and Kirk 1971, Kirk, McCarthy and Kirk 1968), or Lerea 1958 with reference to the Michigan Picture Language Inventory.) Such questions abound when one considers the literature on language disorders. They can be answered .only by developing what, for lack of a better tenn, we can call a proper degree of 'syntax-awareness', and this means relating any commentary to some descriptive framework which deals comprehensively and systematically with the syntactic features of the language-in other words, a 'grammar'. Only then will generalizations about someone's competence or incompetence in syntax be able to be assessed, compared, and implemented. We therefore isolate as two central goals for any work on remedial syntax that it be (a) in principle, compre hensive, and (b) systematic. And of the two, it is the need to be aware of system which is crucial. One could produce a comprehensive inventory of the syntactic features of a language; but an inventory provides no explanation, or sense of underJying pattern. To have a list of features. in which item 13 was the definite article, and item 97 the indefinite article would be of little value. At some point in any grammar of English, these two items have to be brought together, and their complementarity of function investigated. And so it is for all aspects of grammar. Grammar is not a random collection of features, nor is it learned in this way (see chapter 4). It is a highly complex system of structures and meanings, and doing a syntactic analysis means establishing how the various patterns in speech can be located within this overall system. It is the task of the grammarian to define all the variables which operate within this system, and to suggest an optimum route by means of which all aspects of the grammar can be displayed and interrelated. There are many possible methods of grammatical analysis available (our own being presented in chapter 3)./ But it should be clear that without some attempt at a comprehensive and systematic procedure, the dangers of an uncontrolled impres sionism come to the fore, and the possibility of carrying out objective and. consistent comparative work considerably recedes. It is at this point that the role of the linguist becomes clear. Establishing a grammatical analysis that is unambiguous and comprehensive is a specialist, time- The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 16 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY and space-consuming task, and the value of much recent published work on syntactic disorders is diminished by authors not being fully aware of this. It is simply not possible to do a clinical investigation of a group of Ps and present a grammatical framework in the course of a few pages, but this is what is regularly attempted in the journals. The likelihood is that the grammatical model will be explained in an approximate or oversimplified way, with only a small proportion of structures and descriptive labels being selected and properly illustrated, and apparently arbitrary distinctions being made, in the absence of any discussion of criteria. For example, Engler et al. (1973), in an analysis that is far more detailed and discriminating than most (and which, incidentally, seems to parallel ours in many ways, cf. below), presents a skeleton analysis of syntactic structures that is of very limited use. There is one main reason for this. It has no mention of criteria,12 and thus the reader is left unclear as to the grounds on which a distinction has been made, and how many other cases of a similar kind there might be. At one point, for example, they say that in Subject-Verb-Object constructions, the intensive verb is subclassified into three types: 'complement taking' (e.g. look, seem), 'senses intransitive' (e.g. taste), and 'middle' (e.g. weigh). The reader who wishes to use their system of analysis, however, is left quite in the dark. How many other verbs like weigh are there? How do you decide whether a verb is 'middle' or not? Nor are Engler et 01. atypical in their inexplicitness. Longhurst and Schrandt, for example, have a paper comparing the merits of four assessment procedures in respect of their ease of applicability (how much technical linguistics was needed, and whether the instructions were clear), inter-scorer reliability, ability to discriminate language differences between children, and ability to describe these differences. They conclude (1973, 248): 'None of the four linguistic procedures investigated [sc Lee and Canter 1971, Engler et al. 1973, Dever and Bauman 1971, Lee 1966] proved to be completely explicit and sufficient for analysing the speech of children', and the same applies to other papers that we have read in this area (e.g. Braun and Klassen 1971). Approaches which explicitly refer to some grammatical theory are strongest in these respects, e.g. those which base themselves on a particular model of transformational grammar (such as Menyuk (1963) in her use of Chomsky (l957a); or Morehead and Ingram's (1973) use and modifica tion of Rosenbaum's (1967) adaptation of Chomsky (1965); or Dever (I 972a, 1973), who uses tagmemic analysis). But even these approaches are extremely selective in their illustration of grammatical features and processes,13 so that it proves difficult or impossible for T to 'follow the model' and use it for himself. T is continually being left to make his own decisions about questions of analysis- U Apart from a footnote stating that there is a problem over deciding whether a construction is a passive or an adjectival complement. But they do not say what the problem is, or how it is to be resolved. 13 Morehead and Ingram (1973) find that their grammars accounted for all but 8-10 per cent of the sampled utterances-but their sample was relatively small. as is suggested by some of their results for the normal sample, e.g. (in their terms) all, vocative and stalive verb particle-shift (e.g. up ball) Iransformations are never used after their stage I, prepositions after stage II, Verb qualifiers (e.g. I just go) after stage I until stage V; object noun retention (e.g. hil it ball) is used only in stage II, and inchoative (e.g. it got red) only in stage IH, The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 17 whether an item should be analysed as X or Y -particularly as regards the more restricted grammatical patterns. And for those approaches which do not relate to any accepted linguistic theory, the problem for the reader is far worse. It should be noted that this criticism is a very specific one. It is not a criticism of the various methods when used by their respective authors; but it is certainly a criticism of the usefulness of the approaches, for unless such work is sufficiently explicit to be generally usable, one wonders why it was published in the first place. There is, however, only one way to avoid this criticism (apart from turning one's papers into a book-as in fact has been done by Dever (I972a) and Lee (1974)). It is impossible to present the entirety of a syntactic system in any meaningful way in a few pages unless one can relate one's outline to some established grammatical canon. In this way the clinician is relieved of the responsibility to have to explain all his descriptive apparatus, or deal with problem areas, as these will be given full discussion at the appropriate point in the grammatical tradition on which he is relying. The problem for the clinician, of course, is then to make a good choice of a grammar, and we shall have more to say on this in chapter 2. But this aside, we have been surprised to find that workers in the field of syntactic disorders have rarely tried to relate their work to conventional descriptive grammars of English. The problem of explanatory clarity thus remains. A related criticism would fall under the heading of representativeness. It may be impossible for a syntactic procedure intended for remediation use to incorporate information of a maximally detailed kind, but at least all the main areas of grammar should be given some recognition, and a balance kept regarding the depth of detail of the description presented. Procedures which go into great detail about, say, morphology, but tell us little or nothing about phrase structure would be clear cases for criticism. One of the criticisms we have of Lee's method of Developmental Sentence Scoring (1966, 1969)14 falls under this heading-though here the strength of the approach is in fact in its information about phrase structure. She selects eight syntactic features to score: (i) indefinite pronouns or noun modifiers (e.g. this, some, each), (ii) personal pronouns (e.g. me, your, myself, who), (iii) main verbs (e.g. see, is, may, have), (iv) secondary verbs (e.g. going to, I like swimming, I want to go), (v) negatives (e.g. not, isn't), (vi) conjunctions (e.g. and, because, than), (vii) interrogative reversals (e.g. is it, could he ... ), (viii) H1h-questions (e.g. what, how . .. ). Now while we accept that, for her purposes, to score every feature in the grammar would be 'clinically impractical' (1971, 319), we nonetheless find her particular selection to be unrepresentative of the syntactic system as a whole, and think that she has been unnecessarily dismissive of other areas of syntax which we hold to be of critical significance in assessing the character of a language disorder-in particular, the order of elements in clause and noun-phrase structure. Lee lists a number of the features not scored separately: 'the use of articles, plurals, possessive markers, prepositional phrases, adverbs, word order, word selection .. .' and says, 'To account, at least in part, for these unscored items, an additional 14 See also Lee and Canter 1971. The approach is reviewed by Bloom (1967), Longhurst and Schrandt (1973), and Tyack (l972b). See pp. J 96-7 below for some comments on Lee's most recent publication (1974). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 18 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY sentence point is added to the total sentence score if the entire sentence is correct in all respects' (Lee and Canter 1971,320). We find such an allocation to a child who can command all such features miserly in the extreme p5 Putting this another way, we find it unsatisfactory that sentences such as Boy eat or Boy eat cookie would each score 0 using her procedure-in our view a complete reversal of linguistic priorities. (It may be contrasted, for example, with the separate attention given to the various 'processes of arrangement' in Engler et al. 1973.) Another example of selectivity is Reynell's Expressive Language ScaL-. This is largely con cerned with distinctions in length and quantity, but a few structural features are incorporated. These however are categorized too generally (e.g. 'word combinations', 'correct use of words in sentences', 'use of subordinate clauses'), and in the one case where specific structures are cited, we cannot understand the basis of selection, from the point of view of developmental progress (see chapter 4). One point is scored for correct use of any three of pronouns, prepositions, conjunctions, and questions other than intonational. But prepositions emerge as early as Stage II in the classification in chapter 4; pronouns develop over a period from Stage III to Stage VI; formal questions are concentrated in Stages II-IV, and conjunctions are the dominant constructional process at Stage V. To score I for any 3 seems a strange juxtaposition of values. Lastly, concerning the problems of selective commentary, it is important to ensure that any information obtained through the use of the syntactic procedure is retained in the final presentation of the results, so that a balanced and wide ranging picture may be had of P's ability in the syntactic system as a whole. Bearing in mind the number of major variables which have to be taken into account, we therefore take the view that any attempt to work out a single syntactic 'score' is going to be of negligible diagnostic or therapeutic value, in respect of the above desiderata. Syntax scores have often been tried, e.g. Lee 1966, Brannon and Murry 1966, Reynell 1969; but we have found them unilluminating, and at times counter intuitive. Sometimes, even, we have the impression that the internal logic of the scoring system takes over the linguistic analysis, and forces linguistically question able rules or classifications, e.g. Lee and Canter (1971, 325) find themselves in danger of overscoring words like 'it was broken', so they assume that these are learned as adjectives: 'a rule was made that if the past participle verb form could be used as an adjective in a noun phrase, the sentence would be scored as copula + adjective, not as a passive'.16 Scores will become more useful only when ,. For these reasons we are not surprised to see Sharf's conclusion (1972). He compared four measures of language development (MLR, NDW (number of different words), LCI (length complexity index) and DSS (developmental sentence scoring) and finds a close correlation between the first two and the last two. He concludes that for relating language growth to chronological age, 'it appears that quantification of structural analysis of language provides no more sensitive measure than quantification of verbal output' (73)-though he accepts that the former are of no remedial use. But this may be no more than a comment on the limitations of the structural analyses cited. 18 Cf. above p. 12 for the similar difficulty over coordination. In a different, but related connection, we note that the listing of transformations is sometimes carried out regardless of their diagnostic relevance: e.g. Morehead and Ingram give a number of categories which seem to have little or no role in their subsequent analysis, some of which are listed in fn. 13 above. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE STUDY OF SYNTAX 19 the influence of extralinguistic factors (such as class and sex) on syntax comes to be better understood, and the concept of a 'matched' sample made more plausible. For the time being, therefore, we prefer to avoid the notion of a score, and think instead in terms of a syntactic profile (see below), in which a wide range of syntactic variables are simultaneously presented, and a qualitative assessment made of the areas of strength and weakness. The integrated set of values present in the profile could of course be given a statistical analysis, and this will be an important feature of any standardization of this procedure; but for routine remedial work, we have found this to be unnecessary. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 2 Some critical factors in devising a syntactic remediation procedure A consideration of the weaknesses in the traditional study of syntax in language disorders naturally suggests the areas which any positive contribution to this field would need to concentrate on. But before going into the specific characteristics of a syntactic procedure, it is important to consider in general terms what the purpose of the exercise is. What exactly do we expect such a procedure to be able to do? We therefore begin with a statement of aims. Aims Our aim is to develop a procedure for analysing the syntactic character of language disorders, capable of being used routinely by anyone involved with the diagnosis, assessment and remediation of language disability. Such a procedure would fulfil a variety of functions, of which we see the following as being the most central. (i) It must be capable of providing a framework for use as a screening procedure, i.e. as a technique for determining whether or not there is a case for more systematic linguistic examination. (ii) It must be able to provide a comprehensive description of P's syntactic output at any stage during the processes of assessment and diagnosis. (iii) It must be able to provide a principled therapeutic methodology. Points (ii) and (iii) are fundamental, and need further discussion. Concerning (to, implicit at any stage is a comparison between the syntactic patterns P has and those he should have. T must work within an integrated developmental perspective-by which we mean, in practical terms, that he must know where he is at any given point in treatment. There are two sides to this, though: (a) knowing what structures have been acquired (and with what degree of mastery), and (b) knowing what structures have not been acquired but which would be appropriate for the stage of development that P seems to be at. This second point is perhaps the most underestimated in clinical language programmes, and unless one is aware of its implications, problems rapidly accumulate without being noticed. T takes a structure, and after training, P makes progress in its use. As a result, T develops the use of this structure, building on it and extending it in various ways, and there is further progress. But caution is necessary. This is only one structure among a number that P may need to develop. By concentrating on one to the exclusion of the others, T runs the risk of developing an unbalanced The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC REMEDIATION PROCEDURE 21 linguistic skill, and of overestimating P's syntactic ability. T has to make himself aware of the entire range of structures that ought to be developing at any given stage, so that he knows clearly what he has not taught as well as what he has. It is sometimes difficult for T to stop working with a structure that has been successfully introduced to P, and to turn to something else, but such decisions are crucial if a balanced syntactic skill is to emerge. l There are other aspects of the need for T to have available a comprehensive descriptive picture of P, at any given stage. One is to be able to evaluate the unexpected structure, when it arises. A typical situation, we find, is for T to be concentrating on eliciting a particular structure, and P comes out with a different structure. The problem for T is to decide what to do with this: should it be ignored, reinforced, modified? Such a decision can only be made, however, if T is able to see how this structure relates to the others that have been used or taught and those yet to be introduced. Any syntactic procedure, to be of use, must allow this kind of ready cross-checking. A second aspect of this situation is to enable a systematic revision or recapitulation to take place between remedial sessions, or whenever a period of time elapses between assessments. A checklist of syntactic patterns, properly organized, we find is especially important when we are faced with apparent regression in behaviour. With dysphasic adults, for example, variations in clinical and mental condition are so frequent that it is essential to be able to plot with some precision the various rises and falls in syntactic ability, otherwise therapy rapidly becomes random and impressionistic. Concerning point (iii), the provision of a principled therapeutic methodology, we see this as the corol1ary of our second aim. A grammatical analysis of a dis ordered individual, by itself, is of restricted value.2 What T needs to know is how to proceed, once the character of the syntactic disability has been detennined. Any system of syntactic analysis which is to be of remedial value must be organized in such a way that some grading of structures is apparent. When should a particular structure be introduced into a remediation programme? When should it be phased out? What alternative structures can be used, if P fails to respond to one's first choice? What is the optimum route through the syntactic system (or, if there is no optimum route, what are the routes most likely to be successful)? Such questions are central, and in providing answers, it is important that the principles underlying any procedure should be made quite explicit. Unless T can answer the question 'Why?' in all of this, there is a real danger of remediation degenerating into a mechanical routine. Aims (ii) and (iii) between them constitute adernand for an integrated perspective 1 In one case, just such an imbalance had been created due to therapy having concentrated on the use of specific parts of speech and phrases. When Peter (aged 41) was first referred to us, he had a good range of noun-phrases, verb phrases, prepositional phrases etc., but next to no aware ness of the main elements of sentence structure, so that his phrases came out with little relationship to each other, and a resulting ambiiUity and unintelligibility. Subsequent therapy had to redress the balance, by concentrating on sentence structure. ' a It is this point which we find constitutes the main difficulty with many of the syntactic analyses intended for remediation, e.g. Myerson and Goodglass 1972. To write 'a grammar' of a patient is only the first stage. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 22 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY in studying syntactic disability, without which we do not feel T can proceed with confidence. It is this, the establishment of confidence in remedial decision-making, with which we are ultimately concerned, and as one prerequisite for confidence is knowing the limitations, as well as the strengths, of a procedure, we should make it clear from the beginning of this book that there are certain clinical decisions involving linguistic reasoning to which we find it impossible to make any contribu tion, in the present state of knowledge. In particular, we are not ready to make any systematic contribution to questions of diagnosis, and this is why we do not cite diagnosis as one of our primary aims. It is unfortunately the case that insufficient Ps have been analysed in terms of syntax to enable any general categorization to be made, other than in the broadest terms (see chapter 6). We share in the widely voiced criticism of some of the general labels that have been used to characterize linguistic disturbance (e.g. 'agrammatism') that they are too inspecific to be of practical use, or leave unclear whether the deficit is to be identified primarily in linguistic or non linguistic (e.g. cognitive) terms; and we do not propose to add to this confusion by constructing a further set of vaguely defined and empirically unsupported categories for syntax. Once the appropriate amount of empirical spadework has been carried out, and a normative framework established, we hope to hazard a classification of syntactic disturbances which may be of diagnostic value. But for the present, our entire orientation is towards the individual, and away from the group. We wish to establish progress in P's syntax, and to control it. Only after this has been done in many cases can we compare individual strategies and responses, and suggest typologies. In this we find our selves at one with Rosenthal et al. (1972) who, after evaluating seven major diagnostic categories, and finding them all to be heterogeneous and ambiguous, also conclude; 'assessment ought to be aimed at detennining individual treatment needs, and not toward assignment of the child to a specific diagnostic category' (135). The same applies to work with adults. Desiderata We have so far talked in very general terms about syntax and its application in remedial work, but we have also at times referred to specific procedures, by way of illustration. We now propose to look more thoroughly at what we feel to be the basic factors that need to be borne in mind in constructing any syntactic remedial procedure. It is clear that developing such a procedure is going to be no easy matter, when one looks at the range of considerations, theoretical and practical, linguistic and nonlinguistic, that affect the issue, and we have not been surprised to find that none of the available approaches satisfactorily meet the criteria that we consider central. We have aU of us learned a great deal from the work cited in this chapter and in chapter I, most of which was developed in the United States in the mid-I 960s; but we have been unable or unwilling to use this in our own work, and we must explain why, as a preliminary to outlining our own approach later in the book. We propose to do this by isolating certain general principles. to which we think any syntactic procedure should conform. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC RE~EDIATION PROCEDURE 23 (i) General applicability. The model developed must be capable of being applied (a) to both normal and disordered individuals, (b) to problems both of assessment and remediation, (c) to both adult and child Ps, in all the domains of remedial work,3 and (d) to the analysis of both comprehension and production. We shall comment briefly on each of these points. (a) The model, as we have already suggested, must be comprehensive, to allow for routine comparison with normal development, at whatever stage, But there is a second reason why any model cannot set its sights lower than a full grammar of the adult language, and that is because very often the focus of T's attention must be on his own syntax. This is particularly so when the disorder is severe, when little expressive language is being produced, or where there is a comprehension problem, In such cases the syntactic procedure swings round to operate on T's use of language. And generally in remediation, awareness of one's own syntactic habits is an essential feature of any evaluation of progress, (b) One of the main points to emerge from the discussion so far is the need to maintain a proper balance in any clinical procedure between assessment and remediation, This means, in linguistic terms, that the procedure has to be evaluated according to the attention it pays to its system of syntactic description and to its system of grading structures; and we shall discuss both these points in detail below, The main difficulty we have had with the approaches we have studied, however, is that on the whole they do not maintain this balance: they are either systems of assessment and diagnosis, with implications for remediation, or they are remedial programmes, lacking perspective in assessment. The most successful attempts to preserve a balance, in our opinion, are those by Lee, Dever, Ingram, and their associates, and we shall frequently be referring to their work, by way of com parison with our own, The general point we would make is that it is not possible to integrate the elements of assessment and remediation without relating one's procedure to some specific grammatical approach and some specific scale of development. (c) The procedure should not insist on a rigid distinction between adult and child syntax. The essential continuity of development from child to adult argues against it, and any such division would give rise to all kinds of practical difficulties, e.g, should one use the adult or the child procedure in analysing the language of adolescents in special education? At a more detailed level of argument, we have noted some approaches where the procedures are inapplicable (without modifica tion) to certain clinical populations. Dever (I972a), for example, takes for granted the use of well-defined intonational units at the beginning of his syntactic remedia tion procedure TALK (Teaching the American Language to Kids), and his system thus cannot be readily used for children whose initial difficulties are largely 3 Here we are referring to the traditional interdisciplinary barriers, which have been very marked in remedial work. The subject of language disability is of interest to many groups'""":. clinicians, mother-tongue teachers, and foreign-language teachers, in particular. The remarkable thing is that each of these has its own journals, conferences etc., which members of the other groups rarely encounter, The points of contact between these fields need to be explored, materials shared. and duplication of work avoided. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 24 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY prosodic (as with certain ESN and 'non-communicating' children, for example) and for many adult dysphasics. (d) The procedure should be capable of being used in the study of disorders of both production and comprehension. Of the two, we feel that the main gap is in the field of production, and it is there that we see a syntactic procedure as being most beneficial. We have accordingly biased our presentation in that direction. But it should be clear that in principle the procedure could be used as a means of structuring and grading sentence patterns for comprehension work, or in trying to establish what range of patterns promoted comprehension difficulties, and we have noted significant similarities between our own approach and some of the work on comprehension, e.g. Mittler et al. 1974. (We do not however assume that production and comprehension develop in parallel: see below, and also Clark et al. 1974.) (ii) Clinical realism. If it is to be convincing, a syntactic procedure must be able to identify the character of a disorder fairly directly and systematically, and should not regularly conflict with the intuition of the experienced T concerning the degree of delay or severity that may be present. It must also be able to cope with the description of the range of abnormal linguistic patterns used by P. At a more everyday level, it must be capable of being learned by T with a minimum of linguistic training, and capable of being used bearing in mind the pressures of the clinical or teaching situations. This last point needs to be amplified, as there is inevitably a tension between accuracy and comprehensiveness on the one hand and speed and simplicity of use on the other. This applies to a syntactic analytic procedure as much as to any other, and it is up to T to judge how best to arrive at the optim..1m balance between time spent on matters of analysis and interpretation, and time spent in actual therapy or teaching. The hard fact of the matter is that if one wants to achieve a complete and accurate understanding of a syntactic disability, there is no alternative but to spend analytic time on it-perhaps 3 or 4 hours, in order to obtain a reason ably full analysis of a half-hour sample (see chapter 5). Many Ts obviously find such an outlay of time impossible, as a routine enterprise, and, unfortunately, opt for techniques which may be less helpful but are more practicable. The sad story of the Bus Story is a case in point. Renfrew's original method was found to be unreliable, and so; 'Eventually it was abandoned in favour of a scheme based on transformational grammar and devised by a psycholinguist. This method, though in many ways more satisfactory. proved time consuming and required considerable practice to achieve consistent results. The busytlinician, for whom the scale was being devised, rejected it on account of its complexity' (Renfrew n.d., 14-15), and went instead for a measure of sentence length. Not all methods of syntactic analysis pose such problems of complexity, consistency and time as transforma tional ones, of course, but the principle at stake here is a general one: simple analyses of complex phenomena only postpone problems; they do not solve them, and they may obscure them. However, with the best will in the world, it is often simply impossible for T to carry out even a reduced analysis of even his most The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC REMEDIATION PROCEDURE 25 interesting and problematic cases, because of pressure of time. In such situations, the only real solution is long-term-an improvement in the T: P ratio, and a more realistic job-analysis which pays due attention to the complexity of this kind of assessment and allows appropriate time to be given to it.« (In the short-term, it is possible for T to be helped if he can enlist the services of a tame linguist to do the analyses for him.)6 In relation to the question of time, though, we would make two points: (a) T should bear in mind that while any syntactic technique seems complex and cumbersome on first acquaintance, familiarity and practice promote accepta bility. Also, it will not prove necessary to practise any technique with the utmost attention to detail on all occasions. Anyone who uses the procedure described in this book (or. any other) more than three or four times will find that he has developed an intuition about the kind of syntactic pattern likely to be important, and will be able to make rapid and reasonably reliable judgements about syntactic abnormality without having to work systematically through the entire procedure. But having said this, there is no gainsaying that initial mastery of the technique will take time: there is no quick and easy way to learn syntactic analysis. (b) But the problem of time must be put in perspective, and assessed over the long-term, in relation to the total anticipated contact time between T and P. The argument here is that time used in analysis and interpretation in the early stages of remediation is likely to save time later on, and obtain more definite progress in the meantime. We have seen the situation obtain repeatedly, particularly in adult dysphasia clinics, where many hours are spent in maintaining a good degree of personal contact between therapist and patient, but with little overall linguistic progress being achieved, due to no time having been put aside for the systematic analysis of each patient's problems. We hope there will develop an increased awareness of the essential role of analysis and interpretation sessions in providing a remediation service, as we see this, in the long run, as being one of the main ways in which clinical time can be saved. (iii) Gradability. The structural features of the model must be presented in terms of some explicit principle of developmental progress, if a systematic remediation procedure is to emerge. The question is, Which? Our answer is on broadly similar lines to that of Norma Rees (1971), in a paper on the bases for decision in language training. She reviews six possible methods, and we follow her classification. (a) The first-and in many ways the most obvious-is to organize structures in terms of some independent measure of complexity, and a number of attempts have been made to do this, mainly working within the framework of generative 4 There is a natural but regrettable tendency to underestimate the time needed to do clinically usable linguistic analyses. The Quirk Report (London. Department of Education and Science, 1972), for example, makes much of the importance of proper linguistic analysis, but does not discuss the extra hours that would have to be found if such analyses were being done routinely. a But here too there is a need for caution. No linguist should be brought in unless he has first spent some time observing clinical situations, and has learned something of the priorities involved. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 26 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY grammar, and suggesting, for example, that the complexity of a sentence is in direct proportion to the number of transformations used to generate it (see Yngve 1960, Hayes 1970, Fodor and Garrett 1966, Brown 1973, for discussion). The problems with all these approaches are succinctly summarized by Rees (1971, 287): The structural complexity basis therefore appears to offer few substantial guidelines for making decisions in language training. The major problem appears to be that no comprehensive and undisputed analysis of grammatical complexity in relation to sentence processing has been made available. A second problem is that, even should a satisfactory analysis for processing sentences by adult language users become available, applying the results to language training would still involve the untested assumption that the degree of difficulty in sentence production for adults corresponds to the order of difficulty children experience in mastering grammatical constructions. Current psycholinguistic theory has no coherent explanation for syntactic com plexity, and any attempts to assert that some structures are 'more basic' or 'more complex' than others, on intuitive grounds, soon land one in difficulty, if anything , other than the most elementary of sentence-building processes are being investi gated. Differences of opinion rapidly arise. Menyukand Looney (1972), for example, assume that appositional structures are more complex than relatives (e.g. Paul, a Boy Scout, knows how to tie knots is more complex than Paul, who is a Boy Scout, knows how to tie knots), which the present authors find debatable. Lee (1969) sees the order of complexity of wh-questions one way; Ingram (1972d) supports a different ordering. Many such examples could be given. 6 Remedial procedures which attempt to operate in this way sometimes begin plausibly enough, but as structures build up, they rapidly become arbitrary, selective, and impressionistic in a word, pseudoscientific. We were unable to use Conn 1971 after its early stages for this reason, for example. Nor were we able to use those approaches where the measure of structural complexity contradicted developmental ordering, e.g. Morehead and Ingram (1973) found that some transformations are earlier than their base forms (e.g. he going before go). Using a linguistic model of structural complexity without a developmental dimension inevitably leads to arbitrariness in assigning values to one's formal categories. Dever's approach, for example, aims to be a purely synchronic descrip tive inventory, which does not predict development: 'our attempt to classify was made in terms of the language patterns which occur, not in terms of the develop ment of children' (Dever and Bauman 1971,15). Each utterance should be capable • Lee, in common with most investigators working along these lines, does not explain why a structure X is more or less complex than Y, and there thus seems to be a great deal of arbitrary pointing in her method, e.g. this/that scores I, these/those scores 4; pronouns score more than nouns (see further Tyack I972b). Lee and Canter say (1971, 319): 'Within each classification, specific words or structures have been grouped into what is believed to be a general developmental order" on the basis of 'presumably similar degrees of difficulty'-but we are given no rationale, and decisions are often puzzling: e.g. Lee (1969) considers is going to as 'more immature' than will. Lee (1974) is much more satisfactory in motivating the scoring procedure, and a number of modifications are made. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC REMEDIATION PROCEDURE 27 of being classified at one place in the system: the scale 7 'requires only that the child be understood--classification then becomes a matter of applying the rules in the scale' (11). His scale, then, 'classifies degrees of approximation to adult English' (17) for clause structure (sentence and phrase structure are to be investigated at a later date). But Dever finds it difficult to match the rank-analyses of the various clauses, and his approach is thus severely limited, to the study of single clauses and clause-types: it cannot be used 'as a scale which indicates generalized develop ment across clauses' (20), which in our opinion is where most of the interest lies. (b) A method of ordering structures in terms of their cognitive complexity is also a theoretical possibility. Rees (1971) reviews the main opinions on the relation ship between language and thought (that language is central to, incidental to, or parallel to thought) and concludes that 'the literature ... appears to offer no unified basis' (297). We agree. There is as yet too much controversy for a single viewpoint to suggest itself for reliable widespread clinical application, and there is a tremendous gap between what is known about general cognitive processes and the detail which would need to be established before a reasonably complete remedial programme could be constructed. A few structures have begun to be investigated from the point of view of cognitive/linguistic influence (e.g. Donaldson and Wales 1970), but no overall normative picture has yet emerged. (c) It is also possible that structures might be ordered in terms of the perceptual difficulty involved in encoding or decoding them. Certainly Ts are usually well aware of the problems involved due to segmentation, sequencing, category-matching etc. in visual as well as Iinguistk modes, and there is no denying that a structural grading in these terms would be invaluable. But at present this reasoning is valid on a very small scale, for example in deciding the relative ease of two constructions. Rees thinks that this principle 'suggests useful answers in some cases' (300) and offers 'promising insights into the relative difficulty of grammatical structures upon which the clinician may base his decisions' (302). The technique is rather ad hoc, however, and the notion of perceptual difficulty is still extremely obscure, so that decisions about ordering are inevitably to some extent arbitrary. (d) Rees also discusses the possibility that mean length of utterance might provide a basis for structural ordering, and reaches a negative conclusion, as we have already had to do (see chapter I). (e) Nor is there anything to be said about her fifth area, which suggests that structures might be ordered on a dialectal basis-that is, T should provide 'practical information about the structures that the child hears most and with which he will obtain the greatest reinforcement' (302). This is more a general orientation for T than an actual procedure, we feel. Frequency information about structures is difficult to obtain (especially for colloquial speech), and even if norms could- be 7 The Indiana Scale of Clausal Development is a classification of the performance patterns of the spontaneous speech of children between 17 and 40 months, using tagmemic clause analysis (cf. Elson & Pickett 1965). On Dever's notion of performance, see further p. 36 below. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 28 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY established, the relation between frequency and complexity in syntax (as in vocabu lary) has hardly been investigated. All that seems certain is that there is no straightforward relationship between them: the most common constructions are not necessarily the least complex, and vice versa. (I) It is Rees's final possibility which, she considers, offers the most grounds for optimism, and which corresponds to the basis on which we ourselves work. This is to order syntactic structures in terms of the normal developmental sequence in the child. Rees finds this 'a useful source of guidelines ... its application has already proved effective' (302). We would add that, from T's point of view, it is unavoidable: 8 consciously or unconsciously, the analysis of a syntactic disorder inevitably involves a comparison with normal development, and the more the comparative basis can be formalized, the more accurate and constructive one's assessment will be. Rees makes the point: 'The assumption that the normal sequence is somehow the "right" sequence for the language-disordered child to follow has not been proved, but neither has it been seriously challenged' (289). It would be, presumably, if (i) some independent m~asure of complexity could be evolved which produced better results-but we have already seen that this is unlikely in the near future; or if (ii) the majority of syntactic disorders were characterized by a predominance of deviant patterns falling quite outside the usage of the normally developing child, and where information about normal syntactic development would accordingly have a limited or negligible explanatory role. To form an opinion on this point, however, one needs to be very clear as to what is meant by the term 'deviant'. This term is used in the literature on disorders in many different senses, of which three are particularly important: a very general sense, in which deviance subsumes all kinds of linguistic disability (including delay); a more restricted sense, which includes Ps where the range of structures used is comparable to an earlier stage of normal development, but the frequency of use of specific structures falls outside normal expectations; and an extremely narrow sense, in which only certain types of structural abnormality are labelled deviant. Our usage in this book inclines to the latter, which is close to the general sense of the term in linguistics: here, a deviant sentence is one which in some respect does not conform to the grammatical rules of the adult language, e.g. cat angry are. In this sense, considerations of frequency of use are not criteria!. Some sentences which would be deviant from the viewpoint of adult grammar are however possible when seen in terms of normal child development, e.g. want horse, and such constructions would not therefore be considered by us to be deviant, if they occurred as part of P's utterance, but instances of delay. Our final definition of deviance, then, includes only those sentences which would be both structurally inadmissible in the adult grammar, and • We therefore find it difficult to accept Lee and Canter's view (1971, 335): 'DSS should not be considered by clinicians as a test of syntactic or morphological development, but rather as a clinical procedure for analysing verbal performance and planning appropriate remedial measures.' But can this distinction really be made? And is it possible to plan without any developmental perspective at all? If not, then it is important that this be made explicit. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC REMEDIATION PROCEDURE 29 not part of the expected grammatical development of normal children (insofar as this can be established by reference to the language acquisition literature). For example, if P said cat the, this would be considered deviant on the grounds that an adult grammar would reject this construction and that this pattern has not been noted as a regular feature of normal language development. It is obvious that this definition is not as precise as we would like it to be: there are too many gaps in the research literature for us to be sure sometimes whether a construction is deviant or not; also, we must not underestimate the extent to which problems of grammatical acceptability (see Quirk and Svartvik 1966) cause difficulties. But as an operational tool, we have found this notion of deviance usefuL It must however be clearly distinguished from other senses found in the literature on syntactic remediation, e.g. Leonard (1972), who emphasizes the importance of different frequencies of use as a criterion of deviance v. delay. Our assessment procedure focuses at present only on the qualitative range of structures used by P. We only ask the question whether, regardless of frequency, the structures used are possible adult or expected normal child sentences. If the answer is no, this is then a deviant sentence for us. If the answer is yes, it is nondeviant-though it may still be deviant, in a frequency sense, for other scholars, e.g. Menyuk 1964, Leonard 1972. This does not commit us to the view that Ps with no deviant sentences are 'delayed' in any simple sense. The concept of delay is at least as complicated as that of deviance. Some Ps are delayed quite literally, in the sense that their speech is a replica of their juniors; some are delayed in certain structural respects only; some are delayed solely in terms of frequency of use. Most definitions of delay contain a number of criteria, therefore: e.g. Ingram (I 972a, 89) and Morehead and Ingram (1973, 340) conclude that disorder is a function of slower onset of time of appearance of a structure, its less frequent and less creative use, and its slower acquisition time. Comparison of different studies is complicated also by problems of selecting and matching the groups of children being compared, and also by the use of linguistic models which focus on different features of syntactic structure. We nonetheless have the impression that most of the studies of the nature of a language disorder conclude that deviant sentences in our sense are relatively uncommon with children, and only slightly more of a problem with adults. It would therefore seem to follow that the scope of a remediation procedure can be modelled on the range of structures present at the various stages of child develop ment, up to and including adult use. Dever, for example, says (1973, 3): 'a descrip tion of what normal children do while learning ~heir native language will constitute an adequate statement of the sequence of goal behaviors of a language teaching program.' And in essence, this view seems to be widely supported: see, for example, Lenneberg et al. 1964, Lackner 1968, Miller and Yoder 1974, Rosenberg 1973, Pressnell 1973. To use a normal developmental hypothesis as a basis for ordering structure has much to commend it, particularly as its empirical support is based on the fewest possible assumptions about complexity of language processing. It is important to realize that claims made on this basis say nothing about complexity, of any kind. Just because a child learns to use structure X before structure Y does not necessarily The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 30 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY mean that X is less complex than Y. All sorts of other factors-of which motivation is probably the most important-interfere with such a judgement. All that can be said is that if it can be shown that there is a normal (i.e. natural) developmental path for syntax, this is itself a motivation for developing a remedial procedure in its terms. How the procedure is to be implemented is of course a separate issue for T. We are not saying that the disordered child must be treated as if he were normal, e.g. in terms of the amount of routine correction of 'mistakes' which he might be given. Specific therapeutic procedures will be necessary. But we are saying that the grounds for selecting the structures to teach, the specification of the teaching goals, and the entire process of assessment can usefully be done within a framework geared to a normal developmental scale. And this is what we shall present in chapter 4. Before going on to this, however, there are a few other points which need to be made about the use of such a scale. To begin with, it must be made clear that, in the present state of knowledge, we see this scale rather as the 'least unsatisfactory' method of all that have been suggested. The scale of chapter 4 is a synthesis of what is known about developing syntax in children. There are, consequently, many gaps. But while language acquisition studies of syntax have a great way to go before anything like a complete account emerges, we feel that enough information is available now to enable the outline of a remedial approach to be drawn with confidence, and for many areas of detail to be filled in. For assessment, as we shall see, it is the well-defined and comprehensive outline which is all-important-the notion of a 'syntactic profile'-and here there is enough data on which to build. A related point is that we maintain that any linguistic remediation procedure based on developmental norms that does not attempt to correlate its structural stages with chronological age is shirking the main issue. Chronological age is the one thing that impinges on T at all stages, and it cannot be ignored. That a child is not 'up to his age' is the basis for most parental concern, and enters into the decision-making process at all stages-referral, assessment. and release. (It is less important in deciding upon the path through a remedial procedure.) If, then, a syntactic procedure is to be of use to teachers, educational psychologists, clinicians, and others, information about chronological norms must be induded. Popular and professional intuitions make the correlation: it is the job of the linguist to explicate it. On reading the language acquisition literature, however, the most noticeable thing is the way in which writers al'oid the issue of chronological norms. Thus Brown (1973) summarizes a decade of attitudes when he writes: 'Though the order of acquisition of linguistic knowledge will prove to be approximately invariant across children learning one language and, at a higher level of abstraction, across children learning any language, the rate of progression will vary radically among children' (1973. 408). We accept that information about ordering is a valuable means of grading structures for therapeutic intervention, and also as a research tool, but by itself it is of restricted value in assessing the kinds and degree of severity of child language disorders-particularly as most Ps are evidencing some stage of language delay. How great is the delay is the persistent question. This is the information that parents want to have, and no amount of talking around the The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CRITICAL FACTORS IN DEVISING A SYNTACTIC REMEDIATION PROCEDURE 31 subject in terms of 'syntactic relativity' avoids this. So the question is, Is the variability in rate of syntactic development pointed out by Brown and others sufficiently great for the whole notion of chronological age norms to be thrown out altogether? This would seem to be an absurd conclusion. All of us have clear intuitions about norms of fluency and expressiveness in young children.9 We are aware that some children are 'very advanced for their age' and that others are not very talkative. In the light of this, it is likely that the emphasis on rate variability in the literature is at least partly due to analysis so far having been restricted to intensive studies of a very few children: differences between individuals become more marked under the microscope, and as a larger range of children come to be studied, we predict that striking similarities in rate of acquisition of structures will emerge. When one compares the empirical findings of language-acquisition studies, in fact, it is surprising how similar dates of predominant development of the various syntactic structures turn out to be. Onset time and hypothetical acquisition time may indeed vary, but means display a remarkable correspondence. For example, every reference we have ever seen to the emergence of two-element sentences (e.g. daddy see) cites the age-range of between 18 months and 2 years. But even when we look at more specific features, such as question-words, specific word-order patterns, modal verbs, and so on, we find that their dates of emergence with a productive role generally cluster within a period of about 6 months. A 6-month variability (±3 months around a mean date of emergence) may seem vague to the theoretical psycholinguist, but it is an extremely precise indication in remedial work, compared with the impressionism and absence of quantitative measures of syntactic development that is apparent. Lastly in this section, it might be objected that, while structural grading based on norms of child development may be satisfactory for remedial work with children, this principle is inapplicable for work with adults. As will be clear from chapter 8, we have not found it so. If an adult displays a disorder capable of being defined in syntactic terms, we find that the assessment of the disability, and its subsequent remediation, can be carried out using exactly the same scale of syntactic development as we use with children. There are certain differences in the use of the syntactic procedure, which will be referred to in later chapters, but these are outweighed by the similarities. Myerson and Goodglass seem to have reached a similar conclusion (1972, 41): Taking samples from patients at various severity levels is analogous to examining speech samples from children at different ages.' This is not to subscribe to a general regression hypothesis for adult dysphasics, i.e. the structures the adult will have learned last will be lost first. We do not know whether this is a defensible hypothesis, evidence of gradual loss of structure being • We know of no systematic work rating people's abilities to make consistent judgements of such levels of ability-whether this varies between mothers, fathers, nurses, linguists etC.-but we have no doubt that such abilities exist and will provide part of the evidence for descriptive decisions. Naturally, any well-developed theory of syntactic ability will have to take into account the variables of socioeconomic background, sex, intelligence, health, and so on; but we do not wish to under estimate the ability of (especially) the mother to take these variations unconsciously into account in judging the level of linguistic development of the child. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 32 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY almost impossible to find, in view of the nature of the traumas involved; but there is certainly evidence against it: e.g. Myerson and Goodglass show that optional adverbials are quite well retained in their Ps, though these are learned relatively late, whereas some quite basic structural relationships had disappeared. Moreover, we do not know-nor, in the present state of knowledge, see a way of knowing to what extent a disorder is a matter of competence or performance, i.e. whether the underlying system is lost, or is there but unable to get out. The latter view seems the more likely (as argued by Weigl and Bierwisch 1970, Sefer and Shaw 1972 etc.), in view of the way in which most Ps we have encountered respond to therapy, but this is speculation. In other words, we have no theoretical reasons for using the child development scale in relation to adults. Our arguments are prag matic: we have found no viable alternative, and by using it satisfactory progress in Ps' use of syntax has been achieved. In this chapter we have discussed some of the criteria we use in order to evaluate remedial procedures in syntax, and which we have tried to follow in developing our own. We present our procedure in chapter 5. Before doing this, however, we must outline the descriptive and developmental frameworks we shall need to use in order to meet the demands made above. Chapter 3 describes the system of syntactic description, chapter 4 the stages of syntactic development in children. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 3 The syntactic framework of description It is clear from the preceding chapters that a number of different linguistic approaches have been applied to the analysis of language disorder. It would be unwise, however, to assume that all of these approaches, and the models they use. are equally easy to work with in the study of a given applied language field. As far as the subject of language disability is concerned, this is certainly not so. We have looked at this field from many positions within linguistics, and find that whereas all theories are in principle capable of describing the character of a language disorder, certain approaches are much more immediately illuminating and easier to work with than others. Before giving a detailed description of the approach we have chosen to work with ourselves, therefore, it may be helpful to outline the main reasons which led to our choice. l For historical reasons, in this area of evaluation, the first question that has to be posed is how usable is the transformational-generative (TG) approach in investigating these matters. It would seem sensible to open any discussion with reference to this view, as owing to the influence of TG theory over the past twenty years, many Ts, having learned something of it, find that their first question is pre-empted by the climate of opinion, and they find themselves asking 'Do I use TG in my work or not?' And certainly, there have been a number of attempts to use TG in one or other of its forms in the study of language disability, e.g. Braun and Klassen 1971, Wachal and Spreen 1970, Spreen and Wachal 1973, Muma 1971, Hass and Wepman 1969, Leonard 1972, Menyuk and Looney 1972. Davis 1973, Morehead and Ingram 1973. Some indeed have claimed that only TG is worth using, e.g. Rosenberg, cited in Dever and Bauman 1971. In addition, there are a number of approaches which incorporate aspects of TG into their work. e.g. Myerson and Goodglass 1972, Miller and Yoder 1974. Lee's work (see above) claims to be transformational in principle, but it is hardly that. as the approach is basically a selection of hierarchically organized structural features, largely operating at phrase level (cf. below), with some analyses of generative origin, especially of the verb phrase, introduced. We ought also to add that frequently work in language disorders seriously misunderstands the nature of a generative grammar, usually by confusing the notions of competence and performance. We have often encountered Ts who see TG as a language-teaching kit. and who think I Until theoretically sound evaluation procedures for grammars are devised, decisions about the choice of grammar to use must inevitably be pragmatic and to some degree personal. The following discussion of alternatives very much reflects Ihese considerations. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 34 THE GRAMMATICAL ANALYStS OF LANGUAGE DISABILITY that what is being advised is to teach a child phrase-structure rules first, and transformations later! There is, then, a great deal of interest in the application of TO in this field, but the results have been disappointing. The main reason is that the most important studies have shown that the salient differentiating features are precisely those not readily describable in terms of the most important characteristics of the TO model. The set of transformatiuns, for example, as defined by Chomsky (l957a or 1965) have turned out to be singularly undiscriminating. The syntactic features which most directly distinguish the various kinds of ability which we want to characterize emerge as being sentence construction types (i.e. patterns of Subj-!ct, Verb, Object etc.; see below), which TO on the whole pays little attention to, and which other models treat in a much more direct way. Muma (1971), for example, uses a 1957 model of TO to analyse 13 fluent and 13 nonfluent 4-year-olds, and finds no difference in the overall number of transformations per sentence. 2 He ends his paper suggesting that alternative approaches might be better, especially in so far as they would analyse performance and pay attention to sentence-types (which did differentiate his groups). Morehead and Ingram (1973) use Rosenbaum 1967 to do a complex transformational analysis of 15 normal and 15 deviant children, selected to represent the five linguistic levels previously determined by Brown (1973). In order to establish the nature of the deficit in the deviant group, they examine a large number of transformations, but find little difference in either the organization or the occurrence of transformational features at a given linguistic level: 'there is considerable similarity in the transformational development of both groups across the five levels of linguistic development' (337). On the other hand, information about constructional type does have value: 'deviant children appear to be signifi cantly restricted in their ability to develop and select grammatical and semantic features which allow existent and new major lexical categories to be assigned to larger sets of syntactic frames' (343). Likewise, Morehead (1972) concludes that comparison in terms of constructional type is 'the only meaningful difference' (4) between normal and deviant children: 'the deviant child's deficit lies not in his ability to develop base aspects of grammar but rather in his ability to develop additional terms and the relations in which to use those terms' (5). (See also Lozar et al. (1973, 29), who found sentence-level units to be the only significant differences between retarded and normal children's syntax.) The main evidence for the difficulty in using the TO model, then, comes from those who have tried to work with it systematically, and it is interesting that, in their search for alternatives, their suggestions all point in the same direction. This direction is precisely that concluded by Longhurst and Schrandt (I973) who, having compared four measures of analysis, say: 'A renewed interest in basic structural linguistic concepts could prove valuable to the assessment of language development in children. This interest should be focussed on such topics as analysing how-verbs develop at six-month intervals or how the various "features of arrangement" 2 There was some suggestion that the nonfluent children used more single-base and less double-base transformations, in fact, but this was not felt to be significant. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE SYNTACTIC FRAMEWORK OF DESCRIPTION 35 develop .. .' (248). This is exactly our position, as will be described below. But it is worth pointing out that a structuralist orientation does not preclude our using certain transformational notions; indeed, the grammatical framework on which our approach is based frequently incorporates them, especially in describing the output of older children and adults. But we have never found an analysis in transformational terms to be useful for more than a small part of the overall picture. We do not wish to speculate at length about the reason for this, but if anything it is likely to be due to the difficulty in working with any model based on the fundamental insight of the TO approach, namely, that language can be seen as a single system of generative process. A TO grammar works by assigning the same initial analysis to all the sentences of a language (of which the earliest and most well-known. formulation was S --.. NP + VP, i.e. 'Rewrite Sentence as Noun Phrase plus Verb Phrase'), and then progressively expanding, removing, or reordering the various constituents by a sequence of rules so that one finally emerges with a detailed specification of the structural features of a sentence. Working backwards, this means that, given the structural analysis of any sentence in the language, it is possible to follow the reasoning which led to this analysis being assigned by examining the various stages in its derivation; ultimately, the sentence will be shown to have been generated from the initial rule, and, by implication, related to all other sentences in the language. The possibility of attributing psychological reality to these underlying, or 'deep' structures has then often been claimed. Now this kind of approach has some plausibility in normal adult language studies, particularly when it succeeds in explicitly and formally relating sentences which we intuitively feel to be closely related, but which traditional approaches to grammar would have kept apart or ignored. But in studies of abnormality, where the intuitions of the 'native speakers' are fragile and indeterminate, to say the least, the justification for assigning a particular derivation to a sentence becomes extremely difficult, and il) usually arbitrary and implausible. We would want to argue that a remediation procedure in particular should make the fewest possible assumptions about the nature of the mental reality underlying speech, and concentrate instead on an exhaustive account of the characteristics of the speech actually produced by P and T. While accepting that there may be a case for the empirical study of competence (in Chomsky's sense) in normal adults, we find this notion, its formalization in transformational terms, and the more powerful notion of innate structure, which is usually associated with it, extremely difficult to work with in the young child, and impossible to work with in evaluating disorders in either children or adults. In such cases the distinction between competence and performance becomes an unreal issue, and the linguist is forced to rely on a minute analysis of whatever speech he can manage to elicit, in order to provide a basis for therapeutic intervention. (This means, among other things, paying particular attention to the intonational, pausal and other prosodic features of both P and T, which we have found to play an influential role in both assessment and remediation, as is illustrated in chapters 7 and 8.) Scepticism about competence models, and emphasis upon performance data, is in fact increasing in the field of language disturbance, e.g. Engler et al. (1973, 196) conclude: 'the task The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 36 THE GRAMMATlCAL ANALYSIS OF LANGUAGE DISABlLlTY is to identify and count constructions in speech samples of real clients rather than to ponder the competence of the "ideal speaker" , (and see also Morehead and Ingram 1973, Lee and Canter 1971). Two points should be noted here. Firstly, the scepticism is being directed at a specifically Chomskyan notion of a competence involving deep structures (which are only one way of formalizing our intuitions.) Everyone must be interested in making claims about the underlying system of the child's language ability; it is the way of approaching this which is under attack. Secondly, one wants to argue for a proper emphasis on performance, not an exclusive concentration on it, as again it is the systematic use of language that counts in any assessment. Confusion over both these points underlies aspects of the discussion in Dever and Bauman 1971, for example. They claim that their approach, derived from tagmemics, is a grammar of performance-but this is not so. Their grammar does pay more attention to features of performance than TO, it is true, but that does not make it a performance grammar in any strict sense nor do the authors want it to be. That their real concern is with the underlying system emerges clearly at many points, e.g. 'the single most important defining feature of any clause is the presence or potential presence of a verbal element as one of the nuclear tagmemes' (17, our italics), or again, and more explicitly: 'In their abstract form any utterance rule could become confused with almost any other rule. Therefore, use of the scale relies very heavily on the knowledge that each native speaker has of his language for determining whether a clause is to be classified as being declarative, question or imperative' (20), and 'there seems to be no efficient way to define these categories except to say that every native speaker of English knows when a clause is to be classified as one of the three' (I 7). 3 There is also a pragmatic difficulty which accounts for our disillusionment with TO. This arises out of the rapid development of that theory since 1957. This has caused inevitable difficulties in all applied fields, as practitioners continually run the risk of being told that their conception of syntactic theory is out of date (even if, sometimes, by only 6 months or a year!). Much of the early work in clinical applications, which tried, for example, to analyse disorders in terms of kernel and nonkernel sentences, would now presumably be dismissed as ill-conceived theoretically (e.g. Lee and Canter's use of the 1957 analysis of the verb phrase; or Davis (1973), who distinguishes kernel sentences, minor transformations (e.g. inflections and function words), and major transformations (change of sentence type». So would Hass and Wepman's views (1969), e.g. 'It now makes sense to consider deep structure as forming the same sort of borderline between syntactic and semantic processes as surface structure forms between syntactic and phono logical ones' (304). The differences between the various positions are by no means trivial, and make it extremely difficult to compare the results of people working with different models, e.g. Muma, and Morehead and Ingram. And if the inquirer • Dever and Bawnan seem to have underestimated the role of 'surface' structure (continuing the generative metaphor) in reaching this conclusion. in fact. After all, if the categories are being used systematically by native speakers. there must be a basis for this. Intonation in particular is a factor, as they recognize, but while they accept that their rules vary 'only in the intonational envelope at times', this is nonetheless 'a feature which is not formally expressed in the scale' (20). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE SYNTACTIC FRAMEWORK OF DESCRIPTION 37 after syntactic truth seeks to avoid the theoretical disputes of the early TO models by turning to the more recent work, there is no solace: controversy and change abound still, and the emphasis is increasingly less syntactic and more semantic, with few frameworks of any descriptive range and depth of detail emerging. Until such time, then, as successful studies of disorders are made within the TO tradition, we prefer to develop our syntactic procedures using other sources. But which? The safest course of action would seem to be to choose a syntactic framework that has been fairly stable over a period of time, and which is sufficiently flexible to permit the use where necessary of insights from .the recent approaches. The obvious place to turn is to the general tradition of structuralist linguistics in the United States and Europe, which was the dominant mode of syntactic investiga tion in the period preceding Chomsky (late 1940s to mid-1950s). Many distinct theoretical approaches developed out of this tradition-perhaps the most weB known being those of K. L. Pike ('tagmemic theory') in the USA, and M. A. K. Halliday (first 'scale-and-category grammar', later 'systemic grammar') in Great Britain. Richard Dever and his associates are currently working on the clinical application of tagmemic theory (see Dever 1972a, 1973, Dever and Bauman 1971, and above, p. 23). We know of no work being done using a Hallidayan framework. There is however a certain amount of published material which does not aim to start from a coherent general linguistic theory, but works within an ad hoc structuralist framework, e.g. Engler et al. 1973, Conn 1971, Thomas 1971, Morehead and Johnson 1972, Ingram 1972b, Foster et al. n.d., Williams and Naremore 1969, Reynell 1969. Most of this, we find,. is clinically highly relevant. At times, indeed, the similarity to our own approach is so close as to be almost embarrassing I Engler et aI., for example, have developed a technique they call the 'linguistic analysis of speech samples' (LASS), which aims to produce a 'diagnostic profile' of spontaneous speech, is eclectic (incorporating ideas from tagmemics, immediate constituent analysis, and TO), and contrastive (samples are compared with the language of the clinician or with other children of the same age and development). They also pay particular attention to the role of the verb and its classification (though Longhurst and Schrandt (1973, 246) claim that this could not distinguish the two children in their study). The parallelism with chapter 5 below is remarkable. There are however important differences. Perhaps the most far-reaching is that Engler et al. do not make any explicit connection with an established grammatical tradition; as already mentioned, their article presents a highly simplified and schematic account of English syntax, with no criteria or discussion to help T handle the problem areas of analysis. Also, because it is not geared to any set of developmental norms, their analysis is burdened with a number of descriptive categories which we have found to be unnecessary. But the general emphasis in their approach (as well as that found in the other references above) is correct, in our view, and we have been much reassured in our own work by becoming aware of these parallels. ~ In our work, then, we have tried to make use of a descriptive framework which reflects the above emphases, and which is easily and directly applicable to the study of grammatical disability. We have chosen to work with a framework The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 38 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY based on a recently published approach entitled A grammar of contemporary English (Quirk et al. 1972). An abridged version is available under the title A university grammar of English (Quirk and Greenbaum 1973), and as this is in many ways a more practicable manual for Ts to use, we add references to this (UGE) as well as to the main grammar (GeE) in the description below. This grammar is the culmination of a decade of work directed by the first author at University College London, where the Survey of English Usage has been in progress since 1960. This survey has as its aim the full description of actual educated spoken and written usage, which would constitute a 'descriptive register' capable of acting as a basis for realistic handbooks of English grammar~in much the same way as the great New English Dictionary has acted as a sourcebook for many smaller 'practical' dictionaries of the language. GeE is the first major synthesis to be based on this data, and in its empirical emphasis one finds in its pages much more information about the norms of spoken syntax than was presented in the earlier grammatical handbooks of English (such as those of Jespersen, Poutsma, Curme and Kruisinga), within whose general tradition this book falls. Most of GeE's examples are taken or adapted from the collections of data constituting the survey, or from the acceptability experiments which were associated with it (see e.g. Quirk and Svartvik 1966). The emphasis on intonation in relation to syntax is a major feature, as is its stylistic range. The scope is 'the grammar of Educated English current in the second half of the twentieth century in the world's major English-speaking communities' (Preface), emphasizing the 'common core' of the English of serious exposition, but by no means ignoring the restricted styles. It gives clear recognition to the existence of variations in use within the language~ to such differences as distinguish speech from writing, formal from informal, American from British. There is no condemnation or neglect of minority usage in this book: it is, above all, a democratic grammar. GeE is intended to function as a reference grammar, that is, as a handbook which can be consulted in order to find out the facts of usage in the language, the regular patterns that constitute the grammatical structure as well as the exceptions to-rules and problem cases. It is not the kind of grammar which aims to present a coherent theory of language structure as a whole (including phonology and semantics) or to maintain a particular theoretical viewpoint that is generally applicable to languages other than English~such as one finds in Chomsky 1957a. There is a minimum of theoretical exposition, or discussion of how the facts of English motivate one to see grammar in one way rather than another. As far as is feasible, the patterns of structure are allowed to speak for themselves, with the emphasis away from general theoretical implications. And conversely, whenever a feature from some linguistic model seems illuminating for an area of English syntax, it is used. This is not a transformational grammar in principle, for instance, but the use of transformations is to be found at various points throughout the book. To some theoretical linguists, of course, this kind of approach to a language is in principle unsatisfactory, as for them the facts of English are of interest only insofar as they shed light on more general processes of language formation, and to propose an 'eclectic' account of a language (I.e. one which brings together The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. i THE SYNTACTIC FRAMEWORK Of DESCRIPTION 39 different theoretical traditions and assumptions without explicitly relating them into a fresh coherent theory) is to bypass major difficulties. But for those interested solely in establishing and interrelating the patterns of structure of English as an end in itself, the lack of any such general theoretical foundation will not be crucial. Of far greater importance will be the comprehensiveness of its lists of features (e.g. irregular nouns, types of sentence connecter) and the clarity of its descriptive apparatus and terminology; and in both these respects we have found GCE to provide less of a problem than the other approaches with which we are familiar. Most of the terms in it will in fact be recognizable to anyone brought up on a diet of school grammar (though of course some of the definitions have had to be modified, in the interests of consistency and precision). Terminology and notation from a theoretical point of view, are of course of rdatively trivial concern (but cf. Lyons 1963, 5). What you call an entity is unimportant, compared with the reasoning and criteria which led you to identify that entity in the first place. But in applied fields of study, terminology takes on an importance far out of proportion to this, and rightly so. It is arrogant of the theoretical linguist to present a linguistic technique for prac:tical use by non linguists without making some effort to bridge the gap from his side between his own technicality and the tradition of the nonspecialist, and one way of doing this is by the deliberate use of familiar terms. People seem more ready to take to familiar terms which have been given new definitions than to accept new terms which go with such definitions, particularly when-as is normal these days-the new terms are not isolates, but complex sets of terminology and notation, which must be mastered all at once, as a system, if any of it is to be learned at all. It is essentially a question of basic motivation, in our view. People seem to be more motivated to master and use a technique if they are not put off by its appearance-and no amount of reasoning from the theoretician that 'appearances don't matter' seems able to alter this fact of busy clinical and teaching life. It therefore ranks high in our list of syntactic priorities. We are, after all, only interested in developing a syntactic procedure if it stands a reasonable chance of being used. The main reasons why we use this approach, then, are twofold. Firstly, it provides as comprehensive a description of contemporary English (at least of its educated 'core') as the present state of knowledge allows, and this ensures that the vast majority of the structures used by P or T will be given some treatment. Secondly, we have found that the organization of this grammar in terms of levels (of sentence, clause, phrase, word), its emphasis at clause level on the functional relationships between the various elements of structure, and its account of the role of intonation in English syntax, permits a direct and economical description of the data of syntactic disability, and correlates well with T's intuitions about syntactic patterning (see further chapter 6 for discussion of disorders located primarily at clause level, phrase level etc.). Before introducing the range of structures from GCE which acts as our syntactic framework, it is perhaps worth outlining the organization of the book as a whole, and noting any major departures we make from it. We shall be cross-referring to the various sections of the book during our own exposition below; consequently The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 40 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY it would seem useful to present some overall conspectus of its contents. (Alterna tively, this summary may be read after our own exposition, in which case the reader is advised to continue now on p. 42.) Cbapter 1 provides a general discussion of the contemporary importance of the English language, of the main senses of the term 'grammar' and its relationship to other aspects of linguistic structure, and reviews the main varieties of the language that any description has to distinguish-notions of standard, dialect, specialist styles etc. For those who have read little in linguistic approaches to the description of English, it provides a useful perspective. Chapter 1 gives a preliminary view of the sentence. It introduces and iUustrates the main elements of sentence structure (Subject, Object etc.). and outlines the basic types of sentence, and the major transfonnational processes which operate on them (question, negation, exclamation. passive and imperative). The chapter also contains a discussion of parts of speech and some of the limitations of this notion. In other words, it provides an overview of the grammar as a whole, and should be read first, in any systematic approach to this book.4 Chapters 3 to 6 study the special characteristics of the various elements of sentence structure. Chapter 3 presents the grammar and semantics of the verb phrase; chapter 4 examines the basic structure of the noun phrase, and related items; chapter 5 studies adjectives and adverbs; and chapter 6 is devoted to the relations expressed by prepositions and prepositional phrases. Cbapter 3 begins by classifying the various classes of verb operating in the verb phrase, makes the distinction between finite and non-finite verb phrases and presents the types of each, and summarizes the characteristics of auxiliary verbs. The tense and aspect fonns of the verb are then outlined, followed by the meanings of the modal auxiliaries. The chapter concludes with a comprehensive classification of regular and irregular lexical verbs. Cbapter 4 begins with a classification of nouns, with particular reference to the division in tenns of proper/common and mass/count. The detenniner system is then outlined, with its main subdivisions into predetenniners, ordinals and quantifiers. The role of the articles in specifying the reference of nouns is then discussed, and followed by a classification of noun types in terms of number, gender and case. The chapter concludes with a classification of pronouns into their many different types. We have followed this chapter fairly closely, as many of the frequently occurring problems of syntactic analysis in English involve items , It should be noted that certain distinctions are made in the description of the verb phrase which we have not found it necessary to incorporate in our own initial analyses, namely the distinctions between 'static' and 'dynamic' verbs, between 'operator' and 'auxiliary', and the verb category labels 'ditransitive' etc. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE SYNTACTIC FRAMEWORK OF DESCRIPTION 41 dealt with here-in particular, items (e.g. enough, one, many, all) that have no single grammatical role, but function in a range of structures. S Cbapter 5 examines the characteristics of adjectives, especially in relation to adverbs, nouns and participles, and describes their use in attributive (premodifying) and predicative (complement) functions. Three classes of adverb are then dis tinguished-adjuncts (integrated within the clause), and disjuncts/conjuncts (which have an evaluative or connecting function). We do not use these distinctions in our approach. Types of adverbial intensifier are then presented, and the chapter concludes with the analysis of comparison. Chapter 6 is largely devoted to the grammar and semantics of prepositions-of place, time, cause etc. It is preceded by an outline of some of the restrictions on their use, an analysis of 'complex' prepositions, and the distinction between prepositions and prepositional adverbs. Other aspects of prepositional phrases, such as their position and internal modification, are then discussed. Cbapter 7 then takes up the study of simple sentence structure from chapter 2, and the main transformational processes, insofar as they affect all elements except the adverbial. This is the most important chapter, as it relates closely to the basis of our approach. It begins by distinguishing the seven main clause types built up out of different combinations of elements, and proceeds to detail the semantic functions of the subject and object elements of structure, and thence to a discussion of concord between the elements. The vocative is given separate discussion. The various processes of negation within the sentence are then distinguished, followed by a description of the formal features of statement, question, command and exclamation. A separate category of echo-utterances is established. The chapter concludes with a brief account of types of formulaic utterance etc.-what our approach subsumes under the heading of Minor sentences (not a GeE term). Apart from this, we follow the analysis in this chapter quite closely. Chapter 8 is entirely devoted to the detailed discussion of types of adverbials, and the reasons for distinguishing adjuncts, disjuncts and conjuncts, thus amplifying the relevant part of chapter 5. The subdivisions are largely based on semantic considerations, and we have not incorporated them into our own approach. Chapter 9 presents a description of ellipsis, coordination and apposition-three ways of expanding or modifying the basic patterns of sentence structure. A wide range of types of elliptical construction is described. The distinction between coordination and subordination is made in clause linkage, and the types of coordination illustrated, particular reference being made to the possibilities of ellipsis. The types of phrasal coordination are then outlined,again with reference lOur only modification is in the use of the term 'Initiator' for 'Predeterminer', and the general label 'Adjectival' to subsume Ordinals and Quantifiers as well as more central kinds of adjective. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 42 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY to ellipsis, and this leads into a description of types of apposition, and related kinds of connectivity. We do not in our approach enter into detail under any of these headings. but we do make use of all the major distinctions. Chapter 10 follows naturally from this, dealing with the links which relate one sentence to another. Semantic. lexical and syntactic means of sentence connection are distinguished, and the syntactic devices described in detail: time and place relaters, logical connecters, substitution forms, features of discourse reference, comparison, ellipsis and structural parallelism. Our own approach is largely within the sentence, but insofar as sentence sequences are taken into account, we follow the distinctions made in this chapter fairly closely. Chapter 11 takes up the distinction between coordination and subordination, and examines how one or more sentences may be subordinated within another. Three types of dependent clause are established on structural grounds-finite, nonfinite, and verbless; the formal indicators of subordination are classified; and the range of functions of these clauses in the sentence presented-noun, adverbial, relative, comparative and comment. The chapter concludes with the rules for the use of the verb phrase in dependent clauses, and the distinction between direct and indirect speech. Chapter 12 follows up chapter 3 by looking more closely at the types of verb phrase in relation to the complement structures which may follow them. It deals with the distinction between active and passive voice, between phrasal and pre positional verbs, and classifies the various types of complement structures. This is a complex area of English syntax, and we make use of only the most general categories in our approach. Chapter 13 follows up chapter 4 in exploring the full complexity that it is possible for the noun phrase to have by taking within itself the structures separately examined in earlier chapters. The distinctions between head, premodification and postmodification, and between restrictive and nonrestrictive kinds of modification are introduced. Chapter 14 deals with the way in which the various parts of the sentence can be arranged to have different kinds of prominence, such as contrast, emphasis and reinforcement. Constructions using it and there are described, along with variations in word order and intonation. There are three Appendices: on word formation; stress, rhythm and intona tion; and punctuation. Our syntactic description The salient features of the method of syntactic analysis we use may be summarized as follows. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE SYNTACTIC FRAMEWORK OF DESCRIPTION 43 It is normal to begin syntactic analysis with the Sentence. 6 There are three main tasks: (a) to identify the sentences in the data; (b) to analyse their structure and function; and (c) to analyse the way they combine into 'connected speech'. A Identification In writing, this is usually no problem: sentences begin with capital letters and end with one of a small set of concluding punctuation marks (full stop, question mark etc.). In speech, there is sometimes a problem: falling intonation tones and the presence of a pause may mark a sentence-ending, but they do not always, and sometimes it is difficult to say if you have corne to the end of a sentence or not. For example, in writing, these two utterances have a different structure: I'm corning at three. And Mary's corning at four. I'm corning at three; and Mary's corning at four. But in speech, these may sound exactly the same, and the analyst does not know whether to count one sentence or two. Usually not more than about 5 per cent of normal adult data causes problems of this kind. Problems of sentence identifica tion become greater in disordered speech and in child language, however, and debatable cases must be classified separately (see further chapter 5). B Analysis of sentence structure Let us assume, then, that we have before us a list of sentences whose identification has not been problematic. What are the most important features of structure to be singled out? It is not possible to summarize all the facts of sentence structure in a single statement. The idea of 'complex' sentence structure is not a single, simple notion, for instance, as the complexity distributes itself around the sentence in various ways: a sentence may be complex when looked at from one point of view, but simple when seen from another. The idea that sentences have different 'levels' of structure, each of which has its own kind of patterning, is central to our description of syntax. An analysis, also, must begin with the most general and abstract facts about sentence structure, before proceeding systematically to the more detailed points which have to be made. In this approach, three main levels of sentence structure are recognized: 1 (i) patterns of sentence and clause structure (ii) patterns of phrase structure (iii) patterns of word structure. In addition, there is (iv), the patterns of sentence connection, mentioned above. e For alternative views, using the morpheme, word etc., see Crystal 1971a. 1 These levels are also used by Dever and his colleagues (see 1971, 1973). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 44 THE GRAMMA TICAL ANALYSIS OF LANGUAGE DISABILITY C Patterns of sentence and clause structure 1 It is helpful to begin by making a very broad twofold classification of all the sentences in the data, distinguishing MAJOR and MINOR sentences. :z Major sentences, essentially, have a Subject-Predicate structure. Minor sentences do not. Major sentences are like: John kicked the ball. We're going to the match at three o'clock. All the children came in when the teacher called. Are you happy in your work? Minor sentences are like: Yes. Oh! Hello. First come, first served. (These are obviously going to be particularly common in informal speech.) 3 We must now state the range of patterns which fall under these two headings. As most of the data will probably be major sentences, this seems the obvious place to start. Major sentences may be SIMPLE or MULTIPLE. A simple major sentence is one with just a single CLAUSE. A multiple major sentence contains a sequence of clauses, which are linked in various ways. A simple sentence would be: John kicked the bal/. A multiple sentence would be: John kicked the ball and then he fell over. John kicked the ball because he wanted to. So what is a clause?8 4 Clauses are of various kinds, but their structure is always defined with reference to a small, fixed set of elements, such as 'subject', 'verb', ·object'. Exactly how many there are will be stated below. But one of them is central: the VERB. The verb is the most important part of a clause, because it detennines most of the patterns which are allowed to appear in the rest of the sentence. It is always possible to tell what the verb is in a clause, as it is the word which belongs to a set of forms which vary in a predictable way: they vary for tense (e.g. I walk/I walked); they have different participle forms (walking/walked); and they can be used immediately after the set of seven words called personal pronouns 8 This emphasis on the clause is at the core of Dever's work, especially the Indiana Scale of Clausal Development (see Dever and Bauman 1971, Dever 1973) based on tagmemic analysis (Elson and Pickett 1965), The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE SYNTACTIC FRAMEWORK OF DESCRIPTION 45 (I, you, he, she, it, we, they). Sometimes more than one word joins together to act as verb, as in He is comi,., (see further below). 5 In a clause, the verb (V) usually has a SUBJECT (S). In its simplest form, then, the major sentence is of the pattern SV, e.g. John's coming. It's raining. The big boys laughed. (Whenever a verb could be used like this, without any further structure following it, it would be caUed an 'intransitive' verb, in traditional grammar.) The only important case where a major sentence verb may appear without a subject is in commands (though even here, a subject is often present), e.g. Go! Look! You go! All you boys look! 6 Verbs detennine the structure of the rest of the clause in various ways. Some verbs, firstly, govern an OBJECT (0), as in: I kicked him. The big boys saw the little girls. S V ° S V ° This is usually referred to as the .: 'V' 'Q' Dlher Problems Conn. Clause Phrase Word vx QX SV , VCID 3 DN vv 2 2 Adj N II- ~inl =9.: sqo AX V part 4-J~ Nell X Dlher NN 2 InlX pi PrN 2 Dlher 2. , :~;~~; - - :ili'A~ll ~ ~df ;£i:~P-~ ~;_A~ -ed VXy Qxyl - .. -en • SVA ~ VOdD; • Adj Adj N Aux ., =~ It'I XY VS , "4- ). J~ NeJ XY ()(her PrDN Pron 2f · do X rl iN Adj N ()(her ..., Xy • S:NP l Xy, v:VP I XY :!~:~1!:.:. ~~P ,/, 0'1 S QIIS '.Clo, -7 -";;'-l-rN-~-Ni Nell V II-~9.: QXYZ I """ I "'M' " D '. N 2 Nell X 'cop &J: eX 3 2 Aux , . ,. 'XcX 3 4 '.lUX Iii"" Olher -I I :I I and, Coord, I J I • P05lmOO. I I' clause ..", how la, c Subord, ( If. I , I ... ~ ,"? Clause: S Pos.moo, I· -er f~ phrase who' Dlher Oause:C/O I ·Iy I a:ic Comparalivc: ( +) ( ) IVP VP Clause IVP VP Clause Initiator Com pi .. Passive Pron Adj seq Modal Concord I Coord Complemen, Del N ;rrea Tense A posilion .... ~ ... ~ V irrea W ordel' r.1 -- - - - -- - -- -- - -- - - -- - - - --- - - - • .. i ciS C Other D,her Discourse SyntQctic ComprehelllliQ1l = A Connectlyity 11 >~ Commen [ Clause III,,,,. SIyl" &+ !~ Empha,i<; Order Dther (fJ::!. Total No. 14-' Mean No. Sentences .,,2. Mean Sentence J ., Sentences Per Turn Length Ii:) O. C/)Islal, p, Flelcher, M. Garman, 19" Uni.e .. ity of Readin, Profile Chart: normal 3t·year·old The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. - fNjo_,"., ~~ V: ' . "C/O, -;Y -"ii,- l-r"-" I ,"0,. I "'~ ',D '. , c l od t d t r T Ot atic O C/)Islal, t A V.aualysed I 1 SYn1bolJ~ Nmse B Respomes I Response V"' IIlIa. I iR I" _---,-E_II_iP_I'_Ca,.-I_M_aJ-,-or_-I i , epet- r Full Struc· , Prob·i SIIm"'"s Type Totals'ilions I I ~ J 4 Major Minor lural 0 lems: ~L~~:::ons_-=-~_"'Po-+;~ _ -I_-+-_--'-_.L-----L~ ~__'___~~~:~~:~-:1 C Spontaneous 261 tti I I Minor " Clause Phrase vr QX 5V 1. veo IS 0 .... " VV seo AX :3 AdJ N J. Ii part t,.1 Neg X Other ., NN 3 In',~ 3 I PrN Z "" OI"Of ",. 2-----I--+--~~-~--+---I----~---~r ' S:NP '0 x' lI,yp -, x· C!O,NP'~\ • AAP I s-vc ° ,. ~~OA. 1210 ,,:;; ;; - j J- t~~- -4-1- SV" 2. "'P YOdO, J Adj Ad) N '\u' 105 liXI Qxtl I .. , xY vs, I "'cg n O,her , Pr ON at. Pron 2.-'0 --"-+-+-----4----+---I--f-----.-:--- I .... , .. dJ N" O,he, 2 t I.:~,~o~~i~'-::x/~p'-:;~- j:i~~t'g v ~i:f1. do XY , S A Conne~ri¥ity 12- ;: ~ 'Comment Clause 40 I Emphallc Order 'f:. QV5 I QXYZ I I ( t) ,5VO.,0, J O,hc. ~J I P. 0 AdJ .... I"l. .... os x i ..:X.:! Aux 'x,x 3 Olhe. Go 7 4-1' IS ,. III No_ Senlences Turn 'Pos.mOO. I 2.1 l' "---~ !.:Iause Poslmod 1 phrase © 0. Crystal. P. Flcll;her. M. Garm~lfl, 1975 Uni .... ersity of Reading Protilc Chart: Ilormal adull Word -mg "'1 pi ,01 oed 'll -<:n 2". Js ~~Ol .' n', -JD au:.: -cSt ·or The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Slimulus Type ~ ... Q .. u .. "ons ..~10~~ C Spontaneous ! \l 1<'1 X " , , 2.' '+ ! Olher~ 26.0 Social " SIi?,eotypes P,oblf'ms Slale",.,,,, 'N' Other Problems SI!"J. 0 .... " vv I Inl.~ i .. l or----.~----------.. --------_1 ~r_. S ':'O .:.:. YV~'" ~.:..C:!f!.~P'~'-'_A.:.~ __ s co'. VCOA.1210AdJN:l1 c,,~ If.' Y A ? . I ... . J J 0 i'oe l , p, . .' i N ... dJ " Olher 2 t --+----I-I.:-~-l:--o-~-'!:'-i-~...-;-'--:-:-X-t-:"~pr-:;~· j:i~~;g Y ~i: f3- ',5YO, l er ~I r .... l .... X ~x.:! XYI S ---+--+-----+-----j-----+----f .. ------.. I "'l' ('<>ord. 1 10 1 • • X,X r . Poslmod 1· .. ------- tag L:: Subord 1 '. I o'it,f Clause co 41 Comparative I 3 'J.,., CI.U5" S :3 -1--... .L------'-...-.--.. '---( -t-)--'----'- I .lI . t © lli · 21 .., l A B C a~ I~ ::9: I~ ;>~ ; J~ U-17M ProbIe_tic I Uomlellijjibl. 6 2 Symbolic Nois<: ] Deviant I Incomplete r, 2o'\mbiauous , R..- Normal Response AbDormal Repet· Elliptical Major Full Struc- Prob- Stimulus Type TOlals itions , 2 3 4 Major Minor lural if lems ~ Questions 3' t '3 5 , la S' I-- 10. 2+- 3. 4.~ 4-I,,. Ollie", ., Spoota_ 23 17 .. Others " MillOf Social 5".1 Probltm.< 10' I Sentence Structure Comm. QUf!SI. Slatemelll 'V' 'Q' 'v,l, '1'1' Olher J Problems Conn. VX QX i I VXY QXY II" XY VS do XY i S QVS QxrZ i • nd halO! !at c s lIJirQ{ Other (+ ) NP VI' Initiacor Complex Coord SV 11 sc/o Nea X Clause V C/O AX Otller '1 : DN Ad)N NN PrN Phrase Word '~I VV 1 V pari "10' '" IntX pi Other -cd LX_ '_SON':. '-'_ X, V:VP 'J_!. + 9f!..:~pJl ~ '_A;..A!_ - ~ I SVCIO 35 VCjOA DAdj N " Cop , -cn SVA VOdO; Adj Adj N Au. -r, 3. '1. NoIXY Other Pr DN I#- Pron I N Adj N Other IlIOn n't 'cop X!..: ~~p.ltx! ~:~p_.~~ XY ~ C/O;~PJJ:: .:. ~!,P ~ SVC/OA AAXY fNPrNP No,V I SVO.,O, Olher I Pr DAd) N Net X eX 2 Au ... XcX Olher 'au ... Coord . I I· Paslmod. I I' clause -cs, Subord. I I· Claus<:; S Paslmod. I phrase I -Iy Clause: C/O i I Comparative Clause NP VI' Clause Passive Pron Adjseq Modal Concord Complement Oct N irrci Ten", A position I V irrea - !,,_ordo: - -l Other Olher f--D_'_·sc_o_u_rs_t' ______________ -l Syntactic ComprehtllSi01l ~- A Connectivity " Commenl Clause ,1m-, Style f~ Emphati<: Order Other u;~ I Total No. IbS Mean No. Sentences 0·; Mean Sentence l·S Sentences Per Turn Length i I iC) D. Cry",I. P. Fletcher. M. GarlTllln, 1975 Un;v ... ily of Readint Profile Chart: Mr .r, third rrotilc The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. c c :;m-- ,. S, "mbi , ... Others " ! i 10' I " ; Major =-~-r-=-_t-I ______ ------------------, " r---:-- '...,c .!i Exc/. a o f! c: J! ,v· ' ' · '" ' ' ' , rm II" XY z t l . " MJ ": "! e A II e "u. ~:~P ~~ xr CIO;~PJJ::..:.. ~~P fJ-. " " l UJl . o t ... t 08 I . I ' -l Ol l i , .1 , rma ;v."il l i 6 The linguistic analysis and treatment of language disorders Before discussing the role of LARSP in relation to pattern evaluation, remedial goals and remedial procedures-the three stages not dealt with in chapter 5- some general remarks about the relationship of linguistics to language disorders are necessary. We hope it is clear from what has been said so far that linguistics has a powerful but restricted role to play in the analysis of language disorders. It can provide a tool for analysis, and a principled procedure, but it is no panacea. It should be remembered, in this connection. that the linguist is very much in the hands of the remedial practitioner, T. In all but a few fortunate research-cum clinical institutions, he is totally dependent on T for his data. and he must at all points liaise with T if optimum use of his skills is to be achieved. This means that one of his first tasks must inevitably be to come to terms with the views, assump tions, techniques, and above all terminology ofT. We have now had an opportunity to examine the therapeutic terminology of language disorders over several years, and on the whole we find ourselves critical of it. We find it inadequate for use as a descriptive framework with which linguistic analysis can be correlated, due to its being too vague, inconsistently applied and impressionistic. We were ourselves misled for a while into thinking that categories such as 'infantile speech'. 'specific language disorder', 'functional language disabil ity', 'child aphasia', 'child dys phasia', 'delayed language development', 'minimally brain damaged', 'expressively disordered" and so on and so forth were discrete, reasonably precise categories. We have learned to know better. We no longer feel that they have even operational value for a linguistic typology, and we have ceased to work with them. We are not competent to argue for or against their value in the wider field of abnormality as a whole, but from a linguistic point of view. it is clear to us that some alternative classification, based on linguistic characteristics, is much needed. Part of the underlying difficulty is an uncertainty among researchers and practitioners as to whether there is such a thing as a distinct language disorder. independent of other considerations, especially of a psychological or neurological kind. Here the views of Piaget have been particularly influential, language develop ment being viewed as dependent on the prior development of cognitive operations (see Piaget 1970). This might well suggest a causal relationship between a language disability and some underlying cognitive deficit, and there is some evidence to suggest that linguistic symptoms we encounter may be the result of a generalized deficit in representational behaviour. Morehead's preliminary analyses (1972, 6) 'support the hypothesis of a general representational deficit in linguistically The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 110 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY deviant children' (though unfortunately he gives no details as to what kind of children he studied apart from saying that they had 'linguistic retardation'). Morehead and Ingram's language deviant children took twice as long over symbol matching tasks as normal children (I973), and Morehead and Johnson argue (1972, 152): 'the less advanced the child, the more necessary it is to coordinate operative knowledge with early representational knowledge such as deferred imitation, imagery and symbolic play prior to direct language training'; and they suggest that 'therapy for pre-syntactic children should begin with those aspects of representational behavior which precede language.' (See also Eisenson and Ingram 1972.) This is something which we see being done regularly in clinics and schools, in fact. However, it does not add up to a view that all language disability is merely a reflex of some other disability. Or-to put it more cautiously-there is nothing to be gained, and a great deal to be lost, by assuming that it is so, and concentrating on the assumed nonlinguistic impairment to the exclusion of linguistic considera tions. There are many arguments which support this emphasis. A rather basic point is that our knowledge of cognitive functioning is, jf anything. much less certain than our knowledge of linguistic structure and use. To pass the responsibility for remediation over to a 'cognitive therapist', therefore, is unlikely to be of specific practical value other than in relation to certain very general processes. The same applies to perceptual 'explanations" (see the critique in Rees 1973). For example, Fraser and Blockley (1973) take the view that the underlying cause of language disability is a defective appreciation of relationships of space and time; and that if this is attacked and improved, language proficiency will improve as a natural consequence. While the hypothesis is not implausible, the 'comprehensive and coherent theory of perception' which they seek (40) does not exist: as they themselves say, 'We do not know enough about the normal development of perception to be able to devise a programme, let alone how it would apply in the case of the perceptually disordered child' (43). A second argument is that while in the very early stages of language development it may be difficult to disassociate linguistic from nonlinguistic deficit, as the child gets older and certainly with the adult-a much clearer distinction emerges. There are numerous areas of linguistic structure which have but an indirect and remote relationship to cognition, e g. concord rules, collocational rules, some word order patterns (e.g. negative placement. question inversion), and these are often areas of error. It seems to us that no amount of cognitive training is likely to improve ability in these areas, and that the case for a specifically linguistically-principled therapy remains. Moreover, as many of these patterns are acquired fairly early, quite a high proportion of children are characterized by problems that seem to be of a specifically linguistic kind. In addition, there are many Ps whose sole deficit, so far as can be established using available assessment techniques, is in language. In all other respects-auditory, motor, neurological, psychological. social, psy chiatric-results are within normal limits. For such reasons, reinforced by a dominant view within human semiotic studies which sees linguistic behaviour as being of a fundamentally different kind from nonlinguistic behaviour (cr. Hockett's notion of design-features for language, and his particular focus on duality of The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS III structure (Hockett 1958, Hockett and Altmann 1968, Lyons 1973), we insist on retaining the notion of an independent language disability, which has to be assessed and remedied within a linguistic framework. Whatever the correct position in these matters, it should be evident that progress in the understanding of linguistic deficit will only be achieved in the context of an accurate and consistent theoretical and descriptive account of language develop ment; and in this respect, we have found many of the explanations of this deficit in psychological terms to be unsatisfactory. This claim needs detailed support, and as a case in point we cite Fraser and Blockley's study (1973) whose main hypothesis is mentioned above. We accept that if there is perceptual disability, then this must be attacked before comprehension and production of speech is likely to develop normally. But the implementation of their theory involves contradictory linguistic considerations, the status of language in their methodology is unclear, and their analysis of P's language behaviour contains misleading assumptions, when looked at from the point of view of the above chapters. Concerning their theoretical position, they claim that their work is based on a study ofthe literature on language over the past 200 years; but they refer only to the more speculative, philosophical writings: no technical works of contemporary linguistics are referred to at al1. 1 What they produce, in fact, is an amalgam of linguistic notions from different theories and stages of development within theories, which is inevitably inconsistent. The following sentence contains at least four incompatible theoretical assumptions, for instance: 'the child has a "schema" of speech when he can transform the elliptical ungrammatical colloquial utterances of his parents into "objects of knowing", into the "kernal [sic] sentences" which obey the universal rules of language and which carry meaning' (39). On the one hand, they say their work is close to a Piagetian psycholinguistic view of language as an integral part of cognitive activity and development (52); on the other hand, their entire discussion is couched in generative linguistic terms, which are in the last resort fundamentally incompatible with a Piagetian theory (cf. Sinclair-de-Zwart's (1969) critique of Chomsky). Then within generative grammar, 1957 notions of kernel sentences are made to exist alongside 1965 notions of transformation and 1970 conceptions of deep structure in terms of generative semantics. As a result, the basic claims of their approach remain very obscure, e.g. that their programme is 'based on the theory that the remedying of the child's perceptual disorder enables him to relate the surface structure of speech to the deep structure' (27). 2 Concerning methodology, the authors of this approach provide little detail about the role of language in their work. They claim that their approach is nonverbal, that is 'we do not treat through the defective medium' (51). But the one child they give us a detailed account of, Simon, had had considerable language contact and therapy before 1 The nearest is an extract from Chomsky'S Language and mimi, and a 1921 introductory work by Sapir. 2 The situation is not helped by the absence of definition of two central concepts, the' "psychic distance" from language', and the' "mental infrastructure" of language' (11), and the standard confusion (cf. p. 5 above) between 'grammatical relationships and syntactical [sic] organization' (10-11), The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 112 THE GRAMMATICAL ANALYSIS OF LANGUAGE DlSABIUTY their programme started, and it is left quite unclear, in the discussion of the visual and auditory sequencing tasks which constitute part of the programme, how much and what kind of language was used to him or around him in the instructions and discussion of the tasks (22-3). It seems unlikely that the programme was carried through in complete silence, and no account seems to be taken of the language contact he was receiving outside the clinic, where he was being seen for one hour each day. In this respect, attributing the improvement obtained to the perceptual hypothesis seems premature. They put the questions: 'is it reasonable that the medium of remedial education should be the one in which his disability lies? Can a disorder oflanguage be remedied by repetitive practice in the medium of language l' (36), to which they clearly anticipate the answer 'No'. But a better answer is 'It depends', as the question contains an assumption which the above chapters argue is incorrect, namely that language is a single, homogeneous structure. If one takes the view that there are levels of syntactic structure, however, then it is perfectly feasible to concentrate on certain aspects of syntax while ignoring others, right from Stage I. We have seen children, apparently like Simon, whose progress had been slow because his therapy had concentrated on the wrong areas of syntactic structure; focussing on the neglected areas produced dramatic improvement. There is certainly some evidence in Fraser and Blockley that they have underestimated the child's linguistic ability, e.g. they say (19) 'He learned the actual action words very quickly but could not learn the structure', yet cite girl running as one of the utterances used, which has excellent structure at clause level; 'he had little idea of word order in sentences', but they illustrate this by That boy sitting -is, which is deviant in terms of verb phrase structure only; scissors in there is called a 'bizarre' construction (19), whereas it is an expected Stage II pattern; elliptical speech is considered ungrammatical (39, cf. p. 13 above). And once again, we are reinforced in our belief that linguistic reasoning and technique provide an indispensable element of any remedial procedure in this area. The one thing that has hit us forcibly since we have begun to work with the language disordered is how heterogeneous a population they are. We have never found two children or adults who are the same in all salient respects. It is quite likely that the relation of cognitive to linguistic disability will be various, displaying different kinds of interaction, severity and dependence. We are therefore not in favour of any narrow attack on a disability based on a single theoretical principle. The more broadly based the attack, using a range of linguistic and cognitive strategies, the more likely we will be to hit the target. In no sense, then, do we wish to play down the importance of seeing language within a cognitive perspective, but neither do we wish to minimize the role of independent linguistic analysis and evaluation, until such time as other research shows this to be redundant. In this respect, then, there are three tasks which remain to be done, using the data of LARSP: pattern evaluation, statement of remedial goals, and statement of remedial procedures. A discussion of these constitutes the remainder of this chapter. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DJSORDERS 113 Patterns and goals This is a comparative exercise, in which the pattern of syntax established in a sample is compared with other samples of P, or other Ps, and ultimately with the expected pattern (in terms of age) for normal children or the norms of the adult language. Detailed studies of the use of LARSP in this way are provided in chapters 7 and 8. Here we simply wish to point to certain patterns which have emerged sufficiently often in our work to make us suspect that they will ultimately turn out to be of typological significance. However, in interpreting a profile, the limitations of our sampJing proCedure (see chapter 5) should be remembered. Perhaps the most important point to bear in mind is that the positive information it contains is always more directly useful and reliable in defining a pattern than the negative (i.e. the gaps): the presence of a score is a positive indication of ability, whereas the absence of a score may mean only that the sample is biased. Pattern 1 A profile presenting a clear case of 'pure' delayed language would be very similar to the one given at the end of chapter 5. The distribution of structures is even across the page, and the child is producing normal quantities of speech; only the age differential is out. The goal here would be to increase the frequency of structures at the next Stage of development, maintaining a balance between levels of structure. It would be hypothesized that, as a balance of structures has been obtained so far, this is likely to continue, and that unless evidence of a specific deficit emerges, all that needs to be done is to widen the range of speech situations in which P can interact and use language. Given increased motivation and opportunity, one would expect progress without structured therapy. Such profi les are not common in clinical therapy, but are more to be found in remedial education settings, where one talks more of 'disadvantaged' backgrounds. The cases of delayed language generally seen in speech therapy clinics present much less of a balanced picture (see patterns 2-6 below). Patterns 2-6 These comprise the majority of cases, where a general sense of language delay is complicated by total gaps in structural ability. These gaps, or areas of specific weakness, characteristically appear (0) at sentence/clause level. (b) at phrase level, (c) at word level, and (d) at discourse level. Pattern 2 This is the most common kind of delay, in our experience. Isolated words and a few phrases may be produced, but there is a lack of any coherent rdationships between them, an absence of sentence patterns, and a high proportion of ambiguous cases. The disorder pattern may be more dramatic if there has been a focus on isolated words and parts of speech in previous therapy or parental interaction. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 114 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY In these circumstances, there is only one initial goal, to establish a solid foundation of clause structure. This means ascertaining in the first instance whether P has the ability to use Verbs-for as we have seen (p. 44), these have a determining role in the development of clause structure-and then systematically building up clause structure patterns (see p. 121) to match the level of phrase structure complexity already achieved. Remediation continues by trying to preserve a developmental balance between clause and phrase structure levels. (Cf. Morehead 1972, who also points to the tendency for Ps to expand phrase structure before clause structure in the early stages.) Pattern 3 This is a pattern which often develops after therapy of pattern 2, but which is also common in an mitial assessment. A reasonable clause structure is established, but phrase structure is weak. There is often a one-to-one relationship between elements of clause and elements of phrase structure, e.g. he kick ball. Often, the only exponent of S is a pronoun. Attempts to increase the complexity of the Subject or Object Noun Phrase lead to a loss of sentence organization, e.g. the man ball (V being omitted), kick big ball (with S being lost). T's aim here must be to slowly build up phrase structure, while reinforcing the pattern of clause structure, beginning with postverbal phrases (NPs as C/O develop before NPs as S, see e.g. Ingram 1972c, 72). Difficulty in maintaining hierarchical organization has also been noticed for adults, e.g. Myerson and Goodglass (1972, 50) say there is 'a correlation between severity of aphasia and the number of specific types of distinction a patient can mark in each phrase in his base structure'. Pattern 4 No word-level development is a pattern which would be evident only if P has generally reached Stages III or IV. Clause and phrase structure would be satis factory, but there would be a much reduced number of inflections, and compound words would tend to be absent. The aim must be to introduce inflections systematically, and this means bearing in mind the following points. (a) Inflections group themselves into systems (singular vs. plural, present vs. past, comparative vs. superlative), and should not be introduced without making clear the overall system of which an inflection is a part. The same principle also applies to grammatical words, such as the, of, to, which are also often absent in this pattern, producing the familiar impression of 'telegraphic' speech. Thus, the plural -s must be introduced in contrast with the singular, the in contrast to a and 'zero', and so on. It is fallacious, one must remember, to conclude that the child who says boy for both singular and plural objects is getting the singular right and the-plural wrong. On the contrary, he lacks the distinction altogether: and it is theoretically possible-though not very likely-for him to use the form boys to refer to both. (b) The cognitive problems posed by the various inflectional endings and gram matical words are of different kinds. Some are simply matters of concord, involving The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LlNGUI'STIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 115 no cognitive difficulty (e.g. 3rd person singular), others are extremely problematical (e.g. comparison). (c) Many inflections appear in different forms, depending on the context, e.g. the allomorphs of plural, possession, past tense, etc., or the variation between -erj-est, and more/most. Presenting stable contrasts is important, particularly if P has poor phonemic discrimination. 3 (For further discussion of morphological development, see Stage II of Brown 1973.) Pattern 5 This pattern displays strong word level development, with correspondingly weak clause and phrase structure, i.e. the reverse of pattern 4. This is a pattern which has from time to time been noticed in ESN children (e.g. Newfield and Schlanger 1968; cf. also Shriner and Miner 1968; Dever I 972b). Morehead and Ingram (1973, 342) suggest that inflections, being more evident features of surface structure are found to be easier for Ps whose general rate of learning is slow; certainly in their data the deviant group used some word structure features more frequently than the normal group (especially -s (number), -lng, contracted copula and possessive's). In such circumstances, the aim should be to focus on clause and phrase structure, as in patterns 2 and 3. Pattern 6 Here there is little or no recursiveness, complex sentence patterns or sentence connectivity. P may produce one-sentence utterances only, or may string a number of sentences together with little or no linguistic or logical connectivity between them. The goal here is to develop sequencing ability, using conventional story telling resources, or games, and to promote dialogue situations, perhaps using question-response games with increasingly long sequences (simplified versions of Twenty Questions etc.). Pattern 7 Here there is a normal distribution for age, but very few instances for a half-hour sample. There may be an abnormally high frequency of repetition. Such samples may indicate a disorder; on the other hand, they may simply reflect a sampling problem (e.g. the child being unwell during the session) or just a quiet child. A further sample should be taken, and questions of personality raised. In the extreme case, of course, the profile might be completely blank: P produced nothing, staying completely mute, or evincing only non linguistic vocalizations. In such I One should note the great functional load carried by lsi in English inflections. Articulatory difficulty here will inevitably lead to reduced morphological distinctions. It should be emphasized, however, that we are here speaking of standard English only. Many varieties of English (e.g. West Indian) do not have -s endings in some functions as a qui!e normal feature of their language system. For a general discussion of sociolinguistic variation of this kind, see TrudgiH 1974. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 116 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY cases, as we have already pointed out (p. 23), the role of a procedure using syntactic techniques is negligible, and those of the psychologist and psychiatrist must precede. The profile might be used in an indirect manner, as a means of assessing the kind of language used by T to P, and the importance of doing this should not be under estimated. But apart from this, when faced with cases of autism, elective mutism, and the whole range of noncommunicating person, for whatever reason, the procedure advocated in this book ceases to be relevant. Pattern 8 The main feature here is the high proportion of 'other' constructions and deviant sentences (cf. p. 28), e.g. he want drink to, kick quickly ball man. It is not common to find this in speech samples. It is more common in the writing samples of adult Ps. Cf. Myerson and Goodglass (1972, 44), who refer to patterns such as OV as being unlikely to be due to a wrong syntactic rule, but due to semantic salience: 'It appears likely that he strings components together in an order determined by the psychological importance of the units constituting the utterance.' In this respect, the output is similar to the constructions produced by deaf adults who are following the conventions of a signing system such as the American Sign Language.' With such profiles, the first thing that must be done is to investigate separately the reasons for the inflated 'other' categories, and to see if specific patterns of deviance can be detected, e.g. whether it is at phrase level, clause level, in the verb phrase only •. and so on. The nondeviant pattern can be approached along the same lines as above (patterns 2-6). Pattern 9 This profile displays a clear scatter of usage ahead of a focal point, and unexpected gaps. This is the main difference between adult and child Ps. The majority of adult sentence patterns will be in or around a given stage, but there will be sporadic strength beyond this, e.g. the bulk of the utterances may be at Stage III, but the auxiliary and modal verbs might be well preserved (Stage VI). This would be a common pattern for many Broca aphasics (cf. Myerson and Goodglass 1972, Spreen and WachaI1973). This pattern is well illustrated in chapter 8. An additional characteristic of adult disability may be noted here: we refer to this as 'incipient structure'. This may take the form either of an unexpected breakthrough in an area that T is systematically working on, or of a sudden emergence of structures which are apparently unrelated to the current concerns of the remediation procedure. In the first case, T's attention will naturally be alerted to what is happening; and often T may justifiably conclude that simply by starting to work on a particular area of language structure contact has somehow been made with a previously 'isolated' or 'blocked' ability, and it may be possible to proceed without further delay to , See Stokoe 1972, Crystal and Craig (in press). ASL and other 'natural' signing systems must be contrasted here with the 'contrived' sign systems such as the Paget-Gorman, or Seeing Essential English, which follow English morphology and syntax more exactly. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 117 the next part of the procedure. Indeed, it may be desirable to do so, in view of the danger of over-using material with the adult patient. In the second case, however, we have problems on two accounts: it is quite possible that T, whose attention is properly focused on the task in hand (say, developing phrase structure in V forms) may fail to notice what is happening elsewhere (e.g. sudden emergence, or increase in the use of, pronoun forms for S or 0), at least for some time. There is also the problem of what to do---stick by the original programme? follow the emergent structure? or attempt to compromise? An example of this difficulty is discussed in chapter 8. With such profiles, the aim would be to consolidate and build upon the structures present at the focal stage, preserving a balance between the various levels of structure. Isolated patterns ahead of the focal stage would be temporarily ignored, but used as the starting point for therapy when the later stage was reached. Pattern 10 This shows a full chart, with very high frequencies, a large number of stereotyped utterances and repetitions, and some low-level errors. Very fluent aphasic speech would produce this kind of pattern. It is often impossible to break into the flow or slow it down. The only alternative is to start from scratch, attempting to build up a new 'dialect', based on the developmental progression above. Pattern 11 The profile distribution is almost identical with the normal adult one displayed in chapter 5, the only main difference being the extremely high proportion of incom plete sentences. These would have to be examined separately to determine any underlying pattern. This profile is typical of Ps with 'word-finding' problems. It is an area about which, for the moment, we have nothing to say, and for which a syntactic remediation procedure seems of little value. Remediation procedures Once some kind of disorder pattern has been hypothesized, the aim of the exercise is to bring P down the LARSP chart in as controlled a way as possible. Exactly how the various structures are to be introduced is the question which now has to be answered, and while the linguist may be able to offer some helpful suggestions, we would emphasize that decisions concerning the choice of remedial procedure are strictly not his to make. Deciding on the appropriateness or otherwise of a line of action we assume to be the final responsibility of T, who has to take into account such questions as motivation, availability of materials, P's general per formance, etc. In this final section, therefore, we restrict ourselves to certain.. general guidelines concerning this boundary area between linguistics and remedia tion, which we have found with experience to be reasonably successful. We list them under two headings: Positil'e and Negatil'e. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 118 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Positive Faced with the problems of eliciting a structure from P, instilling some practice in its use, and planning a sequence of learning patterns, it is essential that T be aware of the many structural techniques which have been devised for these pur poses, and develop an understanding of the strengths and limitations of each technique. It will not always be the case that a structural approach to therapy is what is needed, of course. Particularly with children at the very beginning of therapy, it may be inappropriate to use structural techniques for more than a small part of the time, or perhaps even not at aJl, until T gains p's confidence and some measure of control over his behaviour. Such decisions are not ours to make; but we do advocate that T's intuitive sense of appropriateness, timing and flexibility be supplemented by an informed awareness of what structural approaches have to offer, so that a rational decision can be made. In this, mother-tongue remediation has a great deal to learn from the field of foreign-language teaching, where studies in the use and appropriateness of structural drills have been in progress for over half a century. It is however remarkable how little the remedial language professions have been influenced by this field. We have come across few therapeutic studies which have attempted systematically to take into account the methods of structural approaches to language teaching (or more recent approaches),5 or which have even been aware in general terms of what the foreign-language teaching IL2) methodology has to offer. Of course there are considerable differences between the two fields, which must not be underestimated. In particular, L2 drills presuppose a shared awareness of the conventions and expectations of communication (e.g. a sense of knowing when to respond) which may be quite lacking in the mother tongue context. Also, there is a fairly widespread dislike of 'mechanical' procedures in remediation, where the basic motivation of the learner to perform willingly a structured set of tasks may be absent, and where, accordingly, T needs to approach the use of drill techniques bearing in mind the need for flexibility and moderation. Many L2 drills are more appropriate for the language laboratory situation, or for group classroom use, and it is evident that a great deal of further study of the premisses and expectations of each drill needs to be made before one can make use of them in remediation, or decide in a given instance whether the drill concept can be used at all. But it is equally evident that the L2 literature can provide an invaluable source of suggestions about the nature of language games and structural drills, and the problems of using them, as well as reminders of general language learning principles, all of which are of potential relevance to the task of language therapy. It is for this reason that we refer now to such standard accounts as Mackey 1965 (see especially Appendices A and B) and Stack 1971 (especially chapters 7 and 8), as a perspective for evaluating the suggestions we make. Our analyses of T -P sessions have shown that inevitably T takes the initiative in the majority of interactions, and that T's stimuli can be classified into a number of basic functional types, each of which could be thc basis of structural drill. 6 One exception is the use made of Hornby's structural patterns in Guidelines (Thomas 1971). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 119 1 Imitation tasks. 'Elicited' or 'rote' imitation requires P to repeat the whole or part of an utterance, following an instruction such as 'Say what I say'. A sequence of imitation tasks, with increasing complexity of structure, may be introduced in this way. These tasks are currently attracting great interest in psycholinguistics, as it is thought by some that a subject's performance-in particular, the changes he introduces into a sentence he has been asked to imitate-can indicate his level of grammatical development (see, e.g. Slobin and Welsh 1968, and above, p. 86). But this conclusion is still tentative, and in remediation contexts its applicability is untested. Imitation tasks seem to be of limited value, in that without their large-scale use on a P, they can tell us very little about how far the child has used his own linguistic system to produce the sentences: he might be reconstructing it himself, as far as he is able; or he might be parrotting it. 2 Modelled imitation is much more useful as a teaching tool in remedial contexts. This is our term for the commonly-used technique of obtaining from P spontaneous utterances which follow T's model, as part of their respective roles in a highly-structured (e.g. game) situation. Examples are given throughout chapter 7. 3 Incremental drills, in L2 contexts, refer to cases where the learner is asked to complete a sentence begun by T, sometimes with a fixed phrase, sometimes with any construction. In therapeutic contexts, there is frequent use made of the 'prompt' (e.g. It's a blue-with expectant, rising intonation), but, as this term suggests, it is not usually used as a systematic teaching technique. We find prompts of dubious prognostic value, and of restricted use, as clearly not every part of a sentence can be readily prompted. 4 Replacement or substitution drills are much less commonly used in therapy than in L2 contexts, and we are unclear as to why this should be so, as they are of great potential value in remediation. These are drills in which a stimulus sentence is repeated by the learner with one or more items substituted. The teacher may provide the item to be substituted (e.g. T. The boy was running. He. -+ P. He was running.) or some other means may be found to motivate the substitution, such as pictures (e.g. T. I can see a car -+ P. I can see a boat.). A whole sequence of replacements can be made, in a 'chain' replacement drill or game, e.g. X can Y the Z -+ X can W the Z -+ V can W the Z, etc. 5 There are many other types of drill pattern, some of which we cite here to provide a more adequate illustration of the scope of the language teaching enter prise: none of these, however, seem to be systematically used in remediation. (The names given to each drill vary somewhat in the applied linguistic literature.) (a) Expansion drills ask the learner to add items to a stimulus sentence, e.g. T. He comes on Mondays. Sometimes. -+ P. He sometimes comes on Mondays. (b) Contraction drills ask the learner to shorten a sentence, such as by collapsing_ two sentences into one, omitting a word or phrase, or replacing a complex con struction by some shorter form (e.g. a Noun Phrase by a Pronoun). (c) Embedding drills would require the learner to combine two sentences by The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 120 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY incorporating one within the structure of the other, e.g. T. X is there. X is a Y . .... P. X, who is a Y, is there. (d) Transformational drills introduce structural changes into a sentence, such as the change from active to passive, positive to negative, or changes in word order. 6 Question-answer drills need separate attention, as the use of a question stimulus usually accounts for the majority of T's interaction patterns with P. Apart from the linguistic c1as~ification of questions into structural types (see chapter 3), T needs to be aware of the demands made upon P by the choice of one question pattern over another. For example, it is easier to respond to SV-inversion questions (e.g. with a nod of the head) than to Question-word questions, which usually requir~ some verbally specific explication. Probably the most difficult kinds of question (or question substitute) are those which are intended to elicit a general answer, e.g. What's happening?, Tell me what he's dOing, What can you see? These-'open-discussion' stimuli-are a very common type of question used by Ts in the data we have analysed. Their difficulty is that they give little or no syntactic help :to P in deciding how to answer, and they may presuppose the abi I ities one wants to teach (e.g. by demanding a verb in the response, when P may not have a command of verbs-see further below), or stimulate too wide and unconnected a range dr sentence patterns in response (as is the case with many fluent dysphasics). Handl'ing open-rliscussion stimuli is we believe one of the most advanced skills for P to learn, and he should therefore be introduced to them with caution. The alternative to open questions is to restrict structurally the possibilities of response in some way, and the main technique we use here is one we refer to as forced alternative (FA) questioning. FA questions always conform to the basic pattern Is it X or Y? Their role is to give P the linguistic stimulus required for his answer (thus minimizing non linguistic problems of recall, attention etc.) but without giving him a direct imitation task. P has to choose which alternative to respond with, and in making his choice he is forced to use his knowledge of the linguistic system. To decode an FA he has to know it is a question and has to determine the meanings of the two alternatives. He must then decide which of the two alternatives is appropriate for his response. There is therefore something more than imitation involved in this type of processing. Where the FA, as compared with a simple question, helps the child is in ensuring that he does not have to recall from long-term memory the name of the action or object he wants as he is processing his response (see Huttenlocher 1974, for a discussion of the additional complex.ity involved in recall of object or action names for production). If the right level of FA is selected by T, he should answer appro priately. If too advanced a level is selected, either he will not answer at all, or he will produce a random or echolalic response. The other strength of the FA tech nique is that it can be focused on any part of sentence structure, and can thus be a way of controlling P's ex.posure to different elements or combinations of elements. At the risk of overemphasizing this technique at the expense of others, it may be helpful to illustrate in detail how it is used, and to discuss some of the problems The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 121 which the introduction of a structural drill presents. (A similar level of detail would have to be considered in applying any of the drills mentioned above, of course.) The classical build-up of clause structure, using FA, for a child at Stage [ who can name objects but do little else,8 would proceed in six steps as follows. (We emphasize, once again, the importance of flexibility in adopting any such line of action. FA techniques, like any structural drill, can do more harm than good if used in too mechanical, rigid and unfeeling a way.) (i) Elicit Verb (using toys, pictures, etc. in the usual way), e.g. T is the 'man rUnning/ or is the 'man sleeping/ P (if he is going to answer at all) rUnning/ or run/ or sleeping! or sleep/ ... Notes: (a) Various possible intonations are shown. The rising and level variants may indicate uncertainty on P's part, as of course might other variations, e.g. in articulation. These problems are ignored here, and in what follows, for the sake of exposition. (b) Using the more natural elliptical patterns of adult speech (e.g. is the man running or sleeping) we find hinders the eliciting of the correct reponse; likewise to use pronouns (is the man running or is he s!eeping}-presumably because these structures place too great a comprehension burden on a P at Stage [(but cf. below). (c) This pattern should be used with a wide range of Verbs, making sure that the correct reponse varies between medial and final positions in the stimulus sentence. Intonation should be kept constant (rising-type tone on the first Verb, falling-type tone on the second), except where there is a strong echolalic tendency or recency effect, in which case extra stress may be given to the first Verb, when this is the desired response. In all cases, the structure must be used, once mastered, in as wide a range of natural situations as possible. 7 (d) Correct choice of the inflection is not important at this step (as it is not, in normal development), and should not be allowed to interfere with the development of the clause structure pattern. There should certainly be no tendency to correct P's utterance unless the focus of the correction is on the structure being elicited in this case, the Verb. (e) If P does not respond to a FA question, it may be repeated a number of times, at varying speeds and loudnesses, altering the order of the Verbs, adding cue phrases to focus attention, pointing to a sequence of pictures, or to the Verbs in their written fOrIn, if this is possible. There are no rules about this: it is up to the individual T how long he spends with a particular stimulus question before moving on. It is to be hoped that some variations in the stimulus will emerge as being more conducive to eliciting a response in P than others. • If even naming is difficult, then this must first be combated: but here we find conventional therapeutic techniques to be already available. We assume that in all FA questions, comprehension of the lexical items used has already been established. 7 Ingenious use of FA situations can sometimes be carried too far. We have heard one story (which is probably apocryphal) of a T who, when P fell off his chair and broke a les, asked 'Is your leg broken or is your arm broken 7'! The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 122 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY (j) When P responds fairly predictably, and with little hesitation, to sentences such as the above, T may move on to elicit further structures (see below). Notes (a)-(e) also apply to these subsequent steps. We can offer no guidelines as to how long it will take to get from one step to another. [n some children, it has taken only 3 or 4 hours to get from here to step (I'i) below; in others, regular sessions over a 3-month period were needed before step (l'i) was arrived at. (li) Elicit a, e.g. T is the 'man 'eating an ice-cream! or (is the 'man) 'eating a cake/ Pice-cream! ... Notes: (a) The ability of P to name objects is one thing; the ability to use a noun as the Object of a Verb is quite another. Before we can elicit two-element sentences of a VO type, then, we must be sure that P is aware of 0 as a functional element in clause structure. (b) As the structures to be elicited increase in elements, a FA question inevitably gets longer, and there may come a point where P is unable to retain the whole sentence. If T suspects this, some degrees of ellipsis may be introduced (indicated by the parenthesis in the example). This holds even more strongly for later steps, especially for longer structures after step (I'i). (c) Verbs which have both transitive and intransitive uses make for good con tinuity at this step. A verb like eat can be used for both (i) and (ii), whereas a verb like walk could be used only for (i), and a verb like see could be used only for (ii). (d) 0 patterns are approached before S. for reasons already explained (see chapter 4). (iii) Elicit va, e.g. T is the 'man 'brushing the floor! or 'kicking a ball/ P 'brushing floor/ or 'brush the floor! ... (fl') Elicit S, e.g. T is the man 'eating! or is the hidy 'eating! P man! or man 'eating! ... Note: If P optionally produces the verb as well, this suggests T can move straight on to the next step of elicitation. (I') Elicit S V, e.g. T is the 'man eating! or is the 'lady drinking! P 'man eating! or 'lady drink! ... (I'i) Elicit sva, e.g. T is the 'man 'kicking a ball! or is the 'lady 'eating a cake/ P 'man 'kick balll ... The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUlSTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 123 This is a schematic outline of a procedure for getting from one-element to three-element structures. In real life, of course, all sorts of other factors interfere, as chapters 7 and 8 show; but in principle we think it important to keep some such progression in mind and aim towards realizing it, despite these other factors. A similar procedure could be used for eliciting multiple-element phrase structure, e.g. T is it a roo 'ball/ or a green 'ball/ P red/ or rM 'ball/ ... T is it a 'red ball/ or a 'green engine/ P 'green engine/ ... (We have often seen this kind of elicitation being routinely used by clinicians, though not as systematically, or in as graded a way as our approach insists upon.) Once some complex phrase structures are avaiJable, an attempt can be made to get these used in clause structure. Here, though, it is important not to underestimate the difficulty of making this hierarchical 'jump', and we therefore recapitulate the above progression with a complex 0, e.g. (a) T (b) (c) etc. is the 'boy kicking the 'ball/ or hOlding the 'ball/ is the 'boy kicking the 'big 'balll or holding the 'big 'balIl is the 'boy 'kicking the big 'ball/ or 'holding the red 'balll We have found that when structures reach this level of difficulty, it is essential for T to come to a session well prepared with a range of structures that can be realistically drawn or acted out. It is remarkable how difficult it is to think up plausible structures on the spur of the moment in the middle of a session. Lastly, in this section, it is important to try to establish a reasonable spread of structures within a step, before moving on to the next step, e.g. questions as well as statements. phrase structure as well as clause structure-what we have referred to above as 'balance'. T should always look retrospectively at the profile classifica tion chart, to see what gaps remain from earlier steps, and attempts to fill these should be made at the earliest possible moment. It will always remain a matter of personal decision, though, when T feels it is time to move on, or time to consolidate further, or (as was suggested above) time to stop using the structural approach for a while. Here, there is no rule of thumb. The sense of a P being 'ready' is not something which we are able or wish to make objective. Negative In this section we point to certain lines of action which tend to cause problems for Ts wanting to develop syntactic abilities in a structured way. It is important to emphasize. of course. that these recommendations have no necessary force in relation to other aspects of T's work. where, for example. it may be necessary to go against points I and 2 below in order to make any contact with P at all. Our point is simply that such variability can detract from a structured approach. once a decision has been made to try using one. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 124 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY J Not to ask P questions which presuppose the linguistic ability that it is the purpose of T to establish. If, for example, it has been shown that P is weak: on clause structure, and Verbs in particular, T should avoid asking him 'open-ended' questions which would require the use of Verbs in order for any answer to be formulated, e.g. What's the man doing? (expected answer Verb-ing), or Tell me what's happening, which also presupposes the use of a Verb in the response. What's that Noun usedfor? is another common Verb-presupposing pattern. Tbese questions are easy to introduce into remedial sessions, and they are frequently used; but they make grammatical assumptions which need to be carefully watched. Likewise, if one asks a question which involves three elements of clause structure in tbe production of the response, one is compounding the difficulty, e.g. Why is he crYing?, What happened next? Questions such as this last (used e.g. by Lee and Canter 1971) should be avoided in any structural elicitation sequence, as they are too open-ended to be able to be brought under structural control: they could be answered by almost any type of sentence. Inverted questions (e.g. is he running?, without any alternative) should also be avoided, because of the perfectly normal tendency to monosyllabic responses, yes, no. 2 Not to reinforce the wrong answer. This is very easy to forget, especially with a P who has been making little progres~. and where managing to get anything out of him at all is a bonus. A typical example is as follows: T (pointing to a picture of a man sleeping in a bed) 'what's the 'man doingl P bed! T yes/ that's rightl it's a bed/ isn't it/- it's a nice 'bedl etc. This technique, in which T takes the response obtained and builds further therapy on it, carries with it the obvious danger that T is helping P in what he can already do, and not focussing on the area of disability. P is avoiding the Verb in this example, and T's approach also avoids it. This is a conunon mistake made by parents of language-disordered children; for example, they take their child through a picture-book, naming objects, and reinforcing only the nouns which the child may come out with. 3 Not to use baby talk. But only in the sense of not using the syntactic patterns matching the child's own level. (We have no objection to the use of 'baby' vocabu lary, e.g. doggie, quacky-duck.) Normal speech patterns should be used, though these may be of a simpler structure than in normal adult speech, as happens in normal parental interaction with their children (cf. Gleason 1973). The aim is to use structures 'slightly more advanced' than the child is using (as say Lee and Canter 1971, 318; see also Fygetakis and Ingram (972) There is some evidence that children respond better to language levels a stage ahead of them (Shipley et al. 1969). Once again, these generalizations apply only to children where one is intending to try out a structured approach, and where there seems to be a reasonable chance that some progress will be made The more 'non-communicative' the child is, of course, the greater will be the tendency to use reduced forms of sentences, and in many cases, especially with subnormality, one will often have to have recourse to speaking The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 125 in telegraphic form and in single-word sentences, using emphatic prosody But with such cases, the use of a structured approach is not likely to be particularly helpful. The dangers with not talking in natural syntax to a child with whom one is presenting syntactic structures in a graded way are twofold: (a) T will not speak consistently; and (b) the child will be faced with a world containing greater linguistic variety than is necessary. This happens particularly in talking to deaf children: a policy of speaking slowly and distinctly is not necessarily going to help the child to lip-read more fluently, for instance, as the majority of the situations which he will encounter will involve speech of normal speed. A child who has already made some progress in coming to grips with normal-speed lip movements might then find slow movements more difficult. The use of telegraphic speech also complicates the syntax learning process.s We therefore advise that when using a procedure such as LARSP the vast majority of syntactic patterns used in inter action with P should be normal-in the sense of 'possible adult sentences'. The only slight artificiality we have introduced is in avoiding certain normal adult elliptical patterns, in the interests of preserving the parallelism of FA exercises (cf. above). 4 Not to assume that errors always indicate disability, and need direct correction. It is often the case that errors can give a clear indication of where P is making progress: he is trying to use a new syntactic rule, and getting it wrong to begin with. The point has been observed for adults, too, e.g. by Sefer and Shaw (1972, 88), who see an increase in association errors as a sign that the patient's language abilities are improving (but that retrieval is defective). Nor do all errors need correction immediately. Many errors will spontaneously self-correct, if P is left to himself, and appropriate positive reinforcement given, e.g. P that 'man be goingl T yes/'that's rightl the 'man's going/. This, incidentally, seems to be the normal pattern of correction by parents, who more regularly correct in terms of the truth/falsity of a sentence, automatically introducing the proper grammar as they do so (cf. Brown and Hanlon 1970,45-9, and also Nelson et al. (1973), who show that recasting of syntax in responses that is, providing new sentences, instead of just expanding them-can facilitate acquisition (they were using a 32- to 4O-month sample». One should also note two further points in any discussion of error. (a) One should distinguish 'natural' errors from deviance: a natural error is one which is a predictable part of normal syntactic development (and to some people should not be called 'error' at all); a deviant error is one which falls quite outside the normal patterns of development of the adult language (cf. p. 28 above) . (6) The perspective of this book is standard British English. Regional dialect variation (internationally and intranationally) must be taken into account whenever it occurs. The term 'error' is out of place in any comparison of regional syntactic patterns with standard English. 8 We have heard of one case where P was making no progress with normal speech. but shot ahead when T. in desperation, was reduced to talking in a telegraphic style; but unfortunately no systematic record was made, so it is difficult to be sure of what exactly happened. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 126 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY 5 Not to vary the stimulus sentence too much. This is particularly important with Ps with severe comprehension or attention difficulties, where a slight change in stimulus structure may be equivalent to presenting a totally different pattern. What T has to be particularly careful about here is to control the prosodic com ponent of his speech as far as possible-keeping the intonation and rhythm constant (cf. p. 58). 6 Not to forget about pronunciation, in dealing with syntax. This may happen, for example, if one is attempting to build up inflectional endings in pattern 4 above. It is easy to overemphasize the pronunciation of the endings, so that the overall rhythm and intonation of the sentence becomes disturbed. There is nothing wrong with a clearer pronunciation of endings than normal, as long as the relative weights of these to the rest of the word and sentence are retained. In a different connection, it is important to remember always to interrelate P's phonological and syntactic abilities. If there is an articulation disorder, the emphasis we follow is not to focus on this until syntax has developed to some extent (see further chapter 7). Full pronunciation of isolated words without corresponding sentential ability can later be a hindrance in developing rhythmical connected speech, as has often been pointed out, where the individual words retain their full sound qualities and an overarticulated and arhythmic impression is the result. On the other hand, it is important not to let syntactic therapy go too far ahead of phonology: as sentences get longer and come to be used in a wider range of contexts, the possibility of ambiguity and unintelligibility due to phonological reasons increases. (This may often not be noticed by T, or p's parents, as they are used to interpreting the deviant articulation.) Regular, systematic checks on phonological development ought to be made throughout any syntactic procedure, to ascertain any general progress, to indicate any areas which may fruitfully be given immediate therapy, and to ensure that the phonological abilities which may be needed to expound a grammatical contrast are in fact present (e.g. the -s ending in morphology). 7 Not to try to elicit a structure without checking on comprehension first. Just as P's phonological abilities ought to be related to his syntactic, so must his semantic abilities. Our approach has been oriented towards the analysis of production (cf. p. 24). In investigating the acquisition of syntax, though. it is essential to remove interference from other linguistic levels as much as possible, and establishing semantic ability in the vocabulary used is a crucial first step. Hence, T must check that the names of all objects, events etc. are comprehended by P, e.g. by nonverbal tests, such as pointing, before a structure is introduced. 8 Not to get sidetracked. It regularly happens that, while attempting to elicit a structure X, structure Y appears. This happens particularly with adult Ps, where the attempt to 'uncover' a particular structure triggers off other structures (which are usually at the same stage of development-see for example chapter 8). It is inviting chaos, however, to follow up these incidental emergences in a casual way during the session in which they appear. We see a therapy session as an attempt to put into practice hypotheses about language development which have been carefully worked out in advance of the session. And the hypotheses a particular session tests are only one link in a whole chain of hypotheses which the analyst The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. THE LINGUISTIC ANALYSIS AND TREATMENT OF LANGUAGE DISORDERS 127 is in the process of working through. It is aU too easy, we find, to become enthu siastic when an unexpected structure emerges, and to immediately spend time on it; the danger is that in retrospect it will be seen to have disturbed the generallogic and ordered progress of the approach, possibly making therapy decisions about future grading of structures more difficult As far as possible, we try to get through in a session what we plan to get through. One should always expect the unexpected -but it can be followed up next time. 9 Not to leave too long a gap between sessions. We do not know what the optimum amount of exposure is for syntactic remediation work, but we feel that there is little point in attempting a systematic procedure unless P can be seen for at least one hour a week, and ideally for two or three. For Ps who may be seen once a month or less, there is no point in attempting remedial procedures of the kind introduced in this book. 10· Not to leave too short a gap between sessions. Obviously, there must. be time left between sessions to enable the results of one's approach to be evaluated, and new hypotheses fonnulated. It is not necessary to do a fresh profile between each session-one every month or so might suffice, as a general check on progress. But specific decisions about the choice of structure to be introduced, consolidated or left need to be made session by session. If sessions are left to pile up without any analysis, the point of our procedure becomes lost. 11 Not to rely too much on home practice, without prior training of the contact. Making use of the parent or spouse of P is sometimes suggested as a way round the problems cited in 9 above. But we find that unless the kin are given careful tuition in the kind of thing T is trying to do, they can be a hindrance as much as a help. It is not easy for people with no linguistic training to appreciate the abstract nature of syntax, and the multifarious relationships between patterns that are involved. Training sounds or improving vocabulary are relatively simple tasks for family to get involved with; but syntax is very difficult. A typical error, for example, is to overdrill; to pick on a pattern, drilling it rigidly, and accepting as correct only a small range of possible responses (e.g. disallowing elliptical responses, and insisting on a 'full' sentence each time). With proper guidance, this difficulty can be removed, but it requires time and patience on T's part, and a willing and intelligent home contact, to be successful. Many of these cautionary points, as well as the positive ones, will be illustrated by the two case studies which follow. They have been chosen from among our first attempts to use LARSP intensively, and they thus jllustrate some of the unanticipated difficulties which arose, as well as the strengths of the approach. Chapter 7 applies the approach to a child; chapter 8 to an adult. 9 t cr. also the approach of Gottsleben, Tyack and Buschini (1974). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 7 Case study of a child This account concerns a boy, Hugh, who was 3; 5 when therapy began. At that time he had an effective productive language capacity of single lexical items. From the beginning, LARSP was used systematically for assessment of P's progress and for structuring remediation. After four months of therapy (averaging two 45-minute sessions a week), Hugh had a well-established clause-level syntax; he was producing two- and three-element phrase-level structures, exhibiting phrasal coordination, and using a limited set of pronouns consistently. Despite a still somewhat impoverished phonology, he was conversing in a fashion much more suited to the second half of the third year of life than to the first half of the second year. This chapter presents in some detail the course of the therapy with Hugh over these months, and charts his developmental progress from Stage I to Stage IV. Syntactic profiles of the various stages are presented and analysed, and details of the remediation procedures which were attempted to advance Hugh from one stage to the next are discussed. It will be obvious from the account below that our therapy sessions were somewhat atypical, since there was often more than one adult present. Almost all the sessions were attended by at least one of the authors, at first as an observer, and then, as time went on, as a participant in various activities. Most of the sessions were also attended by the boy's father, who played a full part in session activities. Obviously this number of adults, all able to present the relevant utterance models to the child, is not usually available to T. This does not, however, mean that the procedures we used are not possible in the normal remediation environ ment, with T and P only: on the contrary, most of the techniques we describe can, with appropriate modifications, be readily applied by T alone. This case study was based on a close cooperation between T and a linguist. Each session was taperecorded. As soon as possible after each session, linguist and T discussed it, and planned structures and methods of implementing them for the next session. This kind of cooperation was essential to test the detailed applicability of our procedure, but it is perhaps worth emphasizing that it is neither necessary nor practicable when working routinely. Our approach can be applied by T working independently, once the general technique has been learnt. Background Hugh is the youngest ::>ffour boys, and was born at term after a normal pregnancy. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CIDLD 129 There were no abnormalities apparent in his development generally apart from the language delay. At 22 months he was admitted to hospital after a febrile convulsion which followed influenza. tonsilitis and otitis media. At 3;2 he was again admitted to hospital following a febrile convulsion, which was again associated with tonsilitis. It was soon after this that he was first referred for speech therapy. Audiological examination then indicated that his hearing was within normal limits. At the time he was first referred for therapy, he achieved the following scores on the Reynell Developmental Language Scale: Comprehension 3;9, Expression I ;6. Apart from his language, Hugh, when referred for therapy, was a normal healthy child. No other psychometric data is available for him; test situations had been avoided as far as possible because they tended to produce anxiety in the child. The initial assessment The major initial problem was understanding what Hugh was saying. Until some work had been done at the beginning on his phonological system, it was impossible to assess his syntactic ability. The first two therapy sessions were devoted to obtaining information about his sound system. by getting him to name items placed before him. and also by the administration of the Edinburgh Articulation Test (see Anthony et al. 1971). Data from these sources. and also an increase in our comprehension through familiarity enabled an accurate transcription to be made of the third session, and it was on this that the first profile was based. Table I lists some of the more common substitutions made by Hugh in his utterances during the first two sessions. It might of course be argued that, since his phonology was so deviant, a systematic approach to improve his intelligibility was at least as important for his communication potential as the syntactic inter vention described below. Ideally. one would like to work on phonology at the same time as syntax, particularly as phonological deficiencies can interfere with) productive syntactic development (see p. 126). Hugh, for example, had at the beginning no fricatives at all. At the end of six months, this situation had not improved. Since the alveopalatal fricatives lsI and Izl in English carry a high functional load in English, for signalling the plural, possessive and contracted is, Hugh's continued inability to use fricatives was bound to affect the efficiency of his syntactic system. Our concentration on syntax, however, was imposed by Hugh's reaction to any attempt to work on his phonology. He had shown in the first session that his discrimination was good, and that he could imitate any sound produced by T in isolation. But as soon as T began to work systematically on, say a fricative stop contrast at a particular point of articulation, Hugh flatly refused to cooperate. He was apparently aware of his deficiencies-later on, he would comment on them, saying 'I can't say that' when T required him to produce a sound (or a structure) that he could not use correctly-but he was not prepared to work at developing his sound system. This was in contrast to his attitude to the work on syntax, where he was usually, within the limits of three-year-old behaviour, ready to cooperate, and to take part in whatever activity was organized. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 130 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Table 1 Common consonantal substitutions-Stage I Stops Fricatives Continuants Initial p ...... b t -+ d k ...... g tS -+ d dj'-+d f ...... t:} s ...... w h ...... 2 j ...... {f} Final p ...... b d ...... g k ...... g tS ...... d dj ...... d v ...... b 6 ...... d s ...... d S -+ d Once the therapist had determined what Hugh's phonemic system was (and it was a highly restricted one consonantally (as table I indicates), with some associated vowel distortion) it was possible to concentrate on syntax. The analysis of the third therapy session appears as profile I, and a portion of the transcript is included below (transcript I) to illustrate the kind of communication Hugh was having with T at this point. We will deal with the profile first. The profile is a summary of around 250 exchanges between T and P in the course of a normal therapy session. The salient features of the session are: (a) There is a heavy dependence by T on questions to elicit language from Hugh. Over 80 per cent of all the utterances ofT to which Hugh responded were questions. In general, they were questions about objects or situations in front of the child doll's-house characters, blocks, coloured balls etc. (b) Hugh produced no response to over 25 per cent of these questions. In some cases this was because he was not paying attention, or for some other external reason. In others, the structure of the question was apparently beyond his com prehension.l In the first case, structural difficulties can be ruled out, and external considerations adduced, because he responds to a particular structure, for example what is it, appropriately at one time in the session and not at all at another. In the second case, comprehension difficulties are assumed because a particular question type is used by T often during the session, but never gets a response. One example here is questions of the what-doing type. At several points, Hugh is asked what am I doing? or what are you doing? and fails to respond. His difficulty with these structures is one we will return to later: what-doing questions have to be answered with a verb (cf. p. 124), and Hugh at this stage is having difficulties with verbs. Another example is in questions like what are you sweeping? where the question-word is the object of a transitive verb. These types of questions together produced nearly half the zero responses noted on the profile. 1 Apart from the original Reynell score, which indicated in general terms that Hugh's com prehension was in advance of his production, we made no systematic investigation of the relationship between these two aspects of his language ability. Our approach was pragmatic, keeping a close watch on the structures T used, to determine how Hugh responded to them. Structures to which he consistently did not respond were not used by T for a period of time. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A U_lyoed I Problematk 1 Unin.elligible J & 2 Symbolic Noise 3 Deviant 1 Incomplete 2 Ambiguous 8' B R....- Normal Response Abnormal Repet- Elliptical Major Full Struc- Prob- Stimutus Type T olals 'ilions I 2 I 3 4 Major Minor tural 0 lems ~ Questions "7 2. 4fr 4::., " '11 ,Olh.", 10 I , 2- 8' c Spontaneous Others Prl1blems I Mlaor '"?.., Social 17 Stere01ypes Kr, ----------IF====~============================~ I :~ r ~ i Major ~O .. u r. "i, ii £Xci, C()mm, Quest. 1€ll 'Q' i 'v' I I VX QX I VXY QXY /" XY VS '*' X Y Sentence Structure 'V,S 'N' '6 Other ., Conn. I sv i SCjO , Neg X Clause YC/O AX Other x . S:NP X • V,VP SVCiO VClOA SVA Ne.XY VO.O, Olher Slatemenl Problem. Phrase DN Ad) N NN PrN X ,C/O,NP ,- ------- DAd] N Adj Adj N PrON N Adj N VV Y pari InlX Other X • A:AP --- Cop Au, Pron L,- Other i Word, I -ing I pi -ed ..,n l. =~ r.~ ~ci --+---,-----:----+---+--+-----..... __ .. _-'-------------j lIOn >~ r.l ... -.... 00_ ->- &,'" ."R Iii!: I • hal I i NP ! Initlator Coord - - - Other - S QVS QXYZ SYC/OA ,SYO .• O, I I and I Coord. I tag , C I Subord. I • Clause, S Other, Clause, ClO I i Comparative (+ ) VP Clause Complex Passive Complement XY - V,YP XY CjO:NP XY < A:AP - -- - -- ---- - -- AAXY NPrNP O.her 1 t I .. Pron De. Other Pr D Adj N cX XcX Pas, moo , I clause Postmod. I • phrase NP Adjseq N irreg Neg V Neg X 2 Aux Olher I , VP '100.1 Tense V irreg Discourse f-------------.------~ Syntactic C()mprehell.'Ji()n A ConnectIVity il Comment Clause Ihnr Emphatic Order Other - n't 'cop au, ..,., -er , -Iy I Clause Total No. 12.'7 Stntences Mean Sentence Length /-0 @ D. Crystal. 1>. Fletcher, M. Garman, 1975 Univeroity of Reading' Hugh: Profile 1 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Problematk So epet-t---'---"'---'---1 To~ISr:it~i~o~n~s4-__ +-___ ~_-4 __ 2. ~--+-~~---4----~ 10 , x : t i l ll : D o l V t J. ---I---r---o----+--+--+---~""--,,--'-----------_j ge I tag XY S:NP V V " , : r , I s : .: V 0 t . 1 " c ei .rr ci3 C f- _is_ _ _ _r _e _______________ ~ SYlltactic Comprehell.'Jioll ;1 " l i -Iy 0, P. , i, ,.; 132 THE GRAMMAT1CAL ANALYS1S OF LANGUAGE DISABILITY Tnmseript 1 T P T P T P T 10 P T P T P T P T P T 20 P T P T P T 30 (f) P T 40 P T P T P T P T HUgh/ . look/ . 'what's thfs one/ - - bed/- 'good bOy/ it's a bed/ . 'what shall we 'do with the Iirl/ shall we 'sit her flp/ or 'Jie her down/ down (Ju]/ shall we 'make her sit/ or lie/ dOwn/--- H6gh/- down (Ju]/ down/ . yes what's that for/ - - Iirl (Ju]/ the girl/- *yes/ "'is she 'going to sit/ or He/ . lie/ hrbl UeJ lie/ yes/-- there/ . 'what a'bout grandpal . I mean dAddy/ is 'he 'going to sit/ or Ile/- - sitl sit/ yes/--- 60/ . I've 'bent his 'legs the 'wrong way/ (laughs) 'what's he doing/ - - he's Sitting/ . 'what about MOmmy/ is 'she going to sitl or He/ - sit/ rhhm/- and the bOy/ shall we 'make hlm/ . sft/ or lle/ in the *b8.thl *liel lie/ alright/ good 'boy/- - - lookl 'what's daddy/'got on there/. 'ean you./ yes/- it's 'his tie/ *lsn't itl *yesl what is it/ tie/ his tie/ yes/ have y6u got 'one/ showing toy bed puts doll on bed The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CIDLD 133 .p yes/ . no/ self-correction 50 T no/ - have you 'got one at home/ p yes/. T yes/ *wMt/ P * [bAIY.!j1I1:m 1/ T a. bow tie/ D no/ a little one/ T a little one/ - p rill T a 'little whAt/ D he 'calls everything. 'little/ he he . des'cribed 60 it as baby/ T riun/ a little 'tiel p yes/ T and 'daddy's 'got a. big/- -- p 1 syll. -- 3 sylls. T 'what's tMt/ pointing to cooker 'what's tMt 'Hugh/-- p 2 sy/ls.- - 3 syl/s. T 'what shaH we put in the 'oven! shall we 'put a cakel or a pie/ - 70 P pie/. T a piel p yes/. T 'who's going to make the piel - p me/- [mAn] T hril/ p mel T mUmmY/ . p no. /me! . T you! . 80 P m/ T whO's 'going to 'make the 'piel . p mel . -.---~~~---.-------~--.----. (c) There is a large number of unintelligible responses from Hugh on the first profile. This is partly due to his phonological deviance, and partiy our relative unfamiliarity with him at this time. (d) The number of minor sentence responses, always either yes or no, is also high. The phonetic realizations of these lexical items were un] and [nAm] respectively.2 • The phonetic sequence [pI] also appeared as a tag on the end of identifying responses. For example, at one stage in the session this interchange occurs: T what's that one/ P 'chair [pI]/ The tag occurs widely in the first ten sessions and then gradually disappears. It is never clear what The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 134 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY As far as can be ascertained from the transcripts, Hugh's minor responses are in most cases appropriate to the context. In some cases it is impossible to tell, and in others they are definitely not appropriate, possibly signalling some comprehen sion difficulty. These extracts give examples of the three kinds of minor responses: (i) appropriate T whO's 'that/ - - is 'that Mummy/ P not T 'who is it / . I 'think it's baby/ The context here reveals that Hugh responded appropriately by denying that the item was mummy. Immediately following this, however, comes an inappropriate use. T is holding a baby doll, but Hugh again uses the no response: (ii) inappropriate T 'who is it . / I 'think it's baby/ P not An example of a situation where it is difficult to tell whether the minor response is appropriate or not comes when Hugh is being asked whether or not he would like to do something with the doll's house: (iii) indeterminate T 'shall we 'see inside/ P yes/ The object of an analysis of this kind is to ensure that P is not simply responding randomly with yes or no. (e) There are 50 single-element responses, which can be exemplified from the transcript excerpt: for example, in line 2 bed is an instance of 'N'; in line 31 sit is 'V'; there are also instances which have to be classified as 'Other', e.g. T 'where shall I put it/ P here/ There are in addition 4 instances of Pron, which are also entered as 'Other' at Stage I, and as 'Pron' under Stage III phrase structure. It is on the basis of (e), the single-item utterances on the profile chart, that this child is said to be at Stage I. The question to be asked now is whether this is a realistic assessment, or whether the child's ability has indeed been overestimated. Could the child, in the responses we have classified under Stage I, be imitating or even merely echoing a model which the therapist has given him in stimulus sen tences? As a practical concern, this is an important question to ask, since on it its function is. It is almost always unstressed, and it seems inappropriate to interpret it as yes in the sense of a request by Hugh for confirmation of the identification made by the lexical item to which the tag is attached. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 135 depends whether T proceeds with intervention procedures designed to implement Stage II, or whether a range of functions within Stage I are to be established first. There are several examples in the transcript excerpt for this session of a response by Hugh which consists of an item recently produced by the therapist. In line 6, for example, and again in 16, Hugh's response consists of the final item which appears in the stimulus sentence. In 23, Hugh imitates not the final item of the stimulus but the first verb of the forced alternative (FA) question. And there are various other examples in the full text of the session which could be interpreted as imitation. There are, however, counter-examples to an imitation hypothesis, which suggest that Hugh is not using this strategy exclusively. In line 2 of transcript 1, Hugh gives an identifying response bed which has not previously been introduced by T. Similar examples are girl (12) (accompanied by the []1I] particle) and me (74). And in fact many of the single-element responses throughout the session are of this type. They are all, however, (in the category terms of the adult grammar) nouns or pronouns. Hugh does not produce single-element responses which are verbs (again in adult category terms) unless it has appeared previously in an utterance from the therapist. In many cases, as in 16, he produces the final item of the stimulus sentence. In other cases, as in 23, he imitates an earlier item. In other situations in which T's question requires a verb as a response, Hugh does not give it, e.g. 27, where there is no response to what is he doing? As we have already mentioned, no what-doing question at this stage gets a response from Hugh. It is not what-questions in general which cause difficulty, but the spontaneous production of verbs. A longer example will show the extent of Hugh's resistance to what-doing questions at this time: T 'how do 'we open it/ 'what are you doing/ - 'are you 'what are you doing/ you're - pushing/ aren't you/ 'what are you doing! P [J1I] The detailed analysis of the profile of one of T's first encounters with P yields a good deal of information. In particular, his productive capacity is limited to a single lexical item, thus entrenching him firmly at Stage I. A detailed analysis of the transcript reveals the further information that Hugh's nonrepeated single element utterances are mainly limited to nouns. This information is relevant for the design of intervention procedures intended to implement Stage rI structures, for as we have seen (chapter 6), there is no point in working on SV structures, for instance. if a child is not producing verbs at his single-item stage. The establishment of a repertoire of verbs becomes an immediately priority, and if it is considered desirable to implement Stage IT structures at the same lime, T can concentrate o,p AdjN, for example, where the verb problem does not matter. There are several other phrase-level structures listed in the profile which could be used. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 136 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Stage I to Stage II: remediation The initial assessment is complete, the overall aims of the remediation are clear, and some specific pointers to Stage II structures have been provided by the analysis. The procedures for remediation, however, remain to be described. The profile does not tell you in what order the Stage II structures are to be presented to the child, nor does it help in deciding what input structures are most valuable in achieving the desired output from the child. A good deal of linguist-therapist cooperation in this project has concerned not only the order in which to establish structures at a particular stage, but also determining the most efficient way of supplying these structures to the child. A related point which is by no means trivial is the link between the linguistic input to the child and the situation-the toys the child was playing with when a particular structure was being worked on, for instance. Control of situational variables, the 'things talked about' in the session, which entails control of vocabulary items, was a relevant factor in the child's performance. In the early days particularly, we found that opposites such as big and little, and some primary colours, were the most successful adjectives. Nouns were not so much of a problem, provided that there were not too many of them exemplified at a time. So T found Hugh's performance on AdjN structures much better when the task he was involved in was putting coloured balls on sticks, where the adjectives and nouns are highly restricted, than it was when he was playing with the dolls' house, with all its inhabitants and furniture. The latter is potentially more inter esting for the child, but the linguistic models cannot be tied so closely to it, and the visual or manipulative interest seems to interfere with linguistic concentration. These matters-order of application of structures within a given stage, type of supplied structure and method of eliciting it, and situational variables-may well vary from P to P and from Tto T. However, it is hoped that the remedial procedures described for this and subsequent stages will be widely applicable. Much more information is necessary, of course, before we can be definite about whether there is an order in which structures within a stage ought to be attacked-in other words, whether there is an internal complexity to each stage. The two main remedial methods used were structural elicitation with the forced alternative question, and modelled imitation (MI). These will be referred to frequently below, but in advance of the detailed discussion some general points can be made. With a child as young as Hugh, one of the many problems faced by T is presenting the structures aimed at in as natural and unobtrusive a way as possible. I.t is desirable to avoid an over structured drilling situation in which the child mechanically repeats structures presented. We also want to keep away from the other extreme, a totally unstructured 'bathe the child in language' approach. A suitable compromise is to put the child in a position where he has the requisite structures modelled for him, and is required, by linguistic or situational convention, to produce his own approximations to these structures as a response. The planned use of the FA question (see chapter 6) played a major role in the provision of structures for Hugh early on. Questions require responses, but we found simple questions to be of no help either in modelling a structure, or in getting the child to respond. The FA question, however, The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 137 as we have seen, both requires a response and provides the child with a model of the structure he is expected to produce. The following examples from the first two sessions after the profile (nos. 1 and 2 in Table 2) indicate how T presented AdjN and V structures, and elicited them from Hugh: AdjN T P T P 'is it a big 'chair! or a little 'chair! big 'chair! is 'this a big 'chair/ or a little 'chair/ little 'chair! v T P Hugh! 'are you standing! or sitting! sitting! The usefulness of this structured elicitation is underlined by examples of simple questions and responses to them in these sessions. For the AdjN structures, T is unable to get a two-element phrase from Hugh without intermediate prompts: T 'what's thiu one! p ycHlow! T yellow- P hOrse! T 'good bOy! As for verbs, we have already referred to the difficulty what-dOing questions caused. Here is another example from the same session as the FA example just given: T 'what's he doing! Hugh! P in there! T 'what was he doing! Hugh! P in there! T 'what's the 'girl doing! (No response from P) It was obvious that the FA, as an eliciting structure, was having much greater success at this point than simple questions, and it seems reasonable to attribute its success to the model it gives to the child. Modelled imitation can be illustrated from session 4, when the goal structure was NN, possessor + noun. This session involved two other adults apart from T, one being referred to as 'doctor', and these three and Hugh took part in a game. Toy animals were distributed to each participant, and T began the modelling by saying I want Hugh's pig. The pig changed hands, and then the next adult modeHed the appropriate" structure, and so on. Finally, it was Hugh's turn: T 'now it's Hugh's 'turn! P I doctor hOrsel Not a particularly exciting or ingenious game, but one that was satisfactory for its purpose. It has the essential features of all other games in which we used modelled imitation: (a) a previously determined structure which is used in the game by all non-P participants, or simply by T if she is the only other person, as is most commonly the case; (b) a game situ/ltion which forces P to produce an utterance whose syntax is as near to that of the model as he is capable of, if he The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. / / / P / / / / ! 138 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY wants to join in the game. Often, but not always, most of the vocabulary in model and imitation will be the same, but there will be some differences. Table 2 presents in order the most important Stage II structures that we worked on with Hugh in the ten sessions between the first and second syntactic profiles. The table gives the structure we were aiming at, the method of introducing it, and an instance of Hugh's output. The Notes column gives additional information which is elaborated in the commentary. The general progression is clear enough. At first, verbs in isolation were worked on, at the same time as the first Stage II structures, AdjN. Various examples of the latter were introduced, with a restricted set of lexical items: the adjectives big/little, colours, and nouns like chair, table which referred either to dolls' furniture or to items in the room. The colours were initially used in collocation with the word ball, since the task Hugh was engaged in was to place different coloured balls on to sticks. The NN structure was repre sented by possessor + noun, e.g. modelled as doctor's horse, and reduced by Hugh to doctor horse. He learned this structure very quickly. It was introduced in session 4, and at the beginning of the next session, this sequence occurred: T Hugh/ 'what's this/ P cow/ T whose 'cow/ p daddy 'cow/ T right/ 'what about this/ P dOCtor 'cow/ Table 2 Stage J to Stage II Session Structure Method Example 0/ output Notes 1, 2 AdjN FA big chair V FA walk 3 AdjN MI green ball V FA lie 4, 5 NN MI doctor horse (a) intransitive verb in response to simple question SV MI (b) produces SV sequence PrA FA up there 6 V MI jump SV MI (c) imperative/progressive confusion 7, 8 SV MI/FA sheep running (d) intransitive verb PrN FA on bed (e) potential D? 9, IO VO FA mending car (f) transitive verb In addition to these structures consisting of two elements of phrase structure, Twas also dealing with the verb problem in the first few sessions, using FA, and getting satisfactory results. There were also single instances of V and SV responses in response to simple questions. Encouraged by Hugh's use of two-element structures, and his improvement with verbs, we decided to use MI to elicit SV structures from Hugh. We used toy animals p~rforming actions described first of all by T and then by Hugh-the pig is drinking the water, for instance. When it came to Hugh's The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 139 turn to describe what his animal was doing, however, there was no response. It may have been unfortunate to choose SVO structures as the model, assuming that Hugh would reduce them to SV; but even when SV structures were modelled, there was still little success, and so T moved back to phrase-level, with PrA, a preposItIOn adverb sequence. 3 We surmised at this point that perhaps more attention had to be paid to single verbs in isolation before we returned to SV structures, and so in session 6 impera tives were modelled using M I. The imperative routine had T telling one of the other adults in the room to stand. sit, walk, jump. hop, run, crawl and (very necessary) stop. Hugh then took over the, for him. very enjoyable task of manipu lating an adult in various postures around the room. This session worked well, and so T switched to SV again. using MI with sentences like my pig is jumping. When it came to Hugh's turn he did not respond. Then T saw Hugh looking at his reflection In the rr.irror, and improvised: T 'that's Hugh 'there! - look! 'can you 'see Hugh! 'Hugh wave! Hugh 'wave;' daddy 'wave! doctor 'wave; - Hugh 'wave/ you 'tell them to 'do iti That T took advantage of his attention wandering to model SV structures, but this time consisting of a subject - imperative verb form. The sequence continued: T 'tell daddy/ P 'daddy - waving! Hugh attempts an SV structure, with the pause indicating some uncertainty about it, presumably due to a confusion over the difference between imperative and progressive. This is understandable since (a) the imperatives he was working with so successfully at the beginning of the session did not have subjects, whereas the most recent models did; and (b) the earlier SV structures modelled for him had -ing endings on the verb. The confusion continues, as Hugh continues to instruct people using the -ing ending on verbs: P you 'waving/ D 'miss is 'waving! P 'daddy - waving/ 3 One of the most important attributes that T can bring to these sessions would appear to be flexibility. This may seem paradoxical, in view of the amount of prepianning that has to go mto these sessions, but structuring and a degree of flexibility are not incompatible. As soon as T realized that she was getting nowhere with a planned structure, or that P was tired, she moved on, either to another structure. or to an unstructured play session for the child. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 140 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY So although this interlude demonstrates the interference problems that can arise from modelling similar but different structures within the same session, it is encouraging because, once the Source of interference is identified, it is apparent that Hugh is producing an embryonic SV structure. During the next four sessions, a good deal of time was therefore devoted to two-element clause structures con taining verbs. We began with SV structures with intransitive verbs (note (d» like the sheep is running, which Hugh reduced, in his response to the FA and to MI, to sheep running. Only after this pattern seemed well-established did we move on again to three-element model structures, using transitive verbs (note (I). Models like is Hugh mending the car or breaking the car were reduced to VO, mending car. The other structure listed in table 2 which we have not yet mentioned is PrN. Note (e) refers to an intrusive schwa ([aJ) which starts to appear in D, determiner, position in Pr N structures: T 'where shall we 'put mummy! P 'on [a] bed! The item appeared in a high enough proportion of PrN structures for us to wonder whether or not it was a determiner. However, it did not appear in any other noun phrases, nor was there any other sign of three-element structures at any level, so we decided to omit it from the analysis. After the tenth session, since we had been working on two-element structures with what we felt was some success, another profile was made, again using data from the therapy session, and appears as profile 2. Stage II The profile is an analysis of 174 exchanges between T and Hugh. This is a smaller number than in the last session analysed, which is partly attributable to Hugh producing longer structures than he did before. This kind of variability is inevitable, though, from session to session, and it is very difficult to predict within narrow limits what the total number of utterances will be. The accompanying transcript excerpt is from the middle of the session. The main features of the profile are as follows: (a) The ratio of Other stimuli to Questions is higher in this sample than in the earlier one, approximating more closely to the distribution we would expect to find in normal adult-child interaction. (b) There is a much lower proportion of zero responses. This could be the result of two factors: first, the child's improved language ability, and second, Ts careful control of the complexity of the stimulus structures she uses, after the child's early indication that he is unhappy with some question-types, for instance. (c) There is a slightly smaller proportion of Unintelligible responses. This may reflect T's increasing familiarity with the child's phonological system, rather than any improvement on his part. (d) There are very few spontaneous utterances. P is as yet not initiating The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A U.aa1)'lile4 Problematic I UOinlelligtble 2.0 2 Symbolic NOt"" 3 Deviant I Incomplete I 2 Ambiguous 3 B I), I Abnormal Normal Response Elliptical Major Full frob-Repet- Stimulus Type Totals ,itions I 2 I 3 4 Major Mino 8 lems ,--- "2 1 33W ~ 14-0 Questions- 45 I--- 20 3, Others 3 : Others I , C SpootaneoWl l , , I Minor S3 Social '13 Srereolypes Problems 8-,.. '2. Sentence Structure .... Major '-!~ u u Exc/, Comm, Quesl. Sialemelft .. - c .t ~ !6 u 'v' 'Q' 'V" 'N' 1"7 Other , Problems CIl_ VJ Conn, Clause Phrase Word VX QX sv -r vc/o "7 DN VV , AdjN It- 4- -;1lI =§ SCtO AX V part " r.:b Neg X Other NN IntX 1'1 .,. I,- ~ Z-til::=, PrN Other ..,d X • S:NP X • V:VP X, CIO:NP X + A:AP VXY --~------- ..... - ------------ QXY SVCIO It- VCIOA OAdj N Cop -~ s Clause: S Postmod. 1- .." i~ phrase whot Other Claus~: CIO ·Iy CIl!.:':.1 Comparali~ ( +) (-) NP VP Clawe NP VP Clause Jnitlator Compl .. Passive Pro" Adjseq Modal Concord Coord Comp~eme"t Det N irreg Tense A position >~ V irreg Worder t~. - - - - - -- - - - ~ --- ------ - - --- --- me Other O.her Discourse i SYlftactic CompreheflSum = A Conneclivit)' it .... ~ Comment Clause Inert' Slyle 1'.-.~ Emphatic Order Other fi5~ Total No, "5 Mean No. Sentences '$ Mean Sentence '-3 Sentences Per Turn Length ill D. Crystal, 1'. FIe.cher, M. Garman, 1975 Universi.y of Reading Hugh: Profile 2 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Rest-ses i ~~ b - Others I t ·· . : l!.:':., - .... , -:; • r.>t:> ~ i atic r er . , . 142 TIiE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Transcript 2 10 20 30 T Hugh; - 'what does the fire 'engine 'dol - - [gXk]! -P T P T p T P T P T p T D T P D T p T p T P T p T P T P T 'where does it go/ hm .. 2 sy//s. -- 'where does it gol does it 'go to a fire! yes/ . 'what does it dol when it gets to the 'fire! ladder/ . there's a ladderl yes/ and 'what does the ladder 'do! 'go. up. *1 syll. / *go . 'goes 'up yesj. and then 'when the ladder's 'up/ 'what do the men 'dol - 'water. out! 2 sylls. / they ride in the 'fire 'engine/ nol water! 'comes outl ohl and 'what does the water 'dol - 4 syl/s. - stop. 'firel it 'stops the fire! it 'stops the firel yesl mhm/ - - and 'wlJat do the men 'do 'Hugh/ 'on 'ladder here/ 'on the ladder/ - yes/ 'what do they do on the 'Iadder/. 'go up! they 'go up! yes/ - - and. 'when. 'when they get to the top 'what do they 'do/--- hrill do they 'jump off/ yes/ - no! - 'come downj they 'come down a'gainl be nice/ wouldn't itl just 'going *'up and 'down! *broken 2 syIl5./ - broken 2 5ylls./ 1 syll. broken/ broken/ 40 P yes! yes! yes! T P T P T 'who do you 'think broke iti - do you 'think) broke it! no! 25y1l5. who 'broke iti -- - *'little . boy/ or a 'little girl/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 50 60 70 80 P T P T p T p T P T p T P T p T D p T p T p T P T p T p CASE STUDY OF A CHILD 143 '"1 syll./ - - 'little '"girl/ *a 'little girl 'yes/ 'little 'girls would 'break 'things like that! wouldn't they/ -- yes he 'says/ knowledgeably/ (laughter) - 'where's the ladderl - there/ is 'that the Bidder I yes/ it's 'only a little 'ladder/Isn't itl yes/ 'where's the hose/ - for the water/ 2 syJls. / thereto on thereto yes/ - 2 sylls./ - - 2 sy/ls. ladder. 'come 'up 2 syl/s./ the 'ladder 'comes up there/ yeah/- 1 syU. 'what you gotl 'what are you doingl - 1 syll. 'water 'out here [J1lJ/ 2 syl/s./ 'turning the hAndle/ the 'water 'doesn't come 'out of there/ 1 sy/l. I 'think 'that's the hOse *Hugh/ *no/- 3 sylls./ 2 sy/ls. no/- no/ - 'water 'comes 'out of there/ water/ 'water [J1I] rilhm/ water/ - - 2 sylls. - - 3 sylls. 'where is the fire 'engine Hughl 'down therel 'down therel yes/ 'where is itl 'down there/ P nods high pitch range conversational exchanges, or striking out on his own. The great majority of the structures analysed are responses to some stimulus from T. (e) The number of unexpanded minor responses is still large in proportion to the whole, and the un] particle is still used, though less frequently than before. (f) The non-minor responses now divide into two categories: first, the Stage I, single-item utterances; but also, outnumbering these, are Stage II structures over a fairly wide range. At clause level, there is SY, AX and YO, and at phrase level, AdjN, PrN, VPart, IntX and PrA (classified under 'Other' on the profile). These in the main reflect the structures that were introduced into the remediation during the previous ten sessions. Some examples can be seen on the transcript excerpt: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. .--.~.~-. ---------- 144 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY (line) 19 stop fire 28 go up VO VPart (line) 48 li ttIe girl 79 down there AdjN PrA (g) During this session we have the first appearance of Stage ITI clause structures, and one three-element phrase structure. There are 4 clause structures only, and they are all SVC or SVO. The Stage III phrase structure occurs in this exchange: T and 'What's this onel P 'big - - - red 'ball! The long pause presumably indicates his uncertainty about this structure, and an earlier sequence confirms that he has not mastered the internal syntax of noun phrases with more than one adjective: T 'what's this! HUgh! P 'blue bAll! big! If this is regarded as a mistake in the word order of a single-noun phrase (which could be a matter of dispute, given that there are two tone groups here) then it is one of the very few instances noted of such errors by Hugh. (h) Word-level development is represented at this time by -ing forms on the verbs, and by some plurals. This is entirely expected for the -ing form and the plural in terms of normal development (see chapter 3), and for the -ing form because it was so widely used in our models; but we had not systematically modelJed the plural for Hugh at all. Of course, since Hugh still had no fricatives, the only way we knew he was producing plurals was when he added an extra syllable to words like horse. The usual realization for him of horse was [Ad]. This sequence shows two instances where he adds the extra syllable: T will 'that box 'go in there/ P no! nol T 'why n6t/ P [Ad;ld] in 'there/ T 'why n6t/ P [Ad~.] in/ T yes/ 'there's hOrses in 'there/ It seems reasonable to assume, as T does, that he is marking a plural here, though it cannot be established at this point whether he has been using plurals systematically. At this juncture there are two courses of action open to T, once the profile is analysed in detail. She can either take the spread and quantity of Stage II structures as sufficient, and proceed to establish Stage III structures, or she can extend the range of Stage II structures to include all those 1isted on the summary chart. The disadvantage of the former approach is that it runs the risk of holding back .the development of certain structures which are not worked on, while advancing others, so that an imbalance develops and persists. The drawback to the other approach is that if a child is advancing quickly, and has a particular stage reason ably well established, his motivation and subsequent progress may be affected by painstakingly going through a]J structures at that stage. In Hugh's case, T decided The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. onej was / ! ! ! ! ! CASE STUDY OF A CHILD 145 to advance to Stage III structures, as he had made good progress quickly. He had begun at Stage I at the age of 3;5, roughly two years behind his peers. In terms of the profile, he had now caught up six months in about six weeks, and she did not want to lose momentum. Stage n to Stage m Table 3 shows the structures modelled in the ten sessions between Stage II and the next profile taken. The format is the same as for table 2, but an extra term appears in the Method column. As well as FA and MI, there is increasing use of pictures by T to stimulate Hugh's language. The use of pictures involved either T describing a series of pictures using the appropriate structure, and then having Hugh go through them again; or Hugh simply had to describe them, without prior modelling. The pictures were carefully chosen to fit with the structures being worked on. The use of pictures, in one of these ways, appears under Method as Pies. Table 3 Stage II to Stage III Session Structure Method Example 0/ output Notes 1,2,3 X + O:NP Pics read a book (a) omission of PrDN FA/MI in a bowl grammatical words SVO Pies boy painting fence 4,5 PrDN MI under the box (b) range of prepositions (}Q Imp. V MI don't look (c) SVO AdjAdjN FA little red bus (d) occur later with 6, 7 VOA FA/MI put it in bag simple questions V010d MI give daddy the chair 8,9, 10 SVO Pies dog hitting white ball NegXY MI not a pig (e) negation problem The first structure listed is X + O:NP, one of the early Stage III structures. The notation stands for a structure of two clause-level elements, one of which is an 0; the 0 consists of a phrase which contains at least two elements. The example given in table 3 is of a VO structure with the noun phrase at 0 consisting of a determiner plus a noun. Pictures were used to elicit this and similar structures. Hugh's performance on this structure was variable, with responses sometimes having only single-item noun phrases. Hugh often responded, when VO was the required structure, with a verb with the -ing ending and a single noun, for example climbingfenee. Occasionally, he would give SVO structures when one of the pictures had been modelled using this structure, and at other times he would only respond with one element utterances, as here: T it's a 'boy jUmpingl P nbl girll 'not bbyl girl! T it's a girl 'jumping! P nbl T 'what is itl Hughl P rUnning! rUnning/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 146 THE GRAMMATICAL ANALYSIS OF LANGUAGE DlSABIUTY The picture in question showed a running girl. This sequence illustrates a technique now increasingly used with both pictures and modelled imitation. T, instead of modelling a structure by describing accurately what was going on in a picture, or in the room, would instead give a false description. The intention was to provoke Hugh into correcting her, and it often worked. We began, then, with early Stage III clause-level structures, and then attempted SVOs. Also in the first three sessions, a three-element phrase level structure was introduced, PrON. This was the easiest of the three for Hugh-not unexpectedly, since the sole additional complexity over similar Stage II structures was the systematic use of the determiner. Soon after the structure had been modelled in the usual ways, Hugh was able to use it in response to simple questions: T 'is she in the kitchen/ P no/ T where is she/ P 'in the . bedroom/ Occasionally he was able to do this with clause-level structures as well: T 'what's 'happening in the picture/ Hugh/ P 'girl- 'bouncing - bidl/ In this case T had modelled the structure not very long before. A longer sequence from this period shows how variable his control over structures can be, depending on recency of modelling and other factors: (T has asked Hugh if he has a fish, and then asked where the fish is kept). T d'you keep him in the slnk/ P no/ T in a bottle! - P nol 5 T in a bowl/- - P no/ T in a t!ink/ P no/ T in a drawer/ 10 P nol D 'he's in a bowl! on the fridge/ isn't he/ P yes/ Dwell then/. you 'should've said/ - P 'bowl on the fridge/ 15 A 'bowl on the fridgel P yes/ B where/ P bowl fridge/ Even when the PrON model is given to him here, he does not respond, but perhaps this is because of the simple question stimuli. It is only when his father (0) tells him where his goldfish is (line II) that he gives the response, and then The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 147 he omits the first Pr and D-not surprisingly, since the father's model has two PrDN structures strung together. Finally, on being asked again where the fish is, he gets quite exasperated, and the syntactic coherence of his response breaks down altogether. All grammatical words disappear (note (a» and we are left with two nouns only. Although it is not marked here, the phrase is said with a good deal of force by Hugh, and the rising tone on bowl is a high rising tone; similarly, there is a high falling tone on fridge. So, while he can at this point in remediation deal with some three-element structures in controlled situations, he can revert very quickly to two-element sequences outside these situations. (Compare his later sequence below, where the interchange outside the structured part of the session shows a greater assurance and linguistic maturity on his part.) We worked on PrDN at this point not simply to consolidate three-element structures, but also to broaden Hugh's productive range of prepositions (note (b». He had been consistently making in/on distinctions for some time, but now the models included under/on top of, behind/in front of, and beside. The game session for the modelled imitation here had Hugh describing his own position in relation to a box, as it was thought that self-orientation was easier for the prepositions he found most troublesome, like behind and in front of This seemed to be the case, and he handled the given distinctions well. The remedial procedures described so far have applied mostly to structures under the heading of Statements (and mainly affirmative statements at that). There was little evidence, however, that Hugh was producing, spontaneously or otherwise, other sentence types, particularly imperative and question. Commands were now introduced via MI, in the course of the modelling of another structure. T had placed items in a bag like little red bus, big red bus, little black cow, big brown cow, and was modelling AdjAdjN structures. Participants were required to name the item they wanted, then reach into the bag for the item without looking into the bag. They were told either shut your eyes or cover your eyes or don't look, as here: T don't 'look/ Hugh! . tell 'David/. don't 'Jook/ D 'tell him 'not to look/- 'don't look/ P 'don't look! Despite being presented with two entirely different models in the previous utterance, he used the negative imperative here correctly, and continued to do so. He also used other modelled imperatives, and then. a little later, came this spontaneous example: T 'don't look/. daddy!. 'tell him 'not to look/ P 'don't look/ T don't/ - i stopped him 'this time/ P 'put it back/ Other occasional spontaneous structures at this time were SVO, for example him got g/ol'es. The AdjAdjN structures were easily mastered via FA, and could soon be elicited with simple questions (note (d». . The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 148 THE GRAMMATICAL ANALYS[S OF LANGUAGE DISABILfTY The clause-level structures attempted next were VOA and the indirect object structure, VOPd' The first was elicited initially with FA, and then the range of sentences was extended using MI. The next short excerpt shows Hugh imitating a modelled structure: T 'what do I 'do now/ put it on his head/ tell me/ P 'put it . on [.,] head/ T 'put it on doctor's 'headl 'it's a very big 'head/ Eventually we decided not to use pronouns in models like these, and sentences like put the chair in the tin were preferred. There were two reasons for this. One was Hugh's incomplete pronominal system: for example. there was no recorded occurrence at this point of they or them, 4 which suggested avoiding plural pronouns, at least in 0 position. The second reason was some evidence of word order difficulty in relation to pronouns in 0 position when associated with VPart structures. There were few instances of deviant word order throughout the sessions, as we have already remarked, and the fact that pronouns in 0 position caused a word order problem suggested caution in using them. This sequence gives an example of deviant order: T d'you 'want me to 'lift you upl P yes! T 'what do you say/ P 'lift up mel The VOA elicitation was only partially successful, with highly variable performance by Hugh. This may have been due to the unnecessary complexity of the models. Instead of beginning with simple adverbs like here, there in the A position we relied on the PrDN structures already well established as As, and had models like put the plane in the tin reduced to plane in [~] tin or put in [~J tin or put plane in, depending on how Hugh feIt. And then at times he wouJd, in response to similar models, produce structures like put it on your head. The final clause structure listed is SVO with a two-element noun phrase at O. This was attempted because SVO was a structure he was improving at, and pro ducing spontaneously at times. This line of attack was again only partially success ful, presumably because of the complex NP. We were again trying to do too much too soon, a temptation to be avoided. Pictures were used for this structure, depicting something that could be described with a complex noun phrase at O. So models included: The boy is hitting a yellow ball. The lady is wearing a red coat. 4 An interesting sidelight on this gap in the pronoun system is one of Hugh's responses in the administration of the production section of the Northwestern Syntax Screening Test (administered routinely at this point). The item concerns the distinction between him and their. When the therapist pointed to the picture which needed the response their wagon, Hugh replied with him and him wagon. His inability to use their produces a circumlocution using a syntactic feature, coordination, which is somewhat in advance of his overall ability at this point. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 149 One of Hugh's appropriate responses came in this sequence, though he used a different verb to T in his utterance: T yes! a dog! and a white ball! come on/'tell me 'all abOut it! Hugh! P (3 sylls.) T 'tell me 'all abOut it! p 'dog. hitting. 'white ball! Finally in this series (note (e)) were two sessions on negation. This was an attempt to iron out a specific problem that Hugh had with negative statements. We noticed that he had three ways of negating: (i) No as a minor response to someone else's affirmative statement, then an affirmative of his own. For example if T said this is a red pig, Hugh would say no, black pig. (ii) He denies the assertion with a Stage II negative structure, e.g. not red. (iii) The third method he had was the most complex for remediation, since it involved him repeating an assertion, while at the same time shaking his head. An example of this appears on transcript 3, lines 42-53. Hugh is talking about the propellors on a plane in a picture. At line 43 he says me got them, but what follows indicates that he is denying that his jet has any propellers, and the transcription comments indicate that he was shaking his head when saying this. Since T is concentrating on his spoken output, she misses the headshaking cue, and the father notices it and brings it to her attention (line 54). The game T devised to try to cut out Hugh's dependence on the third method was ingenious. Participants were blindfolded when it was their turn to select items out of a bag and identify them using the model it's an X. Others then used the model it's not an X when someone deliberately identified an item wrongly. Hugh was encouraged to take a major part in telling the person he or she was wrong. The blindfold ensured that headshaking would not be a useful cue for Hugh to use in these circumstances, and he consistently used the not an X negation form in this session. It took some time, however, before the headshaking disappeared altogether. Stage III The next session was analysed to assess progress towards the aim of establishing Stage III structures, and is summarized on profile 3. We shall not discuss this profile in detail, as most of the information it contains is a straightforward development of the tendencies already noted at earlier stages. It is evident that the advance through the stages is being maintained. In addition to a very small number of zero responses, less unintelligible responses, and the first appearance on a_ profile of spontaneous utterances, the number of Stage III structures is in advance of Stage II structures, and there is a wide coverage. There are one or two un expected Stage IV structures, represented entirely at phrase level by structures The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 150 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Transcript 3 T what's the lady 'doing/ - pictures from p [kAg]/ girl/ - sleeping/ Edinburgh T it's a lit/ it's a girl/ is it/ - Articulation Test p sleeping/ . T 'what's she doing/ p sleeping/ . T sleeping/ - - - 'what are those/ p 'on a bird/ 10 T yes/ they're 'on a bird/ it's 'what the 'bird flies 'with/ 'what do we call them/ - p feather/ T wings/ . yes/ said to herself, p ·wing/ rapid, quiet D ·feathers/ T feathers/ 20 'what has your aeroplane 'got/ 'not 'only the birds have g6t them/ but your aeroplane's 'got 'two of th6se/ hasn't it/ p yes/ T 'what Are they/ p wing/ T yes/- 'makes it go 'faster/ p hOuse/ car 'park/ T a gilrage/ ·or a car 'park/ 30 P ·garage/ D he said a ·car 'park/ F ·car 'park/ he said a hUtl a car 'park/ T yes/ p 1 syl/. T it . it's not/ . well you do 'park the 'ear in there/ but it's it's a 'special 'kind of hOuse/ and you 'could get petrol 'there/ what's it 'called/ the p a . car 'park/ 40 T n6/ it's not the car 'park/ the - garage/ p garage/- two/ garage/ 'are open/ 'me 'got - tMm/ - - - 'me 'got theml . shllkes head T 'you've got those/ - rilhm/ - shakes head F have 'you got these HUgh/ p no/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD lSI T no/, he's al'ready told us/ 'it's a *jet/ p *got them/ shakes head 50 T 'those are propellers/ p 2 sylls./ peller/ 'got them/- shakes head T you haven't 'got them/ D no/ ' he's 'still 'shaking his head/ p 1 sy/l, T and what else did we 'have/ p three 'christmas 'tree/ D three 'christmas 'trees/ p yes/ 60 T what else did we 'have/ - p 'paper T ril/ P chain/ T paper 'chains/- p 'paper chain/ T and 'where did you 'have the paper 'chains/ . did you 'have them on the Hoor/ -- p yes/- T did you 'say you trod over them! 'on the 70 'paper trail the chains! - - D nO/ we 'didn't have 'paper 'chains on the floor! 'where did we hitng the 'paper 'chains/ p 'on the - 'on the . christmas 'tree/ D oh 'no/ 'not on the christmas 'tree we didn't/ - - 'what a'bout the 'decorations/ 'where did we 'put the 'decorations/ - -- P er 2 sy/l. / - in the '. 2 syll./ T *on the llinding/ D *no/ - we 'put them on the ceiling/ didn't weI - 80 P 'put them. in the ceiling/ D and 'where did we 'put the ballOons/- p on ceiling/ D *2 sylls. T *who 'blew the bal'loons up/ - p me! D did you! p yes/ F did you. did you 'blow the baI'loons up 'Hugh! or 'did you bUrst them/ - 90 P 'blow 'them up! F did you/ 'good bOy/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A B c ->- &1,' .'R Ui:!. UlWUllyse4 Problematic 1 Unintelligible' 3 2 Symbolic Noise 3 Deviant I Incomple..,' ~ AmbilJUous I i 8. Respouses Sllmulu~ Type QI!J QueSlion< I 12. i o.lhers Sponlaneous Minor S'2. ,., , 5'1 Abnormal I Normal Response Elliptical Major I i Repet-f----,---,----.---1 FuJI , Struc- Totals'itions I 2 3 4 Major:Minor tural 2. , 3 Others 1 Soda! S 2. Problems Sentence I Prob-: lems J i f- i Major ~ (;! Excl. ~I rX, Comm, Quest, VX VXY In xr do XY , S I Complex - - QX I QXY VS .y' 5 'N' IS Dlher S Conn, Clause I SV 2- sero. 3 Neg X 3 X ' S:NP - - -- SVCiO '- SYA Neg XY Z. XY S:NP V C/o. 2. AX 2- o.ther x' Y:VP --- '- Yc/OA I VOdO. , Olher XY V:YP Problems Phrase Word DN , Ad) N " NN 1 VY I -lOg V part 2. 1O lOll' pi PrN Other l 2':" 99.:t:!.P.J.. ~ ._ "':.-A!. ..,d - o Ad) N cOP) .en Ad) Ad) N 2 Au. 2. " 2. 3, PrON Pron N Ad] N S Dlher sen XY C/O:NP xr ' A:AP QVS - - - - - -P·,jP;-NP - - - - - - n" SVCjo.A AAXY Neg V I I SVo',D, OXYZ Other I 1~ tag "'" , Coord. I I • Subord, I I· Clause: S i Other Ciaus<: C/o. Comparative ( +) vp Clau,~e Passive Pmn Comple"",nl Del -- - - ,- - - Other ; Pr 0 Adj N' Neg X 'cop ,X 1 Au~ XcX , Olher aux P<>slmad. I 1 ' clause -esl Poslmad. I· .. r phrase (-) NP AdjS<:q N 'rre, - - - - VP Modal Tense V irreg - - Clause C'orl4,;ord A position W urder - -- - --- i 1-----------.--------1 SYl1lactil' Comprehli'nsiol1 A Connectivity il Commenl Clause rhn'" SlY/I! ~ Emphalic Order D.her I'-'~' I Total No. 10' Mean No. Sentences 'B3 I Mean Sentence ,. & Sentences Per Turn Lenglh -~ @ 0, CrYSla!, P. Flekher. M~ Garman. 1975 University of Reading Hugh: Profile 3 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. ' O . 'I vxr O X Y 'I s I epet-f r-- . JI, 'V O. sv SC O vc 2. . O X, v:vp \ ' ;'100 '- - - V;:";; - f V '10,0 , , MjN " 'IV 2. )mX '- ' ..'t:!.. ':. ! j 2. r N" r . O X-y--'S'-:-N"Cp--X-Y--'-v-:v=p...L- XY ('IO:NI' SVCiOA - - A-AX Y- - P,,j ;- - - \ ' SVO.,O, l . e 05. o 0 - 't > t10&1:"" t o~ .. ~I, . \: :_~:::-'_~_J_'ha_I_Ji_. __ ~ ____ Ji.O._t_he_r...LC_I_au~~~:c_,_O_r--------L--------r-------.---~-~i '!} _ + NP Inili310r Coord me t ,--- . Other cr " ------T--------1-------~ O f _ I-D_iS_co_u_r_se ____________________ -i SYl1lactil' Comp,ehensiol1 ~ ~ -:t Commenl Clause ,.," I-S-,-y..,.I/!-------------·~-------- • "" pha.ic rder Olher I' , :::!.,;:::.j.--------------------,---------'-------.---,.--.-.--.----------------1 , ean o, entences • C 3 t ,. J I 0 O. '---__ . ___________ ~ ___ __L ____________ ---'L _________ ~ _______ ___' yst l. M. . CASE STUDY OF A CHILD 153 containing and; coordination had not been explicitly modelled to Hugh between Stages II and III. There are some examples of Stage III structures on transcript 3. At line 9 is on a bird, PrDN. There is an SVO structure at 43, me got them, and at line 80, a VOA put them in the ceiling, with an appropriate use of a third person plural pronoun (there is a model for this utterance on the preceding line, however). Line 57 has tfjree christmas trees, an AdjAdjN structure. All of these had been part of the remediation procedure. There were two instances in the session of Neg XV, some instances of no plus explanatory assertion, and one instance (in transcript 3) of the perseverance of Hugh's headshaking negation. There were three instances of the copula is in SVC structures. Both examples of Aux are can't-presumably learned as a single item, and not a contraction of can + not, as there are neither instances of can nor instances of contraction in the session or indeed in previous sessions. This is one of the instances of can'/: T 'have you 'drawn a 'picture of your buggy! P nol . 'can't 'draw my bilggyl Towards Stage IV Because of the representation of SVO structures in the sample for profile 3, and its non-elicited occurrence in earlier sessions, it was thought to be the best starting point for expansion towards Stage IV. The first sessions (see Table 4) were concerned with expanding noun phrases in Sand 0 positions in SVO structures: this is what is indicated by XY + S: NP and XY + C/O: NP. At Stage IV, they refer to any three-element clause structure with a noun phrase in either subject or object/ complement position. The model structures were of the type the big horse is jumping the fence, for XY + S:NP, and for XY + O:NP, the horse is eating green grass. We began by building up to the first model in stages-Ihe horse is jumping, the big horse is jumping, because we knew that Hugh was variably imitating auxiliaries, and that he reduced structures too long for him rather unpredictably. This excerpt shows him with the shortest structure, the horse is jumping: T the 'horse. lsI - - P jiunping! T yes/ you 'say it/ P jUmping/ T nol'say the 'whole thfn~1 P 'whole thing! - 'horse is jiunpingl Apart from Hugh's little joke of literal repetition, it seems that the only item in this sequence he has difficulty with is the determiner. The next sequence, from the same session, comes when T increases the number of items in the model by adding an adjective in the NP: T the 'big horse is jiunpingl 'you can say that one for mel P nol The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 154 THE GRAMMATICAL ANALYSIS OF LANGUAGE D!SABILITY T yes! go on! P 'can't say it! D you can 'say it! T you can! P okay! T 'go on! P 'horse. 'is - jumping! T good boy/ the 'horse is jumping! the big 'horse is 'jumping! P big horse 'jumping/ As Soon as another item is added, Hugh cuts out one of the iterns he is already producing, usually the auxiliary. The only way T could get him to include the auxiliary was to shift the tonic onto it, and then Hugh would ornit the adjective from the noun phrase. This trade-off strategy to solve the length/complexity problem that was troubling him was also used for the expanded 0 structures. It is interesting to see Hugh's own reaction to the model given hirn, in the excerpt quoted. He was obviously well aware of the problem on a conscious level. (We could have used the simple present tense form in the models, instead of the present progressive-the big horse eats the grass, for example-and so avoided the length problem, or at least alleviated it. This could, however, have caused other difficulties, cf. chapter 1.) Table 4 Stage III towards Stage IV Session Structure Method Example 0/ output Notes 1, 2, 3 XY S:NP MI big horse jumping (a) influence of tonic syllable XY + C/O:NP MI pig eating the cake (b) assurance in spontaneous exchanges 4,5,6 negation MI daddy isn't horrid questions Pies what boy doing 7, 8 copula MI polar bear is furry (c) range of adjectives Apart from what seemed to be only the partial success of our structured approach towards Stage IV, there were some encouraging signs elsewhere. The first spon taneous Stage IV negative structure appeared, with a noun phrase in 0 position: me not like that pig. And Hugh was showing a new assurance in nonstructured exchanges with T (note (b)). T you've 'cut your eye! haven't you! P no! 'cut my head! T you 'cut your bead! P rruh! T where did you do that! P 'fell on the ground! T did 'somebody push youl or did you just 'fall over/ P 'somebody. push me! There seems to be much less dependence on models for control of structure, and spontaneous utterances are more similar to structures being elicited than formerly; The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 155 e.g. 'me got no 'cowboy, which is a spontaneous utterance from one of these sessions, is an XY + 0: NP structure. The next sessions, 4, 5 and 6, were devoted to the elicitation of negatives and questions. Negatives had consistently been a problem area, but there had recently been hopeful signs of improvement. Questions, in any form, had however been in short supply in Hugh's output, and jf they did appear, were generally only into nationally marked. One obvious reason for this is the nature of the therapy situation. The child in therapy becomes quickly used to answering questions-we have only to look back on the profiles in this chapter to see how important questions are as stimuli-and is rarely in a position where he can learn to ask questions himself. Consequently, when he has advanced to the point where he does have conversations, he rarely asks any questions himself. As parents and teachers of preschool children will be aware, this is somewhat unusual. Another possible factor in Hugh's lack of questions may be the syntactic complexity involved, particularly with the yes/no questions such as is the cow jumping, which require subject-auxiliary inversion. Earlier, Hugh had not mastered the auxiliary, so it was not too surprising that he did not produce questions of this type. He had however, during the first three sessions after profile 3, had auxiliaries modelled for him, though not as a primary aim of the remediation, and consequently both yes/no questions and wh- questions were included in the procedures for the next two sessions. For negation, the structures modelled were SVC of the type 'X is adj', where the adjective was inappropriate, and participants denied the assertion in their responses. After the models had been used successfully, simple questions elicited the same response. A variety of adjectives was introduced, for the ancillary purpose of widening Hugh's vocabulary range in this category; the following exchanges i1Iustrate method and results: T is dOCtor 'horrid/ P no/ T 'what d'you say/ P 'doctor isn't 'horrid/ T 'daddy is a monkey/ P no/ 'daddy not a 'monkeyj T 'daddy isn't a 'monkeyj P 'daddy isn't a 'monkeyj The first sequence is straightforward. The second reveals Hugh's performance as variable, omitting an auxiliary which is unstressed in the first model, but including it when a stressed auxiliary is supplied. The procedure for questions used pictures of children performing different actions. T modelled a question which Hugh had to (a) answer, and (b) imitate in asking T about the next picture. The question might be appropriate or not to the picture, so that negatives could be practised as well, as in this exchange: T 'js the 'boy jumpingj P nO! 'boy isn't 'jumping/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 156 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Most successful in the first session devoted to questions were immediate imitations, as in these examples for a yes/no and a wh- question: Task her/'is th~ 'boy jUmping/ P 'is 'boy jUmping/ T just ask/ 'what's the 'boy doing/ P 'what 'boy dOing/ In the next session, Hugh managed the yes/no questions with a wider range of verbs. An attempt was also made in this session to get Hugh to produce fricatives, as he was still pronouncing is as [Id], for example. But these attempts were strongly resisted by the Child, and, after some fruitless effort, dropped. The final sessions in this series were given over to the copula. SVC structures containing a wide range of adjectives were introduced, to determine whether the copula could be used by Hugh with an} less variability than the auxiliary is. Models were of the form the animal is Adj-the tiger is stripey, the polar bear is furry, the lion is jierce--and Hugh was more competent with this structure than with the auxiliary + Ring form. Any comparison, though, is vitiated rather by the difference in length between the two kinds of models. After eight sessions a further session was analysed and summarized on the profile classification chart. This was intended as a guide for future remediation, as we were at this point unsure about Hugh's progress. We had begun by modelling early Stage IV structures like X + S: NP, had advanced to Stage IV negation and question types with partial success, and had seen evidence of Hugh's spontaneous (or, at least, non-elicited) production of other Stage IV structures-coordination, Neg X and NPrN. At the same time, his performance on some Stage HI structures, notably Aux and Cop, was variable. The summary appears as profile 4, and a transcript excerpt is included also. The profile captures a further steady progression down the page, well into Stage IV. The early structures, which we had worked on in some detail, were well represented. Neither the Stage IV questions, nor the negatives, however, appear at all. The most interesting feature of the analysis of sentence structure is the wide spread appearance of phrasal coordination (see lines 45, 70, 72 of transcript) which is something that Hugh has picked up without it being specifically modeHed. This is true also for the modal can in line 65, and the AdjAdjAdjN structure len red raCing cars (line 84). In addition the adjective same, realized phonetically as [lam] by Hugh, had not been introduced into the session dealing with adjectives. Those sessions had left their mark, however, since the examples in the transcript show that he is using same as an adjective I ike furry or stripey in an SVC structure, describing an attribute of an individual noun rather than comparing two nouns: P [;n)] lorry 'same/ and [;lI)J racing. car. 'same/. 6 • We were unable to determine a consistent interpretation for the phonetic sequence [;)1)], which sometimes seemed to be ,hat, at other times simply the. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF A CHILD 157 Tnmscrlpt4 T this speedboat/ is black/ slow delivery p yeah/ - T daddy/ Is 'this 'speedboat 'black/ D n6/ . 'this 'speedbloat 'speedboat isn't 'black/ (laughs) p y~s/ T no that's brown Hugh/ p oh/ . brown/ brown/ brown/ - oh/ 'motorboat. 'go here/ 10 T 'that 'motorboat 'goes there/ . rapidly the racing-carl is red/ - slow delivery p no/ green/- T go on/ 'tell 'Doctor Fletcher/ p [:)1)] 'green/ and [;lI]] 'green/ - T 'tell 'Doctor 'Fletcher about thIs one/ p [;ll)] samet - is/. the/ samet . same [lam] [;>I]] lorry 'samet and [;>1)] racing. 'car 'samet. T m/-- 'what's [lam] 'Hugh/ 20 D the same/- p samet T oh the same/ p yes/ green/. and/. green/ T yes/ the 'same colour/ - now look/ those 'twol are the 'saffi.e cars/ aren't they/ p riun/ 1 syll. racing 'car/- - T this lear . the same 'Hugh/ slowly p yes/ 30 T it's the 'same. colour/ isn't it/ p yes/ - sAmet . c6lour/ . is/ . the/ . lorry/ T the 'same 'colour as the 'lorry/ p ril/ T yes/ p 'two motor 'boat/ T two 'motor boats/ alright/ . are they the 'same colour/ p er not T 'what are they/ 40 P red/- T 'they're. different/ aren't they/ - different 'colours/ p m/- T Hugh/ are 'those cars/ . the 'same cars/- p green 'one/ and a red 'one/ T yes/ . and Hugh/ look/ - 'this is 'a P (laughs) T Hiigh/ - look/ 'let's 'look at these 'two for a minute/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 158 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY 50 P T P T P T P T P T 60 P T P D T 70 P T P T P D T P T P 80 T P T P T P T 90 P T P T P 2 sy/ls. 'this is a racing carl m/ and 'this is a . taxi. *'car/ *yes/ . m/ 'so they're. different/ aren't they/ 'two racing 'car/ 'two racing 'cars! 'two motor 'boat/ 'two motorboats/ yes/ . are 'these the 'same colourl - - - 'what have 'you done/ ~ - you've dropped it/ that's 'what 'you've done/ - canl workl . Hugh! 'can work/ - can! workl 'yes it can 'work/ . now listen! 'let's 'play this 'game 'first! are 'these 'cars the 'same cOlour I - nol . 'yellow and green I (laughs) 'what's that! green 'carl and al - 'green. lorry! a green/ carl and a green lorry! alright! - '[dl)] 'go here/ - - - 'put 'them 'like thllt/ you 'put them like that! 'what colour's 'that 'saw Hugh/ me 'want/ a redl 'red racing 'carl. red you 'want a 'red racing 'car! - mm/-- do you 'think I've got a 'red *racing 'carl *yes/ - here! no/ - yes/ - - - that racing 'car/ is redl yes/ 'got. ten 'red 'racing 'car 'in herel t~n 'red 'racing 'cars in 'there! I 'don't 'think I h~vel - shall we 'see how 'many I have got~ - 'one therel and 'one therel - - [dI,)] litnd'rover/ [00] hril/ [dI,)] Umdrover/ 'got a 'Jandrover here/ yes/ you've 'got a landrover/ but that's not/ 1 syll. nol we're 'looking for 'red racing 'cars/ aren't wei 1 syll./ [Ant] 'red 'racing/- - one/ 'don't think there are 'any 'more 'red racing 'cars Hugh/ . I think 'that's the lot/- 2 sylls./ 'get 'out -- nowl - 'get 'out excited slowly slowly rapidly, slurred The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A UD_lysed I Problematie I UninlcliiaibJe $ 2 'iymboHc Nois~ , J Deviant I Incomplete 2. : Ambl.uous B RIISJIODMS Normal Response Abnormal I Elliptical Major Full r/I I Struc- I Repel- 1 : Pro!1- S'imulu. Type Totals ilions I 2 3 4 'Major Minor tural 0 i lems ~ Questions '7 C. 2.1 141 13 : 30 3 '- I 40-1 Olners e,.1 , to t.i S 1 10 I C Spontaneous 12. , 1 Others 1O S'o Social 50 SrereotyfX?S Prob/~ms IS. '0 J ~, Major Sentence Structure _~, u Comm, Quest. Slole_nl t::J: IS Excl, Viel £ 'v' 'Q' -y-/O 'N' 10 OtiJer 10 Problems i Conn, Clause Phrase Word I VX QX SV ~ vejO 1 DN VV I seo .a. 3 ,o,dj N I, S -In, =~ AX V part &S roJ; I Ne. X Otner '+- 'NN 3 In' X pi ! . ., i I PrN S Olher 2- a. III ::: I i 1 - 1. X ' C/O:NP 2 X • AIAP -ed X ' S:NP X V:VP VXY ----- ----r----- -----QXY i SVCiO S VCIOA 3 D Adj N' Cop Ct- -en a~; I" XY VS ; SV A I VO.,o, Ad) Adj N 1. ,",ux 1- ro:; Nell X Y 3 OliJer.3 Pr ON S Pron '- 3, · " do XY i ' N Ad) N Other.a. cii~ &en XY S NP,5, XY V:VP I Xl' <::/~I I ~'her Clause: S Poslmod I' ..,r ,,"" phrase I :f~ "hal I Clause: C/O -,., ·Iy 1Il_ Comparali¥C' ( +) ( ) NP VP Claus!! NP VP Clouse I Jnitiator Com pie. Passive Pron Adjseq Modal Concord Coord Complement Del N !rrci Tense A posilion >~ V incg Worder ro1 - - - - - -- - -- - - - - -- - - - - - - - ------ - - · '. -,., 1Il_ Other Other Discourse Synloclic Comprehension I = A CQnnectilfily /I I >- Commenl Clause ,ltf'rt' Slyle I ro f I • 'If!. EmphatIC Order 01 her Vi::!- Total No. It,..O Mean No, Sentences ·C;b I Mean Sentence ,. , I Sentences Per Turn Length I @ D. Crystal. p, Fletcher, "I, Garman, 1975 Uni •• rsity of R .... dinl Hugh: Proll Ie 4 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. UII,/ ' tt Il 1' ! :! 'V' i l If. I l ; l . . ,,60' -i ,-,." T, ,' ' . l . I· I· ! ~- ll . ! ---- , l . . l 160 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY There is a sharp increase in the number of spontaneous utterances in this session, and no zero responses at all. The proportion of Other stimuli is much higher too. Minor sentences are still well-represented. Summary Therapy for Hugh began when he was 3; 5 and had an effective productive capacity of single lexical items. When profile 4 was taken. Hugh was 3;9. Four months of remediation structured in terms of nonnal language development had taken him from the kind of language nonnally used by children eighteen months old and younger, to that more usual from two-and-a-half year olds. This chronological normalizing is of course an important aspect of language remediation for the preschool child. particularly in view of the natural parental anxiety that any language problems should as nearly as possible be removed before the child enters school. But as important as catching up in the sense that one might say that Hugh has caught up a year in four months, according to the LARSP, is the kind of improvement that is made. To move from lexical items to a point at which the main features of clause syntax have been established is a major achievement. And in addition, it seems from the evidence available that Hugh has now started to acquire language on his own, although this is not yet systematic enough for him to be left to his own devices without therapy. However. we do not yet know whether this is a general feature of this kind of remediation with a significant proportion of language-delayed children-the information is not yet available. Nor do we know, if it is a general feature, whether there is a particular stage at which it occurs. But in the case of Hugh. the emergence of this range of spontaneous language development has been an encouraging sign. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 8 Case study of an adult Introduction In this chapter we shall try to illustrate the ways in which our procedures for assessment and remediation have been applied in the case of an adult dysphasic. We are very conscious of the fact that adult dysphasics differ very widely in both the severity and nature of their language disorder, and we realize that concentrating, as we shall in this chapter, on just one patient inevitably restricts the scope of our illustration in a rather idiosyncratic fashion. However, it should be remembered that our main purpose here is illustration, not at all a general study of adult dysphasia (which would be beyond the scope of this book), and that representative ness would still elude us even with descriptions of, say, three or four different types of patient. For the moment, we are at the beginning of things, and lack a systematic typology for what we intuitively feel to be different sorts of adult dysphasia. As we have pointed out (chapter 2), we hope that our work will eventually provide a basis for such a typology, in the field of grammar. However, all we require just now of our procedure is that it prove adequate for adaptation to individual cases, so that eventually we shall be in a position to compare certain different types of disorder within a coherent framework. It may be that, in one or two places in this chapter, we have gone into some possibly alarming detail. We have some misgivings that some readers will be put off by this, and assume that our suggestions only lead to further complications, and pointless theorizing, in a job which is already difficult enough! Our reason, however, in each case was that we felt bound to be as explicit as possible. Where we took a particular decision, we wanted to set it out in such a way that the arguments, and the underlying assumptions, behind it could be examined, as fairly as possible. Quite possibly some of these arguments and assumptions are not as good as they should be; so they are opened to inspection here. Certainly, our theorizing was not pointless, since it guided our practice; our proper (and much more serious) concern now is to assess the credentials of the theorizing in this light. Most of the close discussion concerns the description of the first syntactic profile and certain points of detail that arose in the remediation procedure; we hope that the information supplied in such passages will be treated as supplementary rather than as com pulsory, especially on a first reading. We may close this introductory section with a few observations about the therapy situation as it applied in the case of the patient whom we shall call Mr. J. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 162 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY The general pattern was as follows. T saw Mr J once a week, each session being fully taperecorded and subsequently followed up, a few days later, by a discussion of the tape between T and ourselves. In this way, we were able to monitor Mr J's linguistic progress; additionally, T had final responsibility for relating the linguistic remediation procedure to other factors (such as health, disposition, family situation etc.). For the first 3 sessions, T saw Mr J as an outpatient, alone; later, she saw him at home (except for 3 semi-sessions in the hospital ward, when he was an inpatient, undergoing a haemorrhoids operation). Seeing him at home made it possible for sessions to vary a good deal in length, depending on Mr J's receptivity (sometimes taking up most of a morning). His wife was also naturally involved in such sessions, and was most willing and able to cooperate with T, particularly in the matter of helping Mr J retain the work done in one session to the next. General and medical background Mr J was born in December 1926, and at the time of his stroke was a warehouse general manager, married, with two daughters aged 10 and 4 years. In July 1971 while on holiday in France he collapsed at a friend's house, and was admitted to Lille City Hospital. His symptoms were reported as: right hemiplegia, right homonymous hemianopia and total aphasia. No predisposing cause was found. The EEG showed a slow dysrhythmia on the left hemisphere; the angiogram showed a complete occlusion of the left internal carotid artery I cm above the carotid bifurcation. At first it was feared that he would not survive, or, if he did, that he would remain beyond reach of intensive rehabilitation. But, in August he was transferred to a hospital in Reading, and a rehabilitation programme was put into effect-consisting of remedial gymnastics, physiotherapy, occupational therapy and speech therapy. In October he was discharged home, it having been decided that no surgery was possible. He continued to receive all three therapies as an outpatient. He was then able to walk unaided, with a stick, but had very little use of his right arm, and was unable to work. During this time we, as linguists, were not involved. As far as his language was concerned, considerable spontaneous improvement seems to have taken place in the first half of 1972, although it is undoubtedly the case that he benefited greatly from speech therapy at that time. The therapy commenced by working on phonation and the imitation of sounds, concentrating on lip and tongue movements. (For further details concerning Mr J's linguistic ability at this time, see Child 1972.) As his comprehension gradually improved, he went through a period of depression and frustration which caused additional problems, as also his unwillingness to attempt speech for fear of making mistakes. Many areas were worked on, in the light of results obtained from the Minnesota Test for Differential Diagnosis of Aphasia (see SchueU 1965), such as reading aloud, auditory memory and sequencing, word-finding, questions and answers etc. However, the main difficulty, noted time and again, was his inability to retain work between sessions. His first speech therapist left the hospital in January 1973, resulting in further depression; and unfortunately, there was another break in therapy when his second speech therapist also left, at the end of the The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 163 following summer. The present speech therapist started work with him in September 1973, and her preliminary assessment of his linguistic ability at that stage was as follows: Comprehension: still considerable difficulty; simple commands had to be repeated. Speech: only single words for the most part; severe word-finding difficulty; some persistent articulatory problems. Reading: inidividualletters were distinguishable, but he was not able to read and unde,:stand a sentence; written material could, however, be used as a cue to word-finding. Writing: able to write a few words (with left, non-preferred, hand only). The first syntactic profile This was established on the basis of a general elicitation session on 25 September 1973. T was newly assigned to Mr J, and wanted to discover as precisely and comprehensively as possible the extent of the language control that he had. The session was fully taperecorded, and a half-hour sample was extracted from it and transcribed for the purpose of the type of analysis set out in chapter 5. Twas using throughout the session a combination of pictures (from magazines, for the most part, especially colourful full-page advertisements) and questions about the pictures-trying to allow enough time between questions for spontaneous speech to be produced, without letting gaps of silence become uncomfortably long. We shall spend some time discussing the main features of this profile, as it differs, inevitably, in many respects from what one would expect from work with a child (chapter 7). The total number of sentences in the sample was 243,1 with a further 52 utterances entered under category A: of these, 17 were 'unintelligible' (consisting of an undecipherable noise, usually of one syllable, or, quite frequently, an alveolar click); 30 were 'incomplete' (e.g. 'box. syl/. probably for boxing, where 'syll.' simply indicated a monosyllabic undecipherable sequence); and 5 were 'ambiguous' (e.g. boy in response to what's he dOing). This leaves us with 243 sentences that we could straightaway analyse exhaustively and a further 35 which might yield to analysis SUbsequently. However, first we noted, in category B, how many of these 243 sentences came as responses to what T said (rather than as spontaneous, initiating productions on the part of the patient). We found in all just 61 responses to the questions of T and we were able to categorize them as follows: 61 normal, 7 abnormal, with 7 repetitions. Under 'normal' we recognized 5 full major types (e.g. the boy is sitting!) where there is a well-formed syntactic pattern (albeit with hesitant intonation) in an appropriate response, with no ellipsis; 19 eJlipticaJ types (with either one or two elements present, e.g. green/, pfil'ling 'it); and 31 minor sentences (mainly involving yes or no-we shall return to this type presently). In the abnormal category we found 1 instances where no 1 This is a rather higher figure for a half-hour sample than that obtained for profiles 2 and 3 (127 and 165 sentences respo;tively); see pp. 174, 175). It may fairly be considered abnonnally high for Mr J., and the very high proportion of minor sentence types seems to provide the explanation. This feature of the data is discussed further on pp. 164-5. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A ProbIemade I Uninfelliglble 17 2 Symbolic Notse 3 [)eyianl I Incomplete 30 2 Ambi.uous S Normal Response Abnormal Slimulu.~ Type ,Repet- Totals i ilions Elliptical Major Full Struc- Prob-i 4 I Major Minor tural S lems Questions " ~ 17 2. S 37 , Others '''0 I- lJ fit 11,. I, e Spoulaneous 12. 10 Il. Others Minor 11:.5 Social "S Prohlllln:! Stereotypes K~, -----------r==================================~ ~ Major ig .. ~ " u Co_. Quest. Itl " Excl. State_fit .!! .!~: " 'V' r'N' 1,.2 , 00_: Jl 'v' 'Q' Other Problems Conn, Clause Phrase Word vx QX sv 12. V Cia 2- ON " VV SCIO AX Ad) N l. .1 -inB =9.' v pan 2.c,. It .... ! Neg X Other NN I Int X I ,e~ ! pi (I)'::.i PrN , Orher 8" ! X ' SIN':,IJ_ X ,V:VP I) X, CIO:NP X + A:AP -od VXY VC/~A- -l 0 ~.lj;; -, - -Cop - - - 4 QXY SVC/O 2- ..,n ilH- I~I XI' VS SVA VO.O, I Adj Adj 1'1 Au. 'it- / 3. .. N Ne, XI' Other p, DN Pron 14-:~ , do X I' Ii) t'- , iN Adj N Other ..,n XI' • S:NP XI' V:VP Xl' • C/O:NP XI' • A:AP -- - - - -rNPrNP- - -- - - n't , S QVS SVCIOA AAXY Ne. V 2:9: QXYZ I SVO"O, Other ' Pr D Adj N Ne. X 'cop ..... eX I 2 A .. ;-A! (l)t'- XcX 2. Other ~au" and Coord. , I· Pootmod. I I - clause -al how log c Subord I I .. ~ • Clau .. : S Pootmod. 1 ..,r It~ lather phrase ",hal I Clau .. : C/O ~ii ! Compara11ve ·Iy ( +) (-) NP VI' Clause NP VI' Clau.se Initiator Complex Passlye Pron Adjseq Modal Concord Coord Complement De! N "ret! Tense A position -(0 .... V irreg Worder 1t1 - - - - - - -- - - - - - - I- -- -- - -- - - -- .- --- - .... ti3~ Other Other Discourse SyNac/jc Comprehell$ioll A Connectivlt)' i, I ;:- II Comment Claus-e lnt'rf' Sly/e ."R ,Emphatic Order Other Vi:!. I Total No. 24-3 Mean No. Sentences 0·' I Mean Sentence I .. 2-Sentences Per Turn Length © 0 Crystal. P Fletcher, M. Garman, 1975 Universlty.,f Reading Mr.l' Profile I The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. K,~---------'==================================~ ~- . © CASE STUDY OF AN ADULT 165 response was forthcoming in a situation where one was clearly to be expected. (It should be pointed out, however, that this was never the result of uncooperativeness on Mr J's part-he was in all cases simply blocking, and a number of the unintel ligible utterances came from this sort of situation, where Mr J clearly recognized that he was to speak, and was trying to do something about it.) In the 'repetitions' category we put such instances as people/ (in response to what about the people/); notice the rising intonation pattern, presumably carried over, together with the word itself, from T's question. We should note here that we found 61 to be a surprisingly low figure for category B (responses to T's questions). When we examined T's side of the dialogue, how ever, we immediately found the reason; there were only 81 questions in the half-hour sample. For the rest, T had used 29 commands (tidl mel; try it!), 121 statements (J don't think it's a string/) and 69 gap-fillers of various sorts (mM!; yes; ok) that is, 219 non-questions. To these Mr J had provided 170 responses (with 11 further instances of zero-responding). Interestingly enough, the distinction in T's language between 'Questions' and 'Others' seems not to have played a role in determining the nature of Mr 1's responses; the ratio of 81 questions to 219 non questions closely corresponds to that of 61 (responses to questions) to 170 (responses to non-questions), indicating that non-questions were just as fruitful in eliciting responses as were questions. Similarly, an analysis of the sentence types shown in the responses to questions and to non-questions reveals no significant differences. As a case in point, we may here conveniently take up the discussion of the minor sentence pattern, briefly alluded to above. The first thing to notice is the very high proportion of minor to full and elliptical major sentence types in category B (165 to 66). Great weight was being placed on the minor type, which is represented almost exclusively in the data by yes and no, in various combinations. Mr J fre quently used both yes and no, in reduplicated sentences (ah yes yes yes yes yes/yes yes yes yesj; no no no no/), apparently for the purpose of providing a well-known segmental vehicle over which prosodic contours might be established (indicating agreement, exasperation, acknowledgement and so on): we shall symbolize these repeated sequences as Yes' and Nor. Yes and no were also used singly, for affirma tion and denial; no frequently occurred when Mr J recognized that what he had said was wrong (often no in such instances formed part of the overall prosodic pattern of the utterance), though he was generally unable to correct himself. We found that the relative frequencies of these 4 types of minor sentence were as follows: Yes Yes' No Nor Category B 16 9 7 5 37 Overall 90 33 23 19 165 From this it will be readily seen that the frequency pattern in category B (responses to T's questions) is not essentially different from the overall pattern. We may conclude. therefore, that as far as the most frequently occurring sentence types are The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 166 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY concerned, it does not really matter whether T attempted to elicit speech by asking questions or by some other means. And we suspect also that the rather low proportion of questions to commands, statements and gap-filIers on the part of T is by no means unusual (especially in work with adult dysphasics, perhaps). Now we come to category C, which records the number and type of the spontaneous utterances of Mr J. It will be seen that these formed but a tiny propor tion of the total (1.2/.295), smaller even than that of category A. Moreover, of these 1.2 utterances, nearly half (5) were repetitions of some spontaneous production (i.e. self-repetitions on the part of Mr 1). For the rest, we had the following novel (nonrepeated spontaneous) sentences: 1 'too ? 'false ? 'force 2 'little girll 3 good! 4 'twenty-six 5 'thirty 6 'man 7 ?'two . 'young ? too 8 'race. results Of these, the first represented a problem in identification-Mr J was looking at a picture of Muhammed Ali, and apparently trying hard to say something about boxing-perhaps that he felt championship fights were rigged beforehand? or that boxing is too violent? Item 7 also represented a problem, in that it was unclear whether Mr J was saying two young (AdjAdj) or too young (J ntX); this was therefore counted as ambiguous. The remaining items were all analysed as elliptical I-element sentences at clause level, in the usual way (race results being logged also as NN in phrase structure). The uncertain intonation of items 4-6 and the pause in item 8 we ascribed to Mr 1's general lack of confidence in expression during this session, and accordingly did not treat these sentences as incomplete. Finally, there is the stages analysis to which we subjected the utterances we were able to analyse. Under the heading 'sentence type', we recorded 165 minor sentences and 78 major, in all. These totals, of course, included the 33 instances of repetitions that had been logged already in sections B (7 + 16 23) and C (10). This told us at once that 33/244, or 13.5 per cent, of the utterances subjected to stages-analysis were not novel. In this case, 13.5 per cent was an acceptably low figure. so we were able to take the stages-analysis pretty much at face value. Notice that these 33 repetitions are distributed, in a manner we cannot immediately determine from the profile, across the rest of the entries in the stages analysis. For the rest, we had 165 social utterances (see above) within the minor sentence type, plus 51 instances of single-element sentences, as illustrated in the following extract: T 'what are they doing! P 'talking T rhhm/ P 'man The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 167 T rilhm/ P little girl! yesj2 Here, the first and second reponses of Mr J were single-element sentences. Notice (under the heading 'Statement') that I'nouns' (42) were much more common than 'verbs' (8) in this type of sentence; this is an important point, to which we shall return below. In addition, there were 27 other sentences (both full and elliptical) which consisted of 2 or more elements and which could therefore be stated in terms of clause and/or phrase structure. It turned out that all of these fell into the category of statements. (The only possible instance of a non statement in the data was in the novel utterance quoted above 7ittle girl. On the whole it seemed wisest to regard this conservatively, as a statement which has uncertain, slightly rising intonation.) When we came to analyse those sentences which showed c1ause- and/or phrase level structure, we found that there were in all II two-element constructions: one was an instance of SV, with a single phrase-level element realizing each of the clause-level elements S and V; two were VO (with the V consisting in each case of Aux + v); and 7 consisted of two phrase-level elements. In addition, there were 11 instances of three-element constructions which had the following form: Clause level: 2 elements S V /~ /~ Phrase level: 2 elements D N Aux v In each case the clause-level element was expanded at phrase level as DN; and the phrase-level expansion of clause-level V was Aux and v. There were also two instances of three-element constructions at clause level (SVO), and one of a three-element construction at phrase level (DAdjN). Three occurrences of and at phrase-level (symbolized as c) were found in the sample: once followed by a single-phrase element, twice conjoining 2 Ns. The final point to notice at phrase level was the 14 instances of the use of some auxiliary verb-a feature which is characteristic of the three-element stage of normal first-language development. Together with this we took the 24 instances of -ing verb forms (in the word-level column), and noted the presence of a number of other word suffixes. 3 This really concludes our examination of the first syntactic profile, which we have gone into in some detail. It was necessary to do this in order to prepare the ground for a description of the therapy programme which we subsequently followed. Now, however, it is profitable to take a few steps back, and put the profile-particularly the stages analysis-into perspective. We notice at once the unbalanced distribution of our observed constructions and single-element forms: the majority of major sentence types are found under 'N' at Stage I (42/79); and when we look for sentences with more than one element, we find relatively few entries at clause level, more at phrase level, and still more at word level. This 2 = indicates a held sound or syllable, often implying some articulatory difficulty. 3 Interestingly, we found at this point just those which seem to occur earliest in normal first language development: the later, contracted forms, 's, 'cop, 'aux, and the comparative construction involving -er, had to be systematically introduced in the subsequent therapy schedule. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 168 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY general pattern cuts across the postulated normal first-language developmental sequence, but we should not be surprised at this-rather the reverse; however, it clearly highlights the areas that needed immediate attention at the start of the therapy schedule, and we found that developmental norms proved most fruitful as an indication of the way in which the therapy schedule was to be subsequently maintained and developed. The following extracts may help to indicate the nature ofMr J's language control at this stage (from session 1): T p T P T P T p T P T P T P T P 'what's this! erm mhrit/ - 'cowboys ""and "'mhril! 'cowboy 'and wrestler/ 'wrestler 'and well! . the 'horse is 'tied with a rope/ . lsn't it! a rope/ . 'ah 'yes yes! 'what are they 'doing with the rope/ - 'what are they 'doing with the ropel string/ . rilhm/. yes/ -- stringl . stringl mhril/ . 'they're pulling itl piil'lillg 'it/ . 'ah yes/'pu [whispers) "'pulling itl . "'pulling itl yes yesl T 'what are they doing/- - P boYI T "'mhril/ . P "'girll . T mhril/ 'try 'joining it up! - P 4 sylls. T 'ab hlh! P 'the 'boy is/ T goOd! . rllhm/ P 'the 'boy is - 'the 'boy is - yes/- yes/ T right! yes/ . you've got itl sitting! s/wwing picture of cowboy lassoing horse soft. slow soft The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. p T P T sing itl yes/ sitting! 1 sy/J. trY itl listen! sitting! . sitting/ P sing- T the 'boy 'is "'sitting! P *sing- T it 'won't come out/ will it! P '!he 'boy 'is sitt. *ing/ T "'ting! good! CASE STUDY OF AN ADULT 169 soft The therapy procedure: aims and techniques There was little room for argument as to what to do next, once the syntactic profile had been examined; clearly, clause-level structures had to be built up, at least to Stage III, with equivalent phrase-level consolidation, before anything else could be attempted. Then, we would be in a position to do something about the gaps in the Command and Question columns; we felt that commands and questions had priority over exclamations since they would help Mr J to use language in normal social situations, inside and outside the home, as soon as possible. At that point, we would review the situation, and work out a further stage in the remedia tion procedure. However, we felt that it was worth while spending two more exploratory sessions with Mr J before starting the therapy schedule in earnest. There were two main questions to be approached: the first was that, as noted above (p. 166), rather few of the single-element sentences consisted of 'verbs', and we saw our task of developing clause-level structures as crucially dependent on the Y category (see chapter 6); accordingly, we wanted to see how far Y elements could be elicited at all. The second reason was that we wanted to investigate the possibility that Mr J's relative lack of multi-element constructions might be the result of an inability to control multi-syllabic sequences beyond a certain critical length. There was also the additional advantage of being able to assess, fairly informally, how stable Mr J's language ability was at this stage, and hence how representative the first syntactic profile was. As it turned out, these next two sessions provided us with fairly clearcut answers to our queries. As far as syllabic sequencing is con cerned, we found that Mr J's relatively minor articulatory difficulties could not be held accountable for his lack of syntactic complexity-he was able to handle section C.4 of the Minnesota Test with very little difficulty, falling down only on the items light the lamp ([talt IQ lamp]) and easy does it ([i: zi dAd IZ]). SO we felt reasonably confident that his trouble was principally located at the level of syntax, as the profile suggested. Interestingly, the answer to our first question-how far could we elicit verbs, even in single-element utterances-put the relationship of Mr J's articulatory and syntactic difficulties in "a new perspective for us; for, while The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 170 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY it proved encouragingly possible to elicit verbs, we noted that V-function (as opposed to SIOIC function) at the level of syntax tended to increase the likelihood of articulatory problems, even over exactly the same phonological sequence. Thus, for example, the word comb as SIOIC presented no articulatory difficulty, but when Mr J was pressed to use it as a V, he at once started to lose articulatory control: T 'what is she doing/ - P yes yes/- T P T erm 'the 'girl 'is. 'comb. 'co. 'the 'boy 'is. combing/ cOmbing/ yes yes/ combing/ soft We now began to consider the possibility that even the articulatory problems might, to some degree at least, be cleared up by intensive work on syntax (i.e. just the reverse of thinking that the syntax should wait upon phonological work). Of course, we recognized even at this stage that syntax was only one factor pre cipitating the articulatory problem; word-finding difficulties also aggravated it and we shall have more to say on this subject below. It looked, therefore, as if Mr J's articulation was generally good as long as his confidence (in handling language material especially) was undisturbed; we subsequently noted many occasions on which articulatory problems seemed to arise as a secondary result of trying to control something too difficult at the level of syntax; or of word-finding difficulties and consequent disturbed sentence structure; or of gellerallow spirits. It is worth pointing out here perhaps that Mr J generally had very good control of intonation (see what wa'5 said earlier regarding the use of yes and no) and relied on this a great deal at this time in order to communicate with his family and with T. T 'what's this when j'm/ I'm/ P erm- oh/- T 'I am/- 'd'you 'get my . erm . actionsl I've 'got a 'piece of 'cake and I'm/ 'maybe my actions 'aren't very 'goodl try "'this onel P "'no no no/ mel T nO/ well'" 1 . P "'no no no/ mel no no nol demonstrates eating action The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 171 We discovered very quickly that whenever loss of intonational control occurred it provided a reliable indication of temporary loss of comprehension, or of inability to grasp the structure of a syntactic combination appropriately. Subsequent to these two sessions (nos. 2 and 3), T fell ill for three weeks, and so the start of the therapy procedure was further delayed. However, it was encouraging to note that Mr J apparently retained a lot of the verbs material used in these sessions during this period; lack of retention had always been a major problem with him, according to the earlier therapists he had been assigned to. The therapy procedure took as its first concern the task of building up basic SVand SVO patterns. For this purpose T used pictures (e.g. of a man digging, or a girl sewing.a dress), in association with specific forms of questions (which we shall say something about in a moment). The idea was to progress from single element elicited forms (we are talking all the time here of clause-level elements, each of which may have more than one phrase-level element realizing it), where the single element might be S, 0 or V; then on to two-element constructions of the form SV or VO; and eventually to three-element constructions of SVO form. We considered it important to use verbs such as sewing, digging, eating, smoking etc. since they have the very convenient property of functioning either with or without a direct object (0) (see chapter 6). The virtue of this is that it was possible for T to hold V constant for both SV and VO forms of the two-element sentence type; and to continue with the same V when proceeding to the three-element type. In this way we hoped to maximize the benefit gained from working on SV and VO separately when it came to fusing them. It may be useful to set out thi., discussion as follows: Step 1 Step 2 Step 3 Elicited forms (clause-level elements) Single element Two elements Three elements SIO v sv VO SVO = «(D)N) (a) man (a) dress etc. digging, sewing etc. the man is digging the girl is sewing etc. digging the garden, sewing a dress etc. the man is digging the garden, the girl is sewing a dress etc. Now, in order to elicit these elements, both singly and in combinations, Thad recourse to three principle types of stimulus (see chapter 6): the 'open discussion' type is illustrated by the following examples: 4 The desired responses are indicated in each case after the arrow; = shows the phrase-level structure of c1ause-Jevel elements; ) indicates progressively more complex responses as remedia tion continued. Types (0) to (c) are relatively simple, while (d) was reserved for a later stage in remediation. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 172 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY lea) 'who is in this plcture/ (b) 'what is he/she doing/ (e) 'what is she sewing/ (d) 'tell me about this picture/ ( ...... 8 ~ =(D)N» ( ...... V (> VO>SVO» ( ...... 0 ( = (D){N» ( ...... SV/SVO) The second consists of providing a list of alternatives, with no intonational cueing as to how large the set of alternatives may be (we refer to these as 'open alternatives'): 2(a) is it a man/ or a girl! or ... etc. (b) is she eating! or drinking/ or ... etc. Finally, there is the 'forced alternatives' technique, occasionally extended to include more than two alternatives: 3(a) is it a man! or a boy! (b) is he drinking/ or eating! or smoking! (e) is he 'eating an apple/ or 'smoking a pipe/ Both open alternatives (OA) and forced alternatives (FA) supplemented the open discussion type of stimulus and were called on by T only when required; moreover, FA was used only when OA failed to elicit a response. In general, it was found that FA had to be used consistently at the start of the programme; one measure of progress was the ability of T gradually to fall back more and more on OA, and then to open discussion, and still elicit fully structured responses (although J(a), (c) of course, as expected, tended to yield single c1ause-element responses). The eventual efficacy of J(d) to this end was taken to be an encouraging sign of successful therapy, since it represented the most open type of stimulus possible (that is, involving least structuring on the part of T). As a further point on elicitation techniques, and on FA in particular, it should be noted that a three-term FA (FAa)-see 3(b) makes greater demands on P than a two-term (FAa). This is for two main reasons: first, there is the simple consideration that the chance of guessing the right alterna tive drops straight away; but, secondly, we have also noted in some Ps, especially those with apparent short-term memory impairment, a definite tendency to select the last presented term (a recency effect), or the first presented term (an initial effect). In such cases, T, by using FA:!, may 'bury' the appropriate term in the middle whenever this procedure seems appropriate.:; The distinction between 3(b) and 3(c), with the latter including both V and 0 within the FA domain is an important point, but one that will be more conveniently taken up shortly. Summary of therapy procedure Before going on to give a detailed account of the main points of the therapy procedure, it is desirable to give a broad outline of the work that it was possible to get done in the first eight months of therapy, from September to May. As will be seen, the picture is not an entirely straightforward one: 6 In general, we found FA3 sufficient for this purpose; no apparent advantage was found with a four-term FA technique over OA. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 173 Date Session Work done Comments 25 Sept. I Exploratory: attempts to stimulate First syntactic profile spontaneous speech using pictures + done. questions. 2 Oct. 2 Syllable sequencing (including phrases from Schuell); verb elicitation via actions. 9 Oct. 3 Recap on sequencing; then on verbs; some work on new verbs. Therapist ill; 3-week gap. 2 Nov. 4 FA work on V, N, Adj; ~ V. ~ S, .... 0. then -+ SV. -+ YO, .... SVO. 9 Nov. 5 FA on V + 0; FA on 3 verbs; mass/ count nouns; she is > she's etc. 16 Nov. 6 Recap SVO; spontaneous SVAs; Spontaneous use of reinforcement attempts; conunands SVO at home. (open/close etc.) Impending operation. 23 Nov. 7 Continue commands; adverbial/pre- Operation postponed. positional function of in/on; more commands. - Seen in ward 3 times; recap. of Operation performed; introduction of requests (May I have success; no regression . . . . ). 7 Dec. 8 Recap requests; up/down + come/go in adverbial and prepositional function; clause connection. 14 Dec. 9 Concentration on directional (vs. relational) contrasts; to/from. 21 Dec. 10 Recap of prepositions/adverbs; Starting to slow introduction of questions (Where ... ) down. Christmas break. Depression. 11 Jan.- 11/12 Only two sessions attempted, both 1 Feb. unrecorded. 8 Feb. 13 Recap SV, YO, SVO; sequencing No confidence. 15 Feb. 14 FA work on pictures; then open Much improved. questions; finally Adj (full/empty). Mild fit on 17 Feb.; 22 Feb. 15 More work on Adj; .... neg; attempts but no regression. at SV .... neg (not successful). Second syntactic 1 Mar. 16 Neg with Adj; unhappy .... not happy profile done. 8 Mar. - Consultant's report; great improve- Growing confidence. ment physically and linguistically. 15 Mar. 17 Adj .... neg; SV -+ neg; SVO -+ neg. 5 Apr. 18 Revision, mainly SVO; conjoined Third syntactic profile structures with and; linear sequencing done. better. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 174 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Date Session Wark dane 19 Apr. 19 Conjoined structures, and ...... then; sequencing; questions. 26 Apr. 20 Picture sequences, with written material ; elaboration of structures over the same material. 3 May 21 Further elaboration; two dimensions of structuring of written material. to May 22 Adverbials, with picture sequences; SV ...... SVA. 17 May 23 Adverbials continued; SVO ...... SVOA. 24 May 24 As for 23; FA resorted to. 31 May 25 More -Iy adverbs; FA on new material. Comments Joins Red Cross Club. Difficulty with howl what stimuli; materials problem. Good spontaneous production. As already pointed out, the procedure itself was begun in earnest only on session 4 after a gap of three weeks; thereafter, work progressed quite smoothly down to Christmas. Mr J responded well to FA work on building up clause structures, and T moved on fairly quickly (starting in session 6) to work on com mands, followed by requests and questions. The idea here was to give sufficient language ability to Mr J as soon as possible to allow him to resume an active role within his family. Fortunately, no serious problems were encountered, in spite of the worry and disturbance attendant upon an operation (postponed once) for a haemorrhoids condition. However, a fall-off in Mr J's ability to take in new material was noticed in the last session (number 10) before the Christmas break; and this led into a period of depression which still had not lifted by 11 January (when session II was attempted). Both sessions II and 12 were necessarily outside the scope of the therapy procedure; T was mainly concerned to explore what was wrong, and to try to determine whether continued therapy would be of any benefit. This period of uncertainty continued up to and beyond 8 February when session 13, involving a recapitulation of basic work done before Christmas, was attempted. By session 14, Mr J was showing signs of willingness to accept further work, and of a growing confidence in his speech once more. This remained so even through a mild fit on 17 February, so much so that it proved possible on the next session (number ]5, 22 February) to work out a second syntactic profile (see p. 174), as a guideline for subsequent remediation. This involved work with adjectives. and attempts at the development of clausal negation. On 5 April the third syntactic profile (see p. 175) was made and the latest stage of the programme, leading up to the introduction of adverbial elements, was implemented; after that time, he continued to progress fairJy steadily in spite of a further mild fit (in July). He started to get out of the house much more than formerly. on grocery errands etc. He was also introduced into a local Red Cross club, where he met a variety of The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. A Un_lf_ Problematic I U .intelligible' 2 Symbolic Noise 3~,ant I Incomplete ~I 2 Ambiluous Respoases Silmulus Type " Question, III Others C SpootaneollS ~. if ~ I- i Major ... ~.i 1:1 . Comm. Quest. r.d: • !i ! [xci ".. " ellS ~ 'V' 'Q' vx 1 QX =~ r.:b ~:.: . i VXY I QXY =~ /n XY VS &'oJ I I ... • do xr {ij~ ~ S QVS ::9: QXYZ r.~ ... ~~ Ito", tag .... ~ &'oJ wlral .... ci5~ ( +) NP VI' Initiator Compl .. Coord >~ - - -- - - Discourse = I' A Connectivity .... - Commenl Claus< j; Emphacic Order CIl"!. Total No. Sentences - - II Other 11..1 Normal Response Abnormal 37 tJ.., " Social 4' Stereotypes Problfms Sentence Structure Statement ·v·/4N' 13 Other Problems Conn. Clause Phrase Word Isv , vc/o 1o ON K VV .ing 1 SC/O AX Adj N I v part 6' Neg X Other NN IntX 1'1 PrN Orher ~~_. s~~]_ X .:~:~~ 1'- :!: ~ ~/£:~P_,,! X +-".:.Ai- - ~ SVC/O :I 2. VCiOA DAd) N 3 Cop -<:n • SVA vo.,o, Adj Adj N Au. ,,"4- 14., ! 1 Nel XY 3 Other Pr ON Pron N Adj N OO .. r lIOn '': ~ ,,"~~'- '-' "q"" c 9~'"~, c ~"1 n't SVC/OA AAXr r~pr NP Neg V SV040, Other ! Pr 0 Adj N Neg X 'cop ; eX 2 A!.Ix i XcX Other au' and Coord. I I Postmod. I I • cla.use ..,.t c Subord. I I , Clause: S Postmod. I • -<:r phrase Other. Clause. C/O I Comparative ·Iy ( ) I Clause NP VI' gaiJ$l! Passive Pro" Adjseq Modal Concord I CompJemenl Det N irreg Tense A position I V lrrq: Worder - ~ - - -- -- - - - - - - ------- ---4 Other Sy,lIoctic Comprehension ! Style I ~_--,~ _____ ---.J I Mean No, Sentences 0' I Mean Sentence 2.. , : Per Turn •. Length 10 C> D. Crystal. p, Fletcher. M, Garman. 1975 University or Rudin. Mr J: Profile 2 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. ,_ :!: ~ ~/£:~P_,,! X +_ ".:.Ai- _ X! ""'-~( Xl "'~" n C 9'''N'!!r, ~~1 x tl .. ell ~ Other f- _ s.c. u_ _S.f___ __________ ~ __ --' ,l i l ~_--, ___ A Uuulyoed Problematic r I U nin.ellilible b 2 Symbolic NOise 3 Deviant I Incomplete S", 2 Ambiauou. b i B R..- Normal Response Aboormal I Repet- Elliptical Major Full Struc- Prob- I S.imulus Type Totals it ions I 2 3 4 Major Minor tural S lems I - 60 Questions J'i r '3 5 6. 12.. S' J M Others 10' " 1,*, J8 "'". ~ j C Spolltaneous 13 ,.., "- Others It; I Minor ,~ Social Sz Stereotypes J,. Problems ~I /0, '" Sentence Structure i !- ' Major -~ ... u Comm. Quest. Stale_III l~ c £xc/, 2 c: 'V'" '1'1' 21 .J r,)l ·v' '0' Other Problems Conn. Clause Phrase Word vx OX sv 17 VC/O It ON 1.1 VV S C/o Adj 1'1 • -In, =1$.: AX V part C;o r.~ I NeB X Other NN 'nlX pi .! .. I PrN Other GIl:=' X_._SN~'.J_ K ..:~:V~~,_ X :...g(!.=r:~.pJlX +_A;"A!. __ -cd VXy I I QXY SVC/O 35 VC/OA 0 Adj N II Cop 3 -<:n 5~ ),,' XY vs SVA *' VO.O, Adj Ad) 1'1 Au. ." If. Js r.~ I NeB XY Other P,ON Pron f1'2. ... - j do XY fi}~ N Ad) N Other ..,n XY ..: ~:N~l7.(Y ~ 2':~~1' XY:... C:jON~J!!!. • ~!,p. ~ n'l . S QVS SVC/OA AAXY r~ Pr NP NeB V ::~ QXYZ SVO,O, Olher Pr 0 Ad) N Neg X , 'cop r.~ eX 2 Aux ' .... i 'au1t rn~ XcX Other and Coord. I I· Poslmod. I I " clause .... 1 lro", tag c Subord, I I. .. ~ J Clause: S Poslmod, I .... r .. '" i phrase :t~ whal ' Clause: C/O i cn~ I J Comparative ·Iy ( +) (- ) NP VP Clause NP VP Clawe Initiator Complex Passive Pron Adjseq Modal Concord Coord Complement Det N iTreg Tense A posilion >~ V irreB Worder J r.J -- - - -- -- - -- - -- - - - - - - ------ ... - J ~':! Olher Other Discourse Syttlaclic CompreilellSioll I = I A. Connectivity /I .. : Ilrt'rt' Style r. ~ i Commenl Clause "" ' Other ci:i : Emphatic Order Total No. IbS Mean No. Sentences 0:"1 1 Mean Sentence loS Sentences Per Turn Length © D. Cryslal. ]>, fle,,~er. M. Carman. 1975 University of Readin, Mr J: Profile 3 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. I ,.-- ~ J c I \ © CASE STUDY OF AN ADULT 177 other patients (not all of them, of course, dysphasic). In August, when this chapter was written, he was again worried and depressed, this time at the prospect of being assigned to a new therapist. It was hoped and expected that this setback would prove to be temporary. DetaUs In this section we want to highlight a number of points that came up during the schedule of remediation. The selection of these points has been guided by two considerations: first, we believe that, in spite of the diversity of language disabilities found in adult dysphasics generally, these points are liable to be encountered quite commonly when working in the area covered by a remediation schedule of the type being described here; secondly, they will conveniently serve as representative illustrations of the techniques and aims which have been discussed in rather more general terms above. (i) Linear segmentation A rather serious problem of linear segmentation, or the structuring of syntactic sequences, arose very early in remediation. Already by session 4, we observed the occurrence of SVO and SV structures where an abnormal intonational and rhythmic contour regularly appeared, and seemed to point to an inappropriate structuring on Mr J's part of the internal components of the sequence in question. For the SVO pattern, what would happen is illustrated by: 'the 'boy 'is. 'eating 'a . 'apple '-----v----' ' I S V 0 For SV structures, this has an even more disruptive effect: T whO is itl P 'the 'man 'is. 'fishing 'a T weill that's itl the 'man is fishing/ Even S may be broken up in this way: 'the. 'girl. 'is. 'sewing a . 'dress Notice in these cases how the pauses (.) fail to coincide with the boundaries of clause-level elements, suggesting that for Mr J the sequence consisted of three components which are not quite the same as S, V and O. In the first example. for instance, the first group includes is (which is typically accompanied by an abnormal degree of stress), while the second consists of the main verb followed by the indefinite article (part of 0), which is also stressed. In this last case, one problem became apparent at once; the indefinite article was represented by its most widely distributed form a, whereas in fact apple requires the less frequent form an. If both The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 178 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY the D and N elements of 0 had been encoded together, one would not have expected this problem to arise. However, if the indefinite article is encoded prior to selection of the N (of 0), then this is exactly what one would predict. Equally, Mr J had problems regarding the auxiliary element (is) of V; the true relationship between S and V is such that the distribution of the singular/plural auxiliaries is/are in: The boy is eating The boys are eating exactly parallels the distribution of the simple present forms of verbs (e.g. eats/eat), as in: The boy eats The boys eat. But insofar as Mr J treated is (and presumably also are, where this occurred at all) as one with elements such as the boy, and as distinct from verbal elements, then it was likely that this important parallelism of singular and plural forms of V was going to be missed. We said earlier (p. 169) that we would return to the matter of how far syntactic and lexical factors played a role in the articulatory ability of Mr J, and we noted the possibility that the syntactic difference between noun and verb function of a particular word (e.g. comb) might cause segmental phonological problems. Now, at this point we may see a possible effect of word-finding problems on the non segmental phonological pattern. For what seemed to be happening was that pauses were breaking up the SVO pattern just at those points where major lexical choices had to be made. We hypothesized at the time along the following lines. Words such as boy, eat, apple are often treated, both traditionally and in more recent grammatical theory, as fundamentally different from words such as a, the, is etc.: the distinction has been given labels such as 'major'/'minor' words, 'open'/,closed' sets of words, 'full'/'open' words, 'Iexical'/'grammatical' words. However Qifficult we may find it to define the boundary between these two types in a precise, once for-all manner (cf. Crystal 1966), we immediately perceive the nature and practicality of such a distinction. One way to model the distinction is to visualize the grammar as containing a vocabulary component (or lexicon) in which full lexical items are stored and from which they can be selected and inserted into structures which are specified in the syntactic component. The syntactic component as well as containing rules for enumerating structures (SV, SVO, DN etc.), will also contain sets of grammatical words, which will already be specified in the syntactic structures, prior to the insertion of items from the lexicon. It must be stressed at this point that terms such as 'already' and 'prior to' in this very informal account, must not be taken at face value to imply temporal order; it is rather an article of faith with linguists that one can only impose a 'logical precedence' interpretation on the use of such terms, if they are to be used at all. This said, however, we feel that little harm is done in using them in this case; for it is observed that Mr J produced is con siderably before (temporally) he produced the complex form eating (containing the major lexical item eat), and a likewise before the major item apple. It was The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 179 precisely these temporal pauses that interrupted the intonation contour; and we may ask here if it might not be the case that two fundamental activities, lexical search and syntactic construction, are in fact temporally as well as procedurally related, in such a way that problems with the first of these tended to disturb the implementation of the second. Although there are many interpretations possible within this general position, we took the view that word-finding problems were resulting in an abnormal intonation pattern; and that this in turn may have been affecting the development of basic clause structure (by disrupting feedback from Mr 1's own productions). When the segmentation problem arose, we were seriously concerned to find a solution as soon as possible. Three main possibilities presented themselves straight away: (a) We might restrict the vocabulary as far as possible, so that Mr J could learn, by rote if necessary to begin with, which Vs required Os and which did not;6 (b) We could work much more on the transition from SV and VO to SVO (keeping V constant over the three structures in each case); (c) We could reserve three-element structures until later on in the remediation programme, and concentrate on eliminating hesitation points in the S and V elements before proceeding. However, all these possibilities had the attendant disadvantage of leading to an overuse of basically the same language material; and, while this is of course an important methodological problem with all patients, we suspect that it is usually particularly acute with adults, and certainly necessitated serious consideration as far as Mr J was concerned (see also p. 189), We therefore attempted indirect ways around the problem, of which, again, there were three: (i) We included V + 0 in the FA technique (see 3(c) above) as in: is she 'sewing a dress/ or 'eating a cake/ (ii) We introduced contracted forms, via use of pronouns: the. 'girl is. VERBing/ she is . VERBing/ she's VERBing/ (iii) We alternated a(n)fthe in the D position of 0 elements, and alternated also mass and count nouns in the N position of these elements: apple } With (i), we hoped to develop in Mr J an awareness that V and 0 hang together in an SVO structure as a particularly cohesive unit; by supplying both major o Cf. ch. 6 for reasons for the focus on YO. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 180 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY lexical choices together (V and 0) in each of the forced alternatives, we were able to elicit intonationally sound responses, which could serve as appropriate feedback. Clearly, this was just to get Mr J moving, as it were, and would not be appropriate over a long period. With (if) we hoped first of all to eliminate the possibility of hesitation within the S; and then, by eliminating the SUbject-Aux hesitation point, to make the Aux-Verb hesitation point the first one in the sentence; this might then be eliminated by encouraging Mr J to complete the first lexical choice prior to starting the utterance. Finally, with (iii) we wanted to highlight the dependency between D (a(n)jthe), or lack of D, and the type of N that follows. We could, in principle, have extended this method by taking in plurai count nouns (which, like non-count nouns, do not pattern with a(n», but we felt that this might be counter productive in possibly confusing Mr J. On the whole, methods (i) and (ii) proved to be the most successful; an example of (ii) is this: T p T P T P T whO's 'walking! 'the. 'girl is . wAlking! yes! good! s6! she's! *waIking/ *walking/ yes/ again! 'she's walking! 'very goOd/ confidently But wrong segmentation remained a serious problem right across the range of structures we wanted to produce, and was still not entirely cleared up at the time of writing; for example (from the session on command structures): p 'put. 'back the. 'cigarettes plcmse/ errn 'put. 'back the . ['attr~I]!plcmse/­ 'put. 'back the. [attr61] 'please/ (ii) Incipient structure intake of breath ashtray second ashtray Up till now, we have talked as though the remediation procedure was implemented throughout as it had been conceived on the basis of the initial syntactic profile albeit with modifications and supplementation in order to cope with such problems as the one regarding linear segmentation that has just been discussed. This may well be the usual case when one is working with young children, helping them to overcome a general or specific language delay; but when dealing with stroke patients (and probably adult dysphasics generally), one is always liable to encounter incipient structure (see pp. 116-17) at any level of description-phonological, syntactic, semantic. A case in point arose in session 6 when T was prepared to revise basic SV(O) structures prior to moving on to commands. What happened was a sudden emergence of adverbials during the revision of SVO, as is illustrated in the following transcription: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. m ['attr~I]/plcmse/­ P the 'girl is . 'sitting ·down/ yes/ T ·goOO/ well done/ good/ CASE STUDY OF AN ADULT 181 (NB T reinforces here just as P produces the unlooked-for adverbial down.) P the 'girl is . on her 6wn/ yes/ T goOd yes/ OK/ (P was uncertain of his control of the adverbial here, hence the rising intonation as if to ask 'is this cerrect ?') P the 'girl is . 'in. bed/ T right/ For T this was a particularly awkward-though naturally gratifying in a general way-point for incipient structure to arise, because it happened right at the start of the main commands > requests > questions component of the remediation procedure. We felt that such a progression would demand the full attention of both Mr J and T, and we were reluctant either to ignore or wholly follow the A-elements when they appeared in this unsolicited fashion. In order to explain what we decided upon and why, we must first take a closer look at A-elements in English. At clause level, we may note that A-elements modify basic SV and SVO structures in the following principal ways: (a) (b) (c) SV type SV A John ate hungrily ASV Hungrily, John ate SA V John hungrily ate SVO type SVOA John ate the food hungrily ASVO Hungrily, John ate the food SAVO John hungrily ate the food Of the three A positions here, ta) is traditionally recognized as basic: while (b) preserves the basic SV(O) pattern undisturbed, it carries an exceptional intonational contour (cf. the comma in orthographic representation), whereby the front A element is emphasized; and (c) involves insertion of the A-element within the basic SV(O) structure (though we should notice in this example the impossibility of breaking up V and 0 in this way). At phrase level, A grammar of contemporary English makes the following classification of A-elements, which we list here for convenience: (i) Adverbials (most commonly, those ending in -ly as hungrily, slowly, qUickly etc.) Oi) Noun phrases (e.g. last week) (iii) Prepositional phrases (e.g. in September, in the garden, by stealth) (iv) Finite clauses (e.g. when he hadfinishecf) The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 182 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY (v) Nonfinite clauses (a) -ing (e.g. waiting at the bus-stop) (b) to (e.g. to get the groceries) (c) oed (e.g. terrified but determined) (vi) Verbless clauses (e.g. ap%getio about everything) From this it may be appreciated that types (iv) to (vi) are more difficult than (i) to (iii), presupposing as they do a complex sentence structure. This is true even of type (vi) where the verb of the A-element clause has been elided. Moreover, of (i) to (iii), type (ii) constitutes a relatively small set; and we thought it best to concentrate first on introducing those structures which would have the greatest chance of common use. A further point of difficulty with using type Oi) was that we wanted to emphasize the distinctness of clause-level elements such as V, A, and SID as much as possible, as an aid to Mr J's developing ability to structure them together; and using exponents of A which at phrase-level were not distinct from exponents of SID was felt to be a possible source of confusion. This left us with types (i) and (iii), if we were to introduce A-elements at all. In this connection, it was encouraging that these were just the types that Mr J had spontaneously produced. In this light, we rethought the implications of our original therapy procedure (see p. 168), and noticed a point that probably should have occurred to us before: prior to developing commands, questions, and other socially important structures (e.g. requests), it is probably not enough simply to have reached the stage where P can handle basic SV(O) patterns (as in statements). For commands, questions etc. are not simply transformationally related versions of statements, but are inextricably bound up in their own particular situations, within which they are appropriate forms of linguistic behaviour, and require their own forms of linguistic or nonlinguistic response. Moreover, ability to handle anyone of these types involves both production and comprehension; and very often these different types will be found to complement each other within a single dialogue. We can perhaps make these points clearer with some hypothetical but realistic examples: (a) Give me a cigarette. (command) R { Tell me where you put them. (command) esponses Are they in the kitchen? (question) (b) Where are my cigarettes? (question) R { Think where you last put them. (command) esponses Did you leave them in the kitchen? (question) It is common experience that a simple command or question is not always going to achieve the desired effect, even with the most sympathetic listener; one partieu larJy important supplementary piece of the dialogue is the command type (e.g. Tell me where you put them), and another is the question (e.g. Did you leave them in the kitchen ?). Notice that with the first of these, the speaker oUght to be able to supply the required information immediately, all as part of the overall command, or question-and for this he must be able to control A-elements (providing the The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 183 precise location of the cigarettes, in this example) in his production. Equally, with questions from the addressee, the first speaker must be able to control A-elements (in the kitchen ?) in his comprehension. We therefore decided that what had emerged as incipient structure in Mr J's case was in fact relevant to the aims of our original procedure at this point; and we were also clear about the phrase-structure types that we wanted to introduce first. There was however one final consideration, which concerned our choice of particular lexical items in remediation, and this will be brought out in the next section. (iii) Adverbial/prepositional function Thus far, we have discussed A-element types (i) adverbials and (iii) prepositional phrases as though there were no point of contact between them; but this is not really the case. At word-structure level (at which Mr J showed most control, once he had found the required lexical item) types (i) and (iii) are closely linked, through the class of items such as up, down, in, over, etc. These may function either as type (i) elements, as in: The man ran up The ball went down S V A or as prepositions within type (iii) constructions, as in: The man ran up The ball went down S V the hill the slope A /I~ Pr D NP Now, given the possibility that we could introduce both type (i) and type (iii) patterns via the same lexical items (thus restricting the scope of the word-finding task), we were faced with the decision as to which pattern should be concentrated on first. We saw two advantages to treating type (i) as basic: first, the overall structure is shorter (Le. by a whole NP): secondly, the degree of contrast between up, down, in, over etc. in adverbial function (type (i» is often greater than that between the same elements in prepositional function (type (iii), and we thought that this would help Mr J in developing his A-elements system to begin with. The greater degree of contrast is partly the function of position within the whole struc ture; final position, which often seems to be reserved for information-bearing elements, is occupied by up in The man ran up (i.e. he didn't run down, or across or over); but it is occupied by the hill in The man ran up the hill (i.e. he didn't run up the ladder, or the road, or the mountain). However, we also saw the advantages of linking the two patterns as explicitly as possible, and aimed for a considerable period of overlap in their use in the remediation procedure; and the introduction The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 184 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY of both patterns was made within the framework of the original schedule (involving commands, requests and questions). T started off with commands involving a penny, and a box with reference to which the penny could be placed by Mr J in various locations. Thus Mr J had to comprehend not only the command structure, but also, in order to respond appropriately, the adverbial contrast (which was restricted initially to in versus on). T aimed, later on, to introduce questions in relation to commands (by asking Where is the penny? after a command had been appropriately responded to by Mr 1), thereby also starting to switch roles of production and comprehension between Mr J and T; and, eventually, to get Mr J making the commands (and supplementary questions). To begin with, there was little difficulty, and Mr J responded in the main with appropriate responses (nonverbal-simplY placing the penny in/on the box). Then T started to ask Mr J for a verbal response to accompany the nonverbal (keeping it as simple as possible to begin with-in or on): T 'listen to 'me first of All! - on the box!- "'in the 'boxl P "'in the 'box! on the bOx! in the box! yes! T 'good for you! 'well done! P yes yesl But trouble came when Mr J, having correctly placed the penny (on the box, in this instance) and correctly described his action, was subsequently unable to describe its location (NB using the prepositional function); this was the case even though T reverted to FA at this point: T now! 'put it on the 'box! P on the 'box! yes! T right! "'goOd! P "'yes! T now! 'where is it! - is it in the box/ or on the 'box! nowl P in the bOx/ . no/ . on the box/ . n6! A short while after (in the same session) T again approached this area, working towards a complete SV A. At first, there was some success: T can we 'have the 'whole thing! P on the 'box! T mhm! the!- P the. 'penny is on the 'boxj The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. T well done/ again/ P the 'penny is 'on the box/ T agaIn/ P 'the. 'penny is 'on the box/ . but it was lost again shortly afterwards: P the. 'penny lsi the. 'penny is/. - the. --- shut the door! (laughs) CAse STUDY OF AN ADULT 185 (Shut the door is a well worn routine from earlier sessions on command structures, put in here for humorous effect by Mr J, acknowledging his inability to supply what is required.) Whatever was wrong at this point seemed to be specific to elements such as in/on, since the session (no. 7) as a whole was successful in other ways (see the dis· cussion of commands, below), and Mr J was working cheerfully and well. Apparently we had reached the point of satiation as far as in/on were concerned; and we therefore reconsidered our choice of lexical items for adverbial work. As a result, we concluded that in/on were too close to each other, semantically (and perhaps phonologically too): they both referred to location of X at Y (with the in/on distinction referring to different modalities of 'being at'); and, in the way that we had been using them, they both referred to stable schemata, encoded as 'stative' linguistic patterns.' In particular, it is probably important to note that on can be used in very general 'contact' situations, as in even The fly is on the ceiling (not under/beneath the ceiling». Accordingly, in the next session, T used up/down in dynamic constructions. It should be noted here that up/down are in contrast along two possible dimensions, which we may call the 'relational' (stative) and the 'directional' (dynamic). Thus, on the first of these, we have: 'up' X I I I . Y I 'down' X -which simply means that X is up/down makes sense only in relation to some reference point Y (which may be an earlier position of X). On the second dimension, it is the direction of travel (dynamic) that forms part of the relevant opposition as is shown by the arrows: , On the distinction between 'stative' and 'dynamic', and for the relationship between in and on, see Lyons 1968, 324-5. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 186 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY 1 I 'up' X 'down' X . Y I 1 Here also, of course, some reference point is required, to distinguish movement up from movement down. Essentially the same point has to be made about in/ollt, which T dealt with after up/down; we shall briefly consider the two dimensions of this opposition here, before going on to show its relevance for the way therapy was conducted for Mr J. In the case of in/out, the stative opposition may be pictured as: Y Y 'in' 'out' X Notice that we again need the reference point Y, which in this instance has to be able to contain X. Now, the dynamic opposition may be pictured in two ways: A Y B Y 'in' 'in' 'out' X X X 'out' 'X Notice that in both A and B in refers to movement from a position like that for stative out, to a position like that for stative in, and vice versa. Now, in A the direc tion of travel from one position to another is constant with reference to Y; but in B it is reversed between in and out. So, in A we can say that there is movement which results in an opposition of location which is essentially like the stative opposition; while in B there is an opposition of movement (directional) which leads to a stative opposition of location. Or again, there are two oppositions in B, but only one in A. Now, in order to emphasize the dynamic contrast, we originally decided to link up/down (which T worked on first) to directionally-opposed verbs of the most basic type in English, viz. come/go. Thus, T always used go + up, and come + down (in the context of rockets going to the moon and returning to earth etc.). Our first discovery was not long in coming: Mr J at once, and quite reasonably, ignored this arbitrary pairing, and seemed quite unhelped by our attempt to keep the directional opposition constant in both V and A. He handled structures like the following very well: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. X --I---+ --+--+X +---/---' CASE STUDY OF AN ADULT 187 This {rocket '\ is {gOing } u bus J comIng p ThO {rocket}, (going ld IS bus IS \ coming J own Notice the use of this as D; Mr J seemed to have no trouble in taking this over from T's own description of the drawings that were used for this part of the procedure. Having got this far, T now explored the possibility of eliciting pre positional function of up/down (NB keeping the tone contrast on up/down): This bus is {gOin~ UPd'" } the hill. comIng uwn This proved no problem, and we felt that we might well have succeeded, or be in process of succeeding, as far as our two main aims were concerned: first, to consolidate incipient structure of the A type, and secondly, to link these two A categories by introducing them systematically together. Moreover, it was notice able at this point that Mr 1's good intonational control was starting to modify the intonation of the opposing structures (linking them into a single intonational unit): This bus is going up (the hill); this bus is going down (the hill). The parentheses indicate that this was true of both adverbial and prepositional function of up/down; and we were so encouraged by this that we decided to treat it as ripe for consolidation too. and introduced the conjunction and. We hoped in this way to capitalize on our work with incipient structure at phrase level by using it as a natural framework within which to build up still more clause-level com plexity. Interestingly, the distinction between adverbial and prepositional function came out most clearly here, and in the expected direction; Mr J could handle conjunction of two adverbial function patterns, but prepositional function proved too much: P 'this 'rocket 'is 'going up! and! . ""this 'rocket 'is 'going down/ T "'right/ P 'this 'bus 'is, 'going 0 up the 'hill/ T good! P 'this 'bus 'is. going/ . now! 'the 0 yes! - - T mhrh! P 'this 'bus, 'is 0 'going 0 down/ 0 'the 0 hilli. T right! P and! . 'this 0 'bus. 'is. 'gomg . 'down the hllJl 'yes 'the--- T 'which 'one is going down/ show me! which one/ The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 188 THE GRAMMATICAL ANALYSIS OF LANGUAGB DISABILITY P T p T P T up/ that's right/ up/ and/ down/ down/ yes/ goOd/ up/. And/. down/ good/. pointing Notice how here the adverbial function of up/down still eHcits an appropriate response. It may be that sentence length is the limiting factor here: but. as we pointed out earlier, it is confounded with vaguer notions such as degree of semantic contrast, which may be at least equally important. Two final points may be made here. The first, which takes up the issue intro· duced above, concerning in/out, shows the importance of directional opposition; T had been using in/out in connection with a dog and a kennel where the movement of the dog was constant: (Clearly, this kennel had a back door!) Mr J found this problematic, rather to our surprise: his production of in/out descriptions showed uncertain intonation, long pauses and occasional loss of the opposition (even for adverbial functions of in/out). At length, Mr J took the rubber stamp which T was using, for depicting the dog, and experimented for a while, eventually coming up with the following: He was then able to say in and out confidently (pointing to the appropriate dog). and seemed much happier with this revised schema-even though one of the rubber-stamp dogs was now necessarily upside down.s He was subsequently able • Assuming that the dog had always to be facing in the direction of ils travel, rather than moving backwards-which seems reasonable enough. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 189 to produce full structures involving both adverbial and prepositional function of in/out (NB out of is the complex prepositional group which corresponds to the adverbial out). We are convinced that Mr J was in effect highlighting here the importance of the directional contrast (rather than, say, objecting -to the notion that kennels have back doors), since exactly the same modification was required later on in the session in respect of two trains and a tunnel (one going in, the other coming out). The second. and last, point is that this excursus into A-elements at phrase and word level at once started to yield a limited productivity, and this has steadily improved ever since. The first instance that we noted was the following: P T p T p T the 'boy is . [opskoJJ! yes! wClIl/ yes yes yes/ yes! the 'boy is . 'hopscotch. phi:ying/ the 'boy is . 'playing. 'with a . hopscotch! yes yes/ 'jolly good/ OK! right! Given the close relation between: and . {football} X plays with a volleyball X plays {football } volleyball 'hopscotch' humorously laughing laughing we were not too worried by this. The main concern was to get Mr J's syntax moving again, and to leave finer matters regarding the boundary of syntax and semantics to a much later date. We were confident. for example, that an utterance such as: the 'boy is . football! did not indicate that Mr J thOUght that a particular boy was a football. Perhaps the most encouraging point regarding the hopscotch example was the degree of self-correction that was involved. (iv) The plateau It is convenient at this point to say something about the difficulty that Mr J showed in taking in structured therapy, from 21 December to 15 February. T had had considerable success in introducing command structures, leading to requests, and was now approaching questions; and, as has been discussed above, definite progress was being made simultaneously with A-elements. We decided that, in view of the problems in conjoining structures such as: This bus is going up the hill This bus is going down the hill, The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 190 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY it would be as well to avoid further clause-level complexity of this type; it might well prove more suitable to build up and- constructions from phrase-level first (red and green pen ,. red pen and green book; running up and down ,. running up and walking down). T therefore decided to focus on wh- questions in the session on 21 December (number ]0); this being the last session before the Christmas break, we looked forward to being able to complete the introduction of all the major structures in this first part of the remediation procedure. It should also be said, however, that we were conscious of the need to move on as soon as possible to new material and fresh structures: Mr J was understandably showing distinct signs of boredom with the material involved with in/out, up/down etc. We naturally wanted to concentrate our efforts mainly on those structures which would socialize him as rapidly as possible. This session was the first one where T was accompanied by one of the authors; Mr 1's relatively poor performance in this session was at first put down to this potentially disturbing influence, but we later revised our opinion about this, as will be seen below. After a brief warm-up on familiar material (involving upjdown) where Mr J's ability was noticeably down on the previous session-T started introducing wh- questions, concentrating mainly on the wh- word where. The first part of this introduction involved a comprehension task only-and this seemed to be perfectly good as far as his vision extended: he was able to point and say There, On the table etc. to questions such as Where is the TV? The next stage required Mr J to ask T where various objects in the room were. Here the same sort of problem was encountered as when in session 7 he had had to command his daughter to perform certain actions (Bring the ash-tray, for example): he often responded with the answer instead of asking the question. (With commands, he had initially tended to perform the action instead of commanding it.) In an effort to overcome this, T and Mr J joined together as a question-asking team, with Mr J's wife and daughter, and the author who was present, as a question responding team. T was now able to provide model question structures where necessary without Mr J's assuming that they were genuine questions rather than models. However, there remained a great deal of uncertainty on Mr J's part; wrong linear segmentation was particularly noticeable in his production, as was a shaky grasp of intonational structure. Moreover, we also found that he could often only initiate a question structure with the name of the object whose location was being questioned; and he would repeat this word until T showed him the word where (written in block letters on a card), whereupon the whole structure would be produced: T P T P T P 'what televisionj 'you ask MIke! Mikel . televisionl nowl re'member 'how we are asking itl yes/ Mikel 'where is the. television! 'where is the televisioni 'where is the television! T shows card The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 191 In general during this session, it appeared that word-finding difficulties were tending to displace Mr 1's attention to syntactic structure; and there was very little carryover even from one part of the session to the other. Material that had eventually been successfully handled after initial difficulties apparently presented exactly similar difficulties when re-presented only a few minutes later on. When T returned, after the Christmas break, Mr J was still generally depressed and unwilling to work through further therapy; at this stage it was an open question as to whether this was a period of regression or simply a 'plateau'. T maintained friendly social contact-and, on a couple of occasions, arrived without a tape recorder and was surprised by Mr J who insisted that a therapy session be con ducted! However, these were relatively ineffectual sessions from the point of view of the past performance of Mr J, and of the expectations embodied in the therapy programme. (v) The development of negative structures Eventually, however, T was able to make a serious revision of basic SV, VO and SVO structures (session 13, 8 February), and to proceed from there to an elicitation task involving description of well-known routines or action sequences such as lighting a cigarette, plugging in and operating a tape-recorder etc. Mr J showed great willingness to cooperate, and a fair amount of retention; but his control of intonation patterns was shaky, and he generally showed a marked lack of confi dence. In an effort to introduce as much variety as possible into the therapy materials, T went on in the next session to work with picture-lotto material which it was hoped would prove to be both graphically simple and yet unchildish. T pro gressed from FA to open questions on this material, and Mr J showed much improved confidence and interest. Finally, T introduced adjectives, relying for the most part on contrast between pairs (such as big/small, tall/short, full/empty etc.) for ease of acquisition. Adjectives were felt to be a suitable area to work on at this stage for the following reasons: first, the first syntactic profile showed rather little SVC clause structure (the ball is red), and few instances of Adj-types at phrase level; secondly SVC clause structure was a natural extension of SVO structure and should therefore cause little difficulty even though it represented new material; thirdly, linking SVC structures (the ball is red) to corresponding phrase structures (the red ball . .. ) would allow simultaneously for development of language control at two levels, and, we hoped, would allow Mr J to intuit the regular relationship between the two. In this connection, it should be noted that we followed general linguistic thought in taking SVC function of adjectives to be basic, and noun phrasal function (pre nominal modification) to be derived from this. Finally, we felt that, having establh;hed adjectives (proceeding from clause structure to phrase structure), we could use this framework for the introduction of negative structures (proceeding from lexical negation and morphological negation and building up to clause level---cf. what was noted above on p. 186 regarding the operation of the conjunction and). This was, of course, an important component of our original programme of therapy for the post-Christmas period. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 192 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Accordingly, after T had asked questions such as: 'is the 'girl 'washing her facet P provided fully structured responses quite regularly (though note the concord problem, which we did not attempt to grapple with at this stage): P yes/ the 'girl is 'washing his face! T then moved on to general adjectives, as illustrated here: T is the 'dress pink/ or green/ (FA) - 'what 'colour is the dress/ (open discussion) P the 'dress is pink/ Subsequently she moved quickly on to contrasting adjective pairs, as in: T is the 'bottle fUlI/ or empty/ Now, Mr J was picking up quite a lot of visual cues from the printed description of the pictures all through this session, so it was quite surprising to us that he showed unresolvable hesitation when faced with this question. In particular, we had reason to believe that up until now Mr J's word-finding difficulties had largely been circumvented by the use of such visual cues. Since full/empty may be characterized as positive/negative lexical items respectively, we were faced with the possibility that a quite separate problem from the word-finding one was being encountered here, having to do with negation. This suspicion was strengthened as a result of the following sort of situation, where T was simply testing comprehension of the negative term: T /is the 'bottle empty! (of a full or empty bottle) P /yes/ . n61 By contrast, asking the positive version of the question generally prompted a single, definite response (as long as the bottle was full): T lis the 'bottle fUll! (of a full bottle) P /yesl If the depicted bottle was empty, Mr J would show the same hesitation as when he was asked the negative version of the question. It should be pointed out here that at this stage he was able to take (perfectly legitimate) evasive action when T tried to elicit a negative sentence structure; as an illustration of tIDS, we may cite the following: T lis the 'cow standingl (cow pictured lying down) P In61 the 'cow is sitting/ Notice here that, apart from the negative particle no, the sentence structure is positive, not negatiVf'; and that Mr J substituted a different lexical item (sitting The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. CASE STUDY OF AN ADULT 193 for standing in T's structure). This is all the more remarkable in view of his word finding problems, of course. Now, the difference between standing/sitting on the one hand and full/empty on the other is that only the latter pair is opposed as positive/negative. So we had the following situation at this stage: Mr J would resort to lexical substitution in order to avoid sentence negation, only so long as lexical negation was not involved. The problem was thus clearly one regarding negation itself, and seemed to arise equally with T's use of words such as empty and with the use of pictures showing an empty as opposed to a full bottle. In the next session (number 15), T began with a revision of adjectives that had been fairly well handled the previous week, and then moved on to the problem of full/empty again. Whereas, however, the bottle depicted had previously been either full or empty, on this occasion T used a picture of a half-filled bottle (i.e. not characterized appropriately as either full or empty). We did not seriously entertain the possibility that Mr J would be able to describe the bottle as half-!uJl/ha!f-empty, and expected that he would therefore be forced back onto syntactic negation. Note that in this situation Mr J did not have to decide which member of the adjective pair was required; what he had to do was negate the structure which was in large measure provided by T. The problem was not a word-finding one now, but one of syntax. What happened was that Mr J showed little ability to meet this problem, in this session; but in the next (number 16) T followed what was to prove a fruitful approach, via those adjectives that permit the possibility of morpho logical negation. Using pictures of happy and unhappy faces, Mr J was eventually able to produce constructions involving unhappy (rather than sad); and from this point it was no problem to move to not happy. Immediately after this success, Mr J spontaneously started to produce a number of syntactic negatives, including the rather impressive: 'this 'man is not so 'tall/ It may well be that this session made contact with incipient structure that was exactly in line with what T was trying to elicit (see p. 179 above). Whatever the case, T continued work on syntactic negation in the next session. in four main steps: (i) First, after revision of unhappy not happy, unkind = not kind etc., T concentrated on the same sort of equivalances with new material, where the adjectives are formed with -ing (participial adjectives) and are intermediate in status between the class of adjectives proper and the class of verbs (unWilling not willing; uninteresting = not interesting; unamusing not amusing etc.). In each case, the not Adj form was introduced via the morphologically negated form un-Adj. (Ii) Then T moved to verbs proper, using the -ing form first of a.1I with intransi tives (not running; not walking etc.). Note that here forms in un- are not possible. (iii), (il') T subsequently moved to introduce verbs with both a transitive and an intransitive use, such as eating. drinking. first without the following 0, and then with. All proceeded smoothly until the final stage, when certain problems were encountered: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 194 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY P 'this 'man 'is smokingj T right/ P 'this 'man 'is not 'smoking/ T right/ that's it! P 'this 'man 'is not / erm T P this 'man is / smoking a 'pipe/ mhmj this 'man isl . confused But these were eventually overcome, and syntactic negation was being regularly elicited: T 'this 'girl is eating/ p 'this 'girl is eating/ 'this 'girl. 'is. not 'eating! T good! that's it! P 'this 'girl is eating! 'this 'girl is not 'eating! T mhml no 'let's 'put the apple in! - 'this girl is/ - - P 'this 'girl is . 'eating. an applej 'this 'girl is eating/ 'this' girl is eating/ T 'this girl is *ifl/ P *'this 'girl is not 'eating/ 'this 'girl is 'eating an Apple/ T good! P 'this 'girl is not 'eating an 'apple! . Conclusion Of course, we must avoid leaving the impression that we are here reporting on a completed case history. Mr J was still undergoing therapy at the time of writing, and was not then in a position to be discharged. Between April and July, three main areas were worked on: (i) sequencing and elaboration of structures over the same material (if) structuring of clause/phrase-level sequences (iii) development of A-elements. (i) takes up the issue of conjunction; it was noted earlier that Mr J found difficulty in clausal conjunction in structures such as: this 'bus is 'going up the 'hill! and! this 'bus is going down the 'hill/ even though the intonation pattern suggested that the desired structure was there in embryo. In an effort to develop clausal conjunction, phrasal conjunction was concentrated on, using elaboration of structures over the same material as is also The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. l il l CASE STUDY OF AN ADULT 195 being used for sequencing tasks. Thus, a picture in a sequence (making up a story) might show a man walking down the street smoking a cigarette; T tried to elicit structures such as: X is walking (down the street) X is smoking (a cigarette) independently at first, and then encouraged Mr J (modelling where necessary) to conjoin them: X is walking (down the street) and smoking (a cigarette). A large measure of success was achieved with this. technique, and the temporal conjunction then was introduced in session 19. (ii) This concerns the problem of linear segmentation, and really represents a new way of trying to tackle it. T used two dimensions systematically in the visual modality, the vertical being used for phrase structure. and the horizontal for clause-level concatenation (using words on cards): I the I was a fat smoking cigarette I man Eventually, we hoped to get Mr J to perform a 'word-anagram' task (attending to the uses of the two dimensions illustrated here) on this sort of material; but at the time of writing it was too early to say whether this would prove possible. or even if this approach in general would be successful. (iii) Further development of A-elements was required, and for this we tentatively set up the following framework of prepositional types of construction, basing our decisions on the linguistic assumptions described above (pp. 1841f.): Locative Temporal in/on/near } ~nder/beside (the table) In front of at the back of in (the holidays) at (Easter) during (this week) Causal with (the hammer) (including passive agent) by (John) That is to say, we are assuming, with Lyons (1968, 298-302) and others, that spatial location is basic with regard to these elements, and that temporal and causal functions are progressively more abstract; and that in (Locative) is more basic than at the back oj, in (Temporal) more basic than during etc. Again, at the time of writing, we were not able to say how successful this approach might be. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 196 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY However, in spite of the ongoing nature of our work with Mr J, we feel that it is perhaps worth concluding by laying emphasis on three general points. Firstly, concerning the remediation aspect of this approach, we feel confirmed in our belief that developmental norms provide the most reliable scale that we have so far in the notoriously difficult task of deciding on 'grammatical complexity' (see chapter 2). We take it as axiomatic that less complex linguistic constructions and elements should be introduced before more complex ones; and we feel that the best way of implementing this axiom is to take the normal developmental sequence of structures when deciding on order of remediation with the adult dysphasic. We have however noted a number of instances where the remediation procedure differs between adult and child, and this leads on to our second point, which is that we do not wish to be tied to a developmental hypothesis which would require us to believe that adult dysphasics necessarily undergo a language regression which is the mirror-image of the acquisition process (as argued for by Jakobson 1941). We make no claims in this chapter about the order in which syntactic structures come to be lost. either at trauma or in subsequent regression; we are only concerned with the order in which they may be most successfully introduced during remediation. Thirdly, it is important to note the extent to which work with adults requires reference to be made to semantic and sociolinguistic considerations in addition to any basic syntactic analysis. Our comments under this heading, however, are regrettably unsystematic, in comparison with our syntactic procedure-but this, we feel, is symptomatic of the present state of these fields (cf. chapter I). We can therefore claim little more than pragmatic justification for such notions as 'degree of semantic contrast' used above. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 9 LARSP 1975-1990 At the end of the first edition ofthis book, we added a two-page Postscript in which we sought to identify the tasks we felt were still necessary before LARSP would become generally applicable, and drew attention to the main points of difference between our approach and that of Lee (1974). A quotation from that section makes an appropriate start to this review of progress in the use of the procedure, 12 years on. First, we require many more intensive analyses ofPs. The number studied so far is still small, and we would like to extend the number and range of cases. [t is important that the analyses be performed by others as well as ourselves, so that we can be sure that there is a system here which is of general applicability across the range oflanguage disorders. Ine"itably, modifications will have to be made as LARSP is tried out in different environments and on different popUlations, and we hope that readers who decide to use this system will keep us informed of their experiences ... Secondly, more information about the normative back ground to LARSP is crucial, to ensure that the assessment and remediation of Ps is based on an account of normal language which is statistically respectable and sociolinguistically informed. These requirements also hold if the gap between individual assessment and the provision of general diagnostic categories is ever to be bridged. As we have seen, group norms of syntactic development and usage are conspicuously absent at present. Our file of information on the use ofLARSP since 1976 is by no means complete, as there is no constraint on users to keep us informed of their work; but our general impression is that considerable progress has been made in relation to both these aims. Our own work, carried out mainly at the speech therapy clinic at the University of Reading, involved LARSP being put to routine use in the special assessment of language disordered children referred from various centres mainly in the south of England. Thirty of these children were given an intensive study in a research pro ject in which LARSP was used alongside other profiles (see below) to establish diagnostic types (Crystal, 1986). At the same time, all of us were involved in in service courses on the use of the procedure, and providing follow-up assistance for those who began to use it: during the 1970s and early 1980s we organized over 50 2-.. 3- or4-day workshops on LARSP, in various parts of the world. In recent years, as the procedure has come to be taught routinely in many institutions, and more publications and software have become available (see below), it has been less The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 198 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY necessary to put on such courses. But in their time, they (and the follow-up days which they generated) were most valuable to us, in that they put us in touch with an enonnous range of clinical cases where we had not previously seen (and in some cases even thought of) LARSP being applied such as deafness, mental handicap, and cleft palate. (To date, the least predictable of all our encounters was to find LARSP being used in a voice clinic - the context being an analysis of a laryngectomee's ability to handle chunks of syntax within the limits of his pattern of inspiration.) Many of the observations made during these courses, and the accompanying visits to clinics and classrooms which arose out of them, provided the motivation for the series of revisions in the use of the procedure which we introduced in 1981, and which is printed on p. 222,£f of this edition. It is difficult to summarize such a wide range of experiences, but we feel able to make certain general observations. I Concerning the descriptive aspect of the procedure, we feel we made the right decision to choose a framework which related to a major model of contemporary English language description, i.e. the approach of Quirk et al. This approach has continued to develop its descriptive power (notably in Quirk, Greenbaum, Leech and Svartvik, 1985), to provide data via software programs (e.g. Svartvik, Eeg Olofsson, Forsheden, Orestr6m and Thavenius, 1982), and to prompt simplified introductions (e.g. Crystal, 1988). These points are of particular importance when T wishes to work in greater detail on a particular LARSP category (e.g. a microprofile of Aux, Pron, or Clause:A) and requires a more detailed description than we are able to provide in this book and our associated publications. 2 We have continued to monitor research into language acquisition, in case new findings made it necessary to change the assignment of structures to particular stages of the chart. We have in fact made one such change, as a result of the Bristol Survey (see e.g. Wells, 1985), moving tag questions from Stage V to Stage IV in the ]981 revision; but in other respects, we have been pleased to see that descriptive studies continue to con finn the validity of our Stage assignments (themselves orig inallya synthesis of findings in the child language acquisition research literature). 3 Concerning the interpretation of the profile chart, we have repeatedly found it necessary to emphasise that the chart plots only the order of emergence of grammatical structures, and does not provide a theoretical explanation for what is there. Why the structures emerge in the way they do is a matter which continues to be a source of speculation (e.g. Radford, 1989). Our aim was less ambitious: to provide an observational framework which would motivate the empirical study of grammatical disability, and provide hypotheses about diagnosis, assessment, and remediation. 4 Concerning the 'mechanical' use of the procedure, we have noted the great diversity of approaches, as Ts have adopted and often adapted it for use in their work. For example, at one school, where the focus of teaching was on Stages I to IV, we saw that those involved had redesigned the chart to focus just on those stages; at another, where the focus was on older children, Stage VI had been expanded to provide further guidelines. A related point is that the vast majority of people seem to use the chart for the developmental stages and section A only, and do not work The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975·1990 199 systematically with sections B-D. We have also noted that, for reasons of time, many Ts who use LARSP in their thinking about patients do not actually work through the procedure and provide profile charts of samples in their case notes. Our impression is that for every 10 patients who are mentally and impression istically LARSPed by T, only I actually gets written up, in the form of a transcription, profile, and accompanying commentary. 5 There is an overwhelming impression that the procedure is for use with expressive problems only. Although work with comprehension problems is more complex, we have nonetheless seen LARSP used successfully with both children and adults where there is concern to establish what range of syntactic structures is being understood. 6 The use of profiles forces T to concentrate on the symptomatology of grammatical disability. It is no longer possible, using such an approach, to restrict oneself to the use of simple labels such as 'language delay' or 'agrammatism', which we feel to be gross oversimplifications. The inadequacy of traditional terminology has been emphasized several times in recent years, as has the demand for published case studies to provide an empirical data base for language pathology (Crystal, 1982b, 1987b). Research studies While routine work with individual Ps continues to provide the primary use and main justification of our procedure, we have been particularly interested in those research studies which have been carried out on groups ofPs, especially where a statistical approach has helped to refine our original descriptions. The first study of this kind of which we are aware was carried out by Bamford and Bench (1979), in the follow-up volume to the present book, Working with LA RSP, in which language samples were taken from 263 hearing-impaired children, analysed using LARSP, grouped profiles obtained, and 56 variables from the chart subjected to a factor analysis, to produce a 'grammatical advancement score' for each subject. Wren (1981) studied profiles of syntactic usage among 6-year-old language disordered children. 30 of these children (identified as having a syntactic disorder through the previous use ofDSS, Lee 1974) and 15 normal children were grouped into clusters of statistically similar profiles (using a procedure known as Q-factor analysis). This showed two groups with imbalanced syntactic development, similar to two of those presented in Chapter 6 above: (i) a pattern of normal clause development with inadequate word and phrase structures; and (ii) a pattern of depressed word and phrase structure with even lower ability in clause structure, problems of inconsistent performance, with a wide gap between maximum and typical performance. In a follow-up study, Wren (1982) analysed the auditory and cognitive processing profiles of these two groups. The children were given a battery of tests to assess receptive and expressive processes relating to grammatical development. The groups were found to differ significantly in terms of their profiles on tests of discrimination, comprehension, memory, sequencing, and rule application of linguistic information. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 200 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Also on children, Haden and Penn (1985) used LARSP to investigate the syntactic and interactive development of a single twin dyad over a 19-month period. One of the twins was language-impaired and received therapy; the' other was normal. Samples were obtained from each twin alone with the experimenter, and interacting with each other and the' experimenter. The results indicated that the presence of the normal twin had a depressing effect on certain grammatical constructions of the impaired twin, and that this effect became less marked after the institution of therapy. Studies using LARSP on adult aphasics have included Easterbrook, Byers Brown and Perera (1982), who analysed the verbal expression offive adult aphasic Ps, comparing their speech in unstructured and structured situations (spon taneous speech vs a picture description task) at intervals over a year. Individual LARSP profiles were made of the samples, and an analysis showed a measurable, longitudinal difference between the Ps' performance in the two interactions - primarily, a greater proportion of discrete, grammatically analysable utterances in the picture description task. Penn and Behrmann (1986) analysed the expressive syntax of 38 aphasics with LARSP, using a hierarchical cluster analysis. Six discrete sub-groups were derived, ranging from least to most syntactically competent. A syntactic continuum for aphasia is proposed, in which fluent, non fluent, and other types are interrelated. The authors observe: 'It would appear that the number of apparently discrete profile types is greater than previous predictions in the literature, presumably because a more comprehensive analysis procedure was applied' (32). This is in no sense a complete listing of research studies which have used LARSP on clinical populations; but these references should suffice to indicate the way in which it is possible to use the procedure to arrive at clinically useful generalizations. At the same time, several modifications to the procedure have been introduced by these authors, and when making comparisons of results, it is necessary to bear these changes in mind. Indeed, at this point it is important to draw attention to the various published suggestions which have been made concerning LARS P methodology, notably by Connolly (1984), with a commentary in response by Crystal (1984b), and also by Penn and Behrmann (1986: 34-5) and Killingley (1981). Comparisons ofLARSP with other procedures have been made, primarily by Miller, Klee, Paul and Chapman (1981), and Hawkins and Spencer (1985). The only complete exposition of the revised version of the procedure, summarized on pp. 222-226, is Crystal (1982a; re-issued 1989). A follow-up collection to Working with LARSP is planned. . The range of uses of the procedure continues to be extremely wide, as illustrated by Klee and Fitzgerald's study of grammatical development and mean length of utterance in normal children (1985; see also Rondal, Ghiotto, Bredart and Bachelet, 1987), and Maxim's (1985) study of grammatical development in the elderly (see Maxim and Thompson, forthcoming). Evaluations ofLARSP include Muller, Munro and Code (1981) and Wren (1983: ch. 4). Detailed illustrations of its use in clinical and teaching settings are given in the various papers in Part 2 of Working with LARSP (Crystal, 1979), and also in Williams (1986), and the later The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 201 chapters of Crystal (1984a). Remediation procedures based on LARSP include 'Language through reading' (Gillies and Hutt (1985), and other related ICAN publications; see the perspective in Hutt (1986»; the Cambridgeshire Language Development Project (CLDP 1979), and also Dennis (1981) and Williams (1981). LARSP informs the approach to the analysis of children's writing and reading in Perera (1984), and is the grammatical basis of two sets of materials: 'Skylarks' (Bevington and Crystal, 1976) and the Databank series of reading book (Crystal and Foster, 1979-83, expounded in Crystal (1984a, 119ft). Quantitative issues In 1984 two Mus (for details see Fletcher, Garman, Johnson,Schelletter and Stodel (1986» began a standardization study designed in part to provide normative informat~on, via LARSP, on a group of normal 3-7 year-old British children, to serve as a reference database for the characterization of language impairment, particularly in primary school-age children. Later on in this section we provide an example of a (partial) LARSP chart based on a number of S-year-olds, from the study. To properly interpret this profile, and certainly to use it clinically, it is necessary to be aware of the methodological issues that lie behind its construction. The provision of profiles which include statistical information, particularly where 'normative' developmental profiles can be provided, would seem to be a valuable clinical provision. Any sensible comparison of an individual P's profile with some reference profile, however, requires that the individual profile be derived in exactly the same way as the reference profile; otherwise there is a risk of misinterpretation. The issues that need to be reviewed begin with transcription. Transcription and segmentation Any statement about the frequency of a grammatical category - e.g. 'S-year olds use on average 13.13 complex clauses (Stage V clause structure) per 100 utterances' - depends crucially on how 'utterance' is defined. Our recommendations for transcription (see p. 90) give guidance about such features as un interpretable speech, overlap, incomplete utterances, and so on. The computer-based transcriptional system developed for our standardization project provides conventions for these features and others that arise if an attempt is made to provide as accurate a record as possible of a conversation (see Johnson (1986) for details). The issue of what constitutes an utterance - or "rather, in the transcription of a stream of speech, what constitutes an 'analysable unit' was not taken up in detail in the first edition of this book (though see p. 58). One reason for this was that for the predominantly pre-school language-delayed or impaired population with which we were mainly dealing, it was very rare for a child's utterance to consist of anything more than a sequence of elements which could be attributed a clause structure or a phrase structure from the set recognized by LARSP. In the standardization project, however, involving older, normal children, a child's utterance might consist of a much longer sequence, which has to be segmented, accordil).g to some reliable procedures, into a series of units analysable into clause sequences (for details, see Garman 1988). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 202 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Sampling Ii has long been recognized that the conditions, in the broadest sense, under which a spontaneous language sample is produce~, can affect the nature of the language produced (see, e.g. Fletcher, Peters and Hixson 1982, Scott and Taylor 1978). It is for example apparent in the conversations from our database that there is general1y a low frequency of child interrogatives. This would be unusual in a mother-child conversation in the home (Wells 1985: Table AI7), but can be explained in our data by the fact that the interlocutor was always an adult female previously unknown to the child. (The decision to collect the data in this way was in order to make the conditions under which our data was collected similar to those of an initial assessment by a speech therapist.) The nature of the profile can also be affected by the general setting of the conversation, and what it is that the participants are talking about. We have analysed our data for two conditions within each sample under which data was collected - stick-on game, and 'free' conversation. In the stick-on game condition, talk is organized around a two-dimensional representation of a house or a farm, onto which the child places plastic stickers of objects which belong to one of these. In the 'free' conversation, the interlocutor chatted to the child about home, the family, school and recent salient events or events about to occur, such as Christmas, a birthday, and holidays. There was no script as such, but consistent efforts were made to get the child to talk about his or her life, and about events outside the 'here and now'. There are some quite clear effects of these different conditions. For example, we mentioned above the generally rather low number of child interrogatives in our samples. Given this, there were still marked differences, between the different conditions, in the proportion of interrogative structures used: 'Free' conversion (FC) Stick-on game (SG) 5-year-olds .06 .II 7-year-olds .03 .11 Number of interrogatives, as a proportion of all clause types used, in two different conditions within samples. The extra interrogatives in the stick-on game condition are either requests for the names of particular objects, or questions like 'Does this go hereT, in relation to the appropriate location of an object at a particular place in the house or farm. Examples of sampling differences can be seen in FC/SG differences in verb forms, at phrase level in the number of modal auxiliaries (AlJXffi on the profile chart), and at word-level in the number of past tense forms (wed on the profile chart), The SG condition has a higher frequency of modals, reflecting the child's tendency to express tentativeness about the location of an object by saying. for example, 'This could go in the bedroom', By contrast, there are many more past The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 203 tense uses in the Fe condition, because the interlocutor is deliberately introducing topics which require reference to the past. Reference to past events is much less likely in the case of the stick-on game, where both adult and child are jointly attending to objects on the table in front of them. Indeed, when we take into account also that the SG condition tends to have a much higher proportion of simpler noun and prepositional phrases (i.e. Stage II phrases), we can see an effect of condition at every level of analysis. The NP/PP difference can be attributed to the fact that, in the SG condition, referents to be identified are generally visible to both participants in the conversation, and hence do not require the use of modification in the form of adjectives or postmodifying phrases. What all this indicates is that in making statements about norms, considerable care is required in stating the conditions under which the samples from which frequencies are derived have been obtained. We are still some way from specifying the variability which pertains within a particular condition within a sample. However, it seems clear from our analysis so far that, certainly at the level of individual categories, frequency differences between conditions arise, which can be plausibly attributed to the contexts from which the data is drawn. Such factors can obviously affect the interpretation of a comparison between an individual P's profile and that of the 'average' 5-year old, for instance. _\'.0 Minor "~y Vocatives Olher Problems ~..::.: :J M' Comm. Quest. Stalemen! cii 0_" "lOr r----+-=---j--- ______ ---l I 'V' :0, I' ~er'l _ '" h o · _-L_O_the_r-L...~_-L. _ C . _p ~..L.-.-:.p_r_a,_e .-.!.. 204 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Such a profile is given on p. 203 (for simplicity, values are given for clauses only). We can summarize the points that need to be borne in mind when making use of it: (i) it is derived from 'free' conversation, between 5-year old children and a previously unknown female adult, in a school setting. (ii) This profile is based on conversations with eight children only. The analysis of data from a much larger sample. of children is required to provide more reliable figures. (iii) The mean frequencies listed in the profile are based on a sample of 40 analysable units. Reliability of the figures would also be improved by larger unit samples. Any comparison between an individual profile and the 'average' profile must use the same definition of analysable unit as that used here. (iv) The entry of clauses on to the profile chart differs slightly from standard LARSP procedure. Complex clauses are entered under the appropriate category at Stage V, but their component simple clauses are not then entered in the relevant category at Stages II-IV. (This modification was made to simplify computation of a simple-to-complex clause ratio in our computer analysis procedures - see below, on SUMLARSP.) (v) As already discussed, individual clause categories may differ quite dramatically in frequency in different conditions within a computational sample. There does however appear to be greater consistency, even with n = 8, for the proportion of clauses at particular stages, across sampling conditions, as we can see from the following table: FC SG II .15 .13 III .46 .50 IV .19 .20 v .16 .20 Proportion of clause structures by stage, for two different conditions within samples (n = eight 5-year-olds). Indices of development and difference An alternative approach to the individual norm comparison, with quantitative profile charts, is to use them to identify subsets of categories which are important either for the discrimination of adjacent age-groups, or for distinguishing language-impaired from normal children. Fletcher and Peters (1984) addressed the latter issue by comparing a group of carefully-defined language-impaired children (mean age just over 5 years) and a group of normal children matched for age and non-verbal intellectual ability. The initial comparison used the majority ofLARSP categories from Stage I-V, together with some extra categories involving lexical types and tokens. The comparison yielded a number of significant differences on individual categories. When these categories were examined together in a multiple regression analysis, only two The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 205 categories emerged as accounting for a significant proportion of the variance. One of these was a grammatical category, unmarked verb forms (verbs unmodified by inflection or auxiliary): the impaired group showed a significantly higher use of these (and so in general lower verb modification). The second was a lexical category, that of main verb type: the impaired group had a lower average type value - they appeared to have a less differentiated set of verbs available. There are two general conclusions for profiling practice that seem to be suggested from these findings. First, that to investigate important areas of deficiency in more detail in language-impaired children, microprofiles (in this case of verb-fonns) win be necessary (see Fletcher (1985:177) for a possible fonnat). Secondly, in order to capture significant discriminations between nonnal and impaired individuals, it may be necessary to use categories which are not part of the LARSP set. The requirement to extend categories of analysis in particular instances is underlined in the search for developmental indices in nonnal children of school age. Fletcher and Gannan (1988) report that in the search for syntactic differences between 5- and 7-year olds, the position and function of adverbials (that is, any clause element labelled A on the chart) is important. The older group are more likely to use significantly more temporal adverbials than the 5-year olds. These rather subtle features do discriminate, where more straightforward measures (proportion of complex clauses, proportion ofNP or VP expansions) do not. They are not however represented directly on LARSP profiles (though they are incorporated into SUMLARSP procedures - see below). Computational approaches LARSP has not escaped the advance of the microcomputer, and two types of approach have emerged to the problem of setting up computer profiles. One, exemplified by the system developed at Reading for the standardization project, involves the analysis of data which has already been hand-coded. Achieving an accurate and exhaustive grammatical analysis of a corpus of utterances by computer is an extraordinarily complex problem, which is still a long way from solution (see for example Leech, Garside and Atwell, ]983). For the type of detailed grammatical and lexical analysis we require, human coding provides at present the speediest and most accurate solution. The SUMLARSP programs (see Johnson (1986) for details) provide a variety of detailed analyses of coded individual transcripts. Outputs include: I A standard LARSP profile chart. 2 A table of frequencies of each individual LARSP phrase type, by clause element. 3 A table of adverbials, given by type and position. (In the extended LARSP analysis, these adverbial types are recognized: Time, Place, Manner, Other, and Problem.) 4 A verb-fonn microprofile, showing which verb-fonns occur with each lexical verb in the sample. The second type of computational approach involves partial automation of The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 206 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY analysis procedures, to produce a LARSP profile. One example, from Britain is Automated LARSP, written by Dorothy Bishop (Bishop, 1984; see Klee and Sahlie (1986) for a review). Another, from the United States, is Steven Long's Computerized Profiling (Long, 1986, 1987; reviewed in Klee and Sahlie 1987; LARSP is one of several profiling procedures included in this approach). Both of these procedures are only partly automated, and work through a combination of software-driven analysis and user analysis. Automated LARSP involves the user throughout, for example in assigning grammatical category codes to words which are not already in the program's lookup dictionary. The LARSP part of Computerized Profiling provides an automatic first pass through the data, and then asks the user to verilY the result, and to make appropriate corrections. Clearly a user of either of these procedures will need to either be, or become, an experienced LARSP user. In evaluating these programs Klee and Sahlie note that Automated LARSP is slower, for experienced LARSP users, than analysis by hand. In their critique of Long's program they imply that user correction will be a not insignificant part of the total time involved in the analysis. Adaptations to other languages Apart from its widespread use in English-speaking communities, LARSP has been adapted to a number oflanguages other than English. There seems to have been little difficulty in applying the general structure of the chart to other languages: the organization into sections A, B, C and D, the seven developmental stages, and the bottom line. Such notions as 'intelligible', 'spontaneous', and 'one element sentence', not surprisingly, transfer to other languages with no greater difficulty than we have encountered in English. Extending this list, into formal and substantive cross-linguistic universals of language development, is a formidable task, and one that is, of course, not solely addressed within the LARSP framework. Yet the development of LARSP-type adaptations for other languages is an important exercise, not least for its consistency of descriptive framework within which comparative data may be collected. It is also not surprising to find, when we tum to consider the morpho-syntactic details of the languages in question, that the exercise has not always been easy, since, clearly, simple 'translations' would be quite inappropriate, given the degree of structural difference that may exist between one language and other. The scope of the task may be appreciated by the following consideration. LARSP may be described, essentially, as an attempt to relate the following three components: (l) a body of data to be analysed; (2) a set of descriptive categories, set within a coherent framework that is comprehensive for the language; and (3) a body of specific information on normal language development. There is no shortage of (I), it seems, in any language! But, under (2) and (3), attempting to adapt LARSP to other languages serves to remind us how much the procedure has been able to build on information that is readily available for The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 207 English. The descriptive requirement has been served through the Quirk grammar: this, in the form of UGE and GeE (now supplemented by Quirk, Greenbaum, Leech and Svartvik, 1985), has represented a comer-stone which is simply not available for many languages (see Chapter 3). Further, we have been fortunate in having English as the most comprehensively-studied language in the child acquisition field; more recently, the picture has started to broaden a little, but still only a little, in spite of the achievements documented in, e.g. Slobin (1985). In spite of these difficulties, a number of attempts have been made. We cannot claim any comprehensiveness for the list below, but we are aware of the following (the list indicates principal researchers, affiliation, a rough date to indicate the time at which research was on-going, and, where possible, a bibliographical reference): Bengali 1. Stokes, City University, (MPhil research, in progress), 1987 Danish Dutch French German Hebrew Hungarian Irish Japanese Norwegian Panjabi K Gregersen, Odense University, 1984 G. Bol and F. Kuiken, University of Amsterdam, 1980 (Bol and Kuiken 1980) L. Verhulst-Schlichting, University of Utrecht, 1982 (Verhulst-Schlichting 1982) J. Rondal, University of Liege, 1986 H. Clahsen, University of Dusseldorf, 1983 (Clahsen 1986) R. Berman, A Rom and M. Hirsch, University of Tel Aviv, 1982 (Berman et al. 1982) Csaba Pleh, Eotvos Lorand University, Budapest, 1984 Christine Hayden, MA dissertation, University of Reading, 1984 (Hayden 1984) Martina Hickey, Ph.D. dissertation, University of Reading, (Hickey 1987) Yuri Hishihori, Hokkaido University, 1984 Julie Feilberg, Trondheim University, 1980 Nita Madhani, City University, (MPhil research, in progress), 1987 Welsh Martin Ball, Polytechnic of Wales, 1979 (Ball 1988) It would be impossible in this chapter to review all of these in any detail, or to attempt to compare or evaluate them in a serious way. What we shall try to do is highlight some of the important issues that have arisen, which may serve to improve our understanding of LARSP as well as its cross-linguistic derivatives. The first of these may be found in clause-level representation, and concerns the copula verb in patterns like SVC, SVA in English: that is nice; it's a ball; I'm in the bath S V C SV C SV A These patterns are located at English Stage III, consistently with the generalization The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 208 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY that 3-element patterns are characteristic of this Stage. But what oflanguages that may have a corresponding verbless 2-e]ement pattern, SC, SA? The situation arises, quite typically, in Hebrew. Berman (1978: ]83-230) notes the occurrence of the 'verbless' type in Hebrew grammar, as in the present tense examples: am ayef; I (am) tired S C hen axayot; they (are) sisters S C hem kan they (are) here -- -- S A In addition to these 'zero-copula' structural patterns, there are others where the copula verb haya, from the root h-y-y appears, as in the versions marking the past tense, the future tense, the infinitive, or the gerund. What does one do in this situation? It would appear that the straightforward solution is to recognize two patterns, where English has one, the SCI A as well as the SVC/ A, and to set the 2-element versions a stage earlier (at least, as a working hypothesis) than the 3-element types. This proposal, however, met with the following objection, from some Hebrew speaking clinicians: does this not 'downgrade' the status of the verbless, 2-element patterns, which are in fact the most commonly-encountered forms of these constructions? One suggestion proposed during the discussion was for schemas such as: S(V)C; S(V)A, to be located alongside other 3·element clause patterns in the language. But it quickly emerged that this could actually represent quite different solutions: (0 one version interpreted the (V) to mean a verbal element that mayor may not appear, while (ii) the other took S(V)C/ A to represent just those patterns in which the verb did not overtly occur (but was 'understood' to be present). According to the first view, then, both verbless and verb-specified versions would be tallied under the S(V)C/A schemas, while the second view would set up S(V)C/A for the verbless pattern, distinct from SVC/ A Yet another version (iii) would use SC and SA until the child was judged to be aware of the 'understood' verb, as indicated by its occurrence in certain marked contexts, and then use S(V)C/ A for the verbless forms beyond this stage. It is not difficult to understand the motivation for such suggestions: there is a concern to document progression from a less mature stage to a more mature one, and to distinguish a mature SC or SA in one language from the immature schemas SC, SA in English Stage II. But it is important to realise that such a proposal, however well-motivated, rests on a misconception. The fact that English SC and SA represent immature constructions is not relevant to decisions about how to represent complement and adverbial patterns in other languages. SC and SA are immature in English only because English syntax has the requirement for copula presence in such patterns. Given that LARSP represents surface syntax, it is bound The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 209 to reflect cross-linguistic differences in syntactic requirements such as these. If Hebrew SC is a mature pattern, then, in some sense there is, in this part of the grammar, less for the Hebrew-speaking child to acquire. (Conversely, we may note that English children have less inflectional morphology to acquire than their Hebrew counterparts: such trading relations are much discussed in the cross linguistic acquisition literature.) So, we should locate Hebrew SC, SA along with the other 2-element clause schemas of the language at Stage II; and their three element counterparts, SVC, SVA, at Stage III. Let us now return to another issue, this time at the level of phrase structure. It is frequently the case that English patterns such as DN or PN, involving distinct word elements, are represented in another language by means of affixes (e.g. Norwegian hus, 'house', husel, 'the house'). Should these be interpreted as phrase level patterns? The answer must be that they are word-structures, belonging within a Word column that is much enriched compared to English LARSP. We shall say more of this below. But there are other constructions, exemplified by Japanese noun-based constructions, which are intermediate, apparently, between phrase and word-level types. Japanese nouns appear in construction with postposition elements such as -0, -ga, -wa, -ni, -to, -ka, ya and -e. Some of these are clear counterparts of English prepositions, as in (examples based on Alfonso 1980): mainichi Tookyoo-e ikimasu every day Tokyo-to (I) go 'Everyday I go to Tokyo' while others are more akin to markers of grammatical relationships such as subject (-ga) and object (-0), which are not explicitly marked in English, as in nihontgo-ga joozu (desu) Japanese excellent (is) 'His Japanese is excellent' nihongo-o hanashimasu Japanese (I) speak 'I speak Japanese' A particularly interesting marker in this connection is -wa, which appears to function as a topic identifier, often translatable (at least non-idiomatically) as something like 'as for .. .': anata-wa nihongo-ga joozu you-as for Japanese excellent 'your Japanese is excellent' Now, the issue arises as to whether we are dealing here with constructions at phrase-level or word-level. Assuming that English prepositional constructions are appropriately treated as phrase level types, then the Japanese adverbial postpositional constructions look similar: Tookyoo-e, 'to Tokyo'; sampo-ni, 'for (a) walk'; sonG heya-ni, 'in that room'; gakkoo-kara, 'from (the) school', etc. In other The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 2lO TIlE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY words, we would think of these having structures as shown in the following tree structure (Po = postposition): PP Np/ ""po I N I gakkoo school kara from PP Np/ ~PO D/ ~N I I sono heya ni that room in On the other hand, there are reasons for worrying about this representation, since these postpositions are (unlike English in, on, under, etc.) not able to stand alone, and are traditionally regarded as 'particles' in grammatical structure - not a very well-defined term, but one which tends to apply, as here, to elements which have doubtful status as words in their own right. If, then, we take the essence of phrase-level structuring to involve the organization of one word with another (within but not between, clause-level elements), then these Japanese examples may not satisfy this criterion. As such, they may be better viewed as types of word-level (stem-affix) construction, which happen to mark relationships at phrase-level: that is to say, the particle is attached to the noun, but signals a relationship between the whole noun phrase and the other elements of the clause. We may illustrate these two aspects of structure as follows: /NP~ PP / ~ D N NP Po 1\ D/ ~N stem affix I I I I sono heya ni sono heya ni A solution which may prove workable, in the context ofLARSP adaptation to Japanese, is (i) recognize phrase structure patterns such as NPo, DNPo, etc. (equivalently to PrN, PrDN for English), at least for the more clearly adverbial types (the locatives presumably being especially important). These will then form phrase-level expansions of A-elements at clause level; (ii) treat the abstract relational particles -ga, -0, and -wa as marking Subject, Object and Topic types of clausal element respectively (which may be represented as S, 0 and T), and regard their phrase-level expansion as noun phrases; (iii) list all the particles at word-level, so that their presence or absence can be monitored independently. It should be emphasized that this is a highly tentative proposal, especially The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. LARSP 1975-1990 211 insofar as it concerns the representation of Topic (T) as an element of clause structure. We shall have to await more detailed information on Japanese language acquisition before we can evaluate how far children's early grammatical constructions provide evidence on this issue. More generally, though, we may recommend the inclusion of elements at both word level and phrase level. wherever their precise status is in doubt, at least as a starting point for collecting relevant acquisitional information. Finally, let us consider an issue of word-level representations. Where morphological formations are rich, as in Hebrew and Irish, they are corres pondingly important in the information they carry concerning the grammatical characteristics ofthe overall profile. But frequently such constructions are learned and used as routines, in more-or-less restricted situations. Thus, the issue of developmental sequence may be very difficult to determine for them (as it is, indeed, for the English set of word-level elements). An idea of the complexities that may arise can be gained from the work of Hickey (1987) for Irish. In place of English progressive, -ing, Irish has 'no regular system for forming the progressive; instead the verbal noun is learned lexically for each verb'. Corresponding to the English plural on nouns, 'the same is true for the plural [in Irishl, which is formed in many different ways and which is also probably learned lexically'. Corres ponding to the English perfective form -en, Irish has a 'verbal adjective, [which) like the verbal noun, is probably learned lexically for each verb'. And so on. In summary, 'the Irish system operates not so much at 'Word' level in the sense used by LARSP, but more often at the individuallexeme level' (Hickey 1987: 81-82). Faced with an enormous amount of variation in both affix forms and in the effects they have on the stems they attach to (especially in 'mutating' certain stem final consonants), the temptation may be strong to try to include too much in one's Word column. Hickey decided, first, to restrict her Irish Word column to 27 general categories (accepting the wide range of variation under them that this inevitably involved), and secondly to observe how these categories emerged according to independent measures of early grammatical development of the three children she studied. The Word column types which figured in the samples that were equated at an MLU value of 1.8 (using morpheme-counting criteria developed specifically for the study) were: ta (the present tense of the verb 'to be'), Vn (the verbal noun), an (the singular article), and some noun plurals. This strategy seems to have paid off, in that it was possible, even at this level of generality, to show not only that these were among the earliest types to be used, but that they were quite heterogeneous in their contexts and representation, suggesting that, at an early stage, this level of development is indeed lexically specific. We may briefly summarize the foregoing remarks by putting forward the following guidelines: 1 no direct translation of LARSP can fill the need for language-specific characterizations of surface grammatical properties in other languages; 2 when faced with structural differences (as in the case of verbless as well as verb specified constructions), it pays to represent this diversity as faithfully as possible; The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 212 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY 3 in case of doubt regarding the level-status of particular elements (as, say, between phrase- or word-level status for Japanese particles), it pays to represent them, initially, at least, at both levels; 4 in the face oflexically-diverse word-level features (as in the Irish inflectional system), a reasonable strategy is to set the descriptive categories at a fairly general level (after all, LARSP itself does not attempt to capture lexically-specific aspects of structural patterning in English). A final point to be made in this section is that it may be difficult to resist the temptation to improve or extend LARSP as one attempts to adapt it to another language. While we would sympathize with such a tendency, our advice would be to proceed cautiously; adaptation is a difficult enough task in its own right, and, arguably, extensions to the system represent a further and complex endeavour, as we have seen ourselves in attempting to improve the quantitative aspects of the procedure. The future In the past ten years, the focus has shifted from a relatively narrow perception of the importance of grammatical analysis in language disability to an integrated view, in which grammatical difficulties are seen as interacting with other aspects of linguistic structure and use. The interactions between the kinds of information provided by LARSP and other linguistic levels are taken up in various places. The profile approach to clinical linguistic analysis is developed for phonology and semantics in Crystal (1982a). Fletcher (1985) incorporates information about verb valency into his syntactic description, and also stresses the importance of grammatical difficulty in relation to children with pragmatic problems (1986). Crystal (1983) identifies a LARSP-based syndrome ('Stage V syndrome') involving reference to other levels, and (1987a) reviews the theoretical questions involved in relation to processing models oflinguistic deficit (an issue also addressed by Wren. 1981 ). Another important development has been an increased awareness of the importance oflongitudinal studies of grammatical disability (Crystal, 1981; Wren, 1981). One of us fortuitously met Hugh, the child described in Chapter 7, and his parents some 12 years after our involvement in his therapy. While his parents said they were pleased with the way his language had continued to develop, it was also evident from their remarks that there were occasions when his language was not all it should be. They commented, for example, that he sometimes got confused when he was tired, or got into a tangle when telling a long story. It is evident, from such tantalising remarks, that more precise data is needed to provide diagnostic information about the course of a disorder, and, ultimately about the efficacy of therapy. But very few follow-up studies of grammatical disability have taken place (cf. the discussion in Crystal, 1984, 133ft), and we know of none using LARSP. We very much hope that ways will be found to enable such longitudinal studies to take place. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Appendix A LARSP Child Data Collection Instructions Not less than 15 and not more than 30 minutes of taped material should be obtained for each child, ideally as follows: (a) approx. 15 minutes in an unstructured, free play situation (using toys which do not make too much noise); books, pictures etc. should not be used unless you find yourself with no alternative; interviewer should play with the child in what he considers to be a natural, appropriate way; if the child stays fairly quiet, the session can be turned into a prompted dialogue (asking the child what he's doing, what's happening etc.); (b) approx. 15 minutes of dialogue, on some aspect of the child's experience not to do with the immediate play situation. Exclude the first few minutes of contact with the child from the above times, especially if he is not at ease with the recording situation in some way. The interviewer should be alone with the child. As soon after the recording as possible (preferably, within 24 hours): (i) Fill out the Recording Data Section below; (ii) Listen through the tape, and write out as much of the child's utterances as you can, concentrating especially on stretches which may cause an outside listener difficulty (e.g. due to immature articulation, family Slang), and giving a gloss to those utterances which may not be clear out of context (e.g. give me that = give me the toy dog; fall down = his lego house has just fallen down; doggy = he has just caught sight of his dog); (iii) write your utterances and each of the child's on separate lines, e.g. Int. What's that you've got? Ch. It's a car. Look, it's making a noise. Int. Can I have one? Have you got one for me? etc. (iv) Fill out the Child Data Sheet. Recording data sheet 1 Where did the recording take place? 2 Date of recording. 3 Anything abnormal in the child's general behaviour, health etc. ? 4 Anything abnormal in the situation, which may have influenced the way he reacted, and which is not obvious from the tape? The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 214 lHE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILIlY Child data sheet Name: 1 Date of birth: 2 Sex: 4 Age of father: of mother: 6 Occupation of father: of mother: 3 Age and sex of sibs: 5 Where living now: 7 Does either parent have a noticeable regional accent? 8 Have either any obvious speech/hearing impediment? 9 Child's medical history: normal birth? any long stays in hospital? any major disability/illness? to Any school/nursery/creche etc. attendance? (state what kind and how long) 11 Is the child in regular contact with other adults at home? (state relationship) 12 Does the child have any contact with languages other than English? (state which) 13 Give any psychological testing scores which may be available: 14 Any other information you consider relevant: Appendix B Child Language Assessment Sampling Procedure Patient: Sex: Therapist: Date of birth: Place of recording: Referred by: Date and time: Reason for referral: Background information (to be obtained before first session): 1 Date and sex of sibs: 2 Age of father: of mother: 3 Where living now: 4 Father's occupation: 5 Does either parent have a noticeable regional accent? (state which): 6 Is the child in regular contact with other adults at home? (state relationship): 7 Does the child have any contact with languages other than English? (state which): 8 Abnormal social circumstances in family background: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. APPENDIX B: CHILD LANGUAGE ASSESSMENT SAMPLING PROCEDURE 215 9 Medical history in family (especially details of speech/language/learning disorder among parents or close relatives): 10 P's medical history (including pre- or postnatal, major disabilities or illnesses, periods of hospitalization): II Any noteworthy features of developmental history: 12 Assessment results already obtained: (a) Hearing tests, if any (give date, details of performance): (b) Language tests, if any (give test used, date and result): (c) Psychological tests, if any (give test, date and result): 13 School/nursery/creche etc. attendance (state what kind and how long): 14 Details of previous therapy, if any (when, where, by whom, why): 15 Do you have any views about therapeutic procedures to follow in the long term? During the session As soon as P has settled, obtain not less than 15 and not more than 30 minutes of taped material, ideally as follows: (a) approx. 15 minutes in an unstructured, free play situation (using toys which do not make too much noise); books, pictures, etc. should not be used unless you find yourself with no alternative; play with the child in what you consider to be a natural, appropriate way; if the child stays fairly quiet, the session can be turned into a prompted dialogue (asking the child what he's doing, what's happening etc.); (b) approx. 15 minutes of dialogue, on some aspect of the child's experience not to do with the immediate play situation. Exclude the first few minutes of contact with the child from the above times, especially if he is not at ease with the recording situation in some way. At the end of the session. establish whether it would be possible for P to be taped at home, talking with parent or sib. One or a series of recordings totalling about 10 minutes would suffice. After the session Listen through the tape, and write out as much of the child's utterance as you can, concentrating especially on stretches which may cause an outside listener difficulty (e.g. due to immature articulation, family ~Iang), and giving a gloss to those utterances which may not be clear out of context (e.g. gi~'e me that = give me the toy dog; fall down = his lego house has just fallen down; doggy = he has just caught sight of his dog). The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 216 THE GRAMMATICAL ANALYSIS OF lANGUAGE DISABILITY 2 Write your utterances and each of the child's on separate lines, e.g. T What's that you've got? p It's a car. Look, it's making a noise. T Can I have one? Have you got one for me? etc. 3 Answer the following questions about· the session: (a) Anything noteworthy in the remainder of the session? (b) Were other people present at any stage? (state whom): (c) Rate the session for typicality, as far as you can, in terms of P's normal health, behaviour etc. : (d) Comment on any notable fluctuations in P's behaviour throughout, especially if the variations are not explicable by reference to the tape: (e) Did P react to the tape-recorder at all? (f) Note any linguistic features which struck you as interesting, and to which you think our attention should be drawn: (g) Evaluate the session (e.g. How well do you think it went? How easy did you find it 1) (h) Evaluate personal attitudes relating to P. (Do you get on well with him? How does P respond to you 1 Any noteworthy parental attitudes towards him or his difficulty 1) Next session planned, if any: Date: Venue: Aims: The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Appendix C Communicative History Form Name of patient: Age: Address: What relationship do you have to the patient? (e.g. wife, son, nurse) Who lives with the patient? (e.g. wife, lives alone, lives in old people's home) Family: Name of wife/husband: Children's names Where do Names of sons Wife's name and ages they live? I 2 3 Children's names Where do Name of daughters Husband's name and ages they live? I 2 3 Names of close relatives (brothers, sisters, in-laws etc.) and any relevant details; only list those whom the patient is likely to mention regularly: 1 2 3 4 5 6 Name Relationship Where living? The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 218 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Names offriends, and any relevant details (e.g. workmates, neighbours): 2 3 4 5 6 List any nicknames or special names used in the patient's family which might come up in conversation with us: Pets: Give names and type of pet belonging to the patient's family: Does the patient see any of the above often? Has anything happened to one of the above recently which might be on the patient's mind? Have you noticed any change in the attitude of any of the above towards the patient as a result of the illness? What sort of a person was the patient before the illness (e.g. quiet, talkative, serious, humourous, lively, short-tempered, thoughtful)? Use as many labels as you can think of. Have you noticed any changes in the patient's behaviour since the illness? Does the patient express any strong feelings as to why his present condition has arisen? If the patient is physically disabled in any way, as a result of the illness, does he/she express any particularly strong feelings about this? Is there any particular activity that the patient is no longer able to carry out which causes particular anxiety? What was the patient's job? Had he/she retired? Was the patient ever in the Services? Give details. Is any work attempted now? Which sports, teams etc. was the patient interested in, if any? Did he/she play any sport before the illness? The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. APPENDIX C: COMMUNICATIVE HISTORY FORM 219 Name any clubs or societies that the patient belonged to. Was the patient a churchgoer? Give details. Television: Programmes and personalities particularly liked and disliked: Radio: Programmes and personalities particularly liked and disliked: Music: Is the patient interested in any particular kind of music? Give details. Does the patient sing, play etc. ? What paper(s) did the patient read regularly before the illness? Were any magazines read regularly? Are any of these still read? Did the patient have any favourite books or authors? Does he/she wear glasses for reading? (If so, would you please make sure that the patient brings them to therapy sessions.) Hobbies (e.g. knitting, gardening, films): give details. Are there any places of particular interest to the patient? (e.g. holiday haunts, week-end visits) Has the patient ever been abroad? Give details. Has the patient ever lived anywhere else for a period of time? Give details. Can you think of anything else we should know about which might stimulate the patient's interest, bring back memories, and generally help to encourage com munication? Can you think of any topics which ought to be avoided, because of their painful associations for the patient? This form should be returned to the speech therapist. Note: This fonn is a synthesis and elaboration of ideas we have seen in use in various clinics: we acknowledge here our thanks to those therapists and teachers who have helped in its construc tion. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Appendix D Sample of syntactic analysis of adult speech A Sentence: I spotted you in town yesterday, John. S V 0 A A Voe Phrase: Pron Pron' Pr N (Minor) Word: -ed B Where? S: Q A I was in the garage behind the police station, and I saw S: S V A e S V P: Pron Pr D N Pr D W: -ed you in the market. S: 0 A P: Pron Pr D N W: N N Pron -ed B Yes. We were trying to find some new curtains. S: Minor S V 0 P: Pron Aux v v D Adj N W: -ed -ing -pl. But what were you doing in the garage? S: c Q S V A P: Aux Pron v Pr D N W: -ed -ing A Looking for a man who could fix my car. ~ V 0 P: W: v part D N -ing s V 0 Aux v D N I think the back wheel's falling off. S: S V 0 ------------------S V ----- P: Pron D Adj N Aux v part W: 'aux 3s -ing The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. -- - -- APPENDIX D: SAMPLE OF SYNTACfIC ANALYSIS OF ADULT SPEECH 221 B Gosh! Did you succeed '1 S: Minor S V P: Aux Pron v W: -ed A I found a very helpful mechanic, but unfortunately S: S V 0 c A P: Pron W: DInt -ed Adj N he didn't have all the parts. S: S V 0 P: W: Pron Aux Neg v I D n't N pl. B S: That's always the problem. What a nuisance! S V A C Minor P: Pron Cop D N W: 'cop 3s -Iy A And there's nothing I can do until he gets some from the factory. S: and there V C A S V s S V 0 A P: Cop Pron Pron Aux v Pron W: 'cop 3s B If I can help in any way, let me know. S: A V imp 0 sS V A P: Pron Aux v Pr D N Aux: Pron v Pron Pr D 3s N I shan't be using my car next week, so if you'd like to borrow it, S: S V 0 A A conn. Aux s S V 0 P: PronNegAux v D N Adj N Pron Aux v v Pron W: n't -ing S: A S: P: W: please do. A Vimp Thanks very much. Minor But I think it'll be ready. c S V 0 S V C Pron Pron Aux Cop 'aux 'aux The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Name A B Vnanalysed ! UninlelllglbJe Responses SIImulu5 T:ope R QueSlions Others C Spontaneous Age 2 Symboll!.: Noise Sample dale In(;om~ pic It Normal Response Type 2 Arnbi~­ U0US J ~tereo­ types I Abnormal Red-I Struc- uced Full Minor tural Repet- Elliptical T Ola lsi l ion s 1--'--'-'-:2:--r- 1 -::-3 -+-1 I I I I Prob lems D Reactions General Structural 0 I Olher I Probkms I I I -:i; Minor QI":": R~spollses FOCGlin'J OI/ler Pl'Oblen1s ... I Comm, Quest ~ ~ :\-1ajor f-----+-----l--------------- - 'v 'Q' 'v Slatemelll 'r-,' Other Pro!Jlems Conn v, Q..I - It'l X Y VSln do \ i - X ItS NI' + S QVS Q,\ I Idg. <111(1 C(lmu { (lMd ~:c ~~; Olher 0<.J. ,,- Olhel ~~ Q,!1(!f ;+1 NP VP ;~ inltldl,lf ~1 Co<}r .. 1 .. ,. !i:~ Olh~1 Dn((lun(' Clause 5V SO SC 1>.eg + V;VP r SVC I I :~: i'ocg ,\ r .\ } + V,VP - SVOA SVCA SVOp SVOC (o0nl ~uh()n.l A s CnmpJrd\1\C Phrase 'u DN VV V() Ad, N \' pOln VC NN 1m A Olner Per-. Other VCA DAd] 1>. Cop VOA Ad] AdJ r-. AU\~I VOp Pr D~ Other Pron;, Olhn .\1 C~P \l-O~P Il+AAP ------------------- AA..X} ~p Pi !'.I J Nc~ \ O~hcr I + i -t o Pn..,tn1"l: dau~ P()~Lnll)d J + rh1d'>l· ) Clall,e Conn Clause rkm..:n! NP D Pc lit'll I1/Uf! Word i pi -ed -en ·h - The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. l -..,--,- 2::-"--'3:-+ l . · · ' ,' I\"i . , . ! r '", !>.c , . - ve ! ,l!>' , ) Cr-,P \l-ONP I " ,I Revised (1981) LARSP Chart: summary of changes As pointed out in the Preface (p. 3), and noted at various places throughout the book, increased use of the procedure has motivated several suggestions for change. We have kept these changes to a minimum, bearing in mind that many clinicians have only recently come to terms with our initial statement. In most cases, the changes help to remove ambiguities encountered in working with the original chart. In a few cases, we have added categories which we have found to be particularly helpful. The most noticeable changes are in the addition of a Section D, and the restructuring of the Stage VI 'error box': the former has especial relevance in devising remedial strategies; the latter we have found of great importance in the assessment of certain clinical populations-of the deaf, in particular. In a few cases, we have removed a category from the original chart, because it has proved not to be as helpful as we had anticipated. The overall 'look' of the chart, however, is the same-and indeed, it would be possible for Ts who wish to continue with the unrevised version to do so (e.g. by ignoring the distinction between Aux M and Auxo, and continuing to use the simple category Aux, as before). We recommended use of the various modifications wherever possible, though, as we feel they help to make the procedure more discriminating and (slightly!) easier to use. Change I. Name of patient, Age, Sample date and type printed on top of chart. 2. Section A Stereotypes added as a further category in the Problematic box. 3. Section B Elliplical Major 4 deleted, and subsumed under 3, which is now printed as 3 +. A new category, Reduced Major in trod uced. ReflSOll for change To facilitate routine clinical reference, and to stress the importance of noting sample type. To enable a more coherent analysis to be made of Stage I, where P might use stereotypes of considerable length. To handle sentences which are partially stereotyped, and which currently have nowhere to go on the chart (except, unsatisfactorily, into Stage VI Error). Hardly ever used. To avoid the frequent problem of how to handle sentences which omit clause elements, but which are not elliptical (i.e. retrievable from the linguistic context). The main problems have been the Stage II 'immature' sentences, viz. Sc. SO, Neg X and some AX. Al present, for example, daddy a car and going 10 10wn would both be called Elliptical Major 2 responses. lUustrarion of new procedure John Smith 3;2 6 Jan 81 12 mins free conversation (with toys and picture) The whole sentence (whether wholly or partially stereotyped) put into Section A. E.g. (from some deaf writing): My brother and me saw a cal. Mv brother and me saw a dog. My brofher and me ... etc. Reduced Major sentences are marked Separately, e.g. (in reply to T what's happening): man in garden, man ball (i: in children from 5 to 10. Cambridge, Mass.: MIT Press. CHOMSKY. N. 1957a: Syntactic struClures. The Hague: Mouton. - 1957b: Review of B.F. Skinner, Verbal behavior. Lg. 35, 26-5K - 1965: Aspects of the theory of syntax. Cambridge. Mass,: MIT Press. - 1968: Language and mind, New York: Harcourt, Brace and World. - 1970: Deep structure, surface structure and semantic interpretation. In R. lakobson and S. Kawamoto (eds,). Studies in general and orienla/linguistics. TolJ'o. CLAHSEN, H. 1986: Die Profilana{vse. Berlin: Marhold. CLARK, R., HUTCHESON, S. and VAN BUREN, P. 1974: Comprehension and production in language acquisition. JL 10, 39-54. CLDP 1979: Cambridge Language Development Project. EARO, The Resource and lechnology Centre. CONN, P. 1971' Remedial syntax. London: Invalid Children's Aid Association, Occasional Papers l. CONNOLLY,1.H 1984: A commentary on the LARSP procedure. BlDC 19, 63-71. COOPER R.L 1967: The ability of deaf and hearing children to apply phonological rules. JSHR 10,77-86, COWAN, PA, WEBER, 1., HADDINOTT, BA and KLEIN, 1. 1967: Mean length of spoken responses as a function of stimulus, experimenter and subject. Ch. Dev. 38, 191-203. CROMER, RF. 1970: Children are nice to understand: surface structure clues for the recovery of a deep structure. HJ Psych. 61, 397-408. CRUITENDEN, A 1974: An experiment involving comprehension of intonation in children from 7 to 10. J. Ch Lang 1,221-31. CRYSTAL. D. 1966: English. In Word classes, special volume of Lingua 17,24-56. - 1969: Prosodic systems and intonation in English. Cambridge: Cambridge University Press. 1971a: Linguistics. Harmondsworth, Middx.: Penguin, 2nd edn. 1985. - 1971b: Prosodic systems and language acquisition. In P. Leon (ed.), Prosodic feature analysis. Montreal: Didier, 77-90. - 1972a: The case of linguistics: a prognosis. BlDC 7, 3-16, - 1972b: Syntax matters. In Proceedings of 1972 National Conference of Teachers of the Deaf. - 1973a: Nonsegmental phonology in language acquisition: a review of the issues. Lingua 32, 1-45. 1973b: Basic linguistics and Language acquisition. Cassette tapes and slides. Reading University: Reading Centre. - 1974a: Neglected linguistic principles in the study of reading. In UKRA Proceedings of 11th Annual Study Congress. - 1974b: Review of R. Brown, A first language. JCh.Lang. 1,289-307. - 1979: Working with LARSP. London: Edward Arnold. 1981: Research trends in the study of child language disability. Symposium on Research in Child Language Disorders, University of Wisconsin-Madison, 1-21. - 1982a: Profiling linguistiC disability. London: Edward Arnold. - 1982b: Terms, time, and teeth. BlDC 17, 3-19. - 1983: Clinical psycholinguistics. Folia Phonialrica 35, 1-12. - 1984a: Linguislic encounters with language handicap. Oxford: Blackwell. - 1984b: Suffering a relarsp: a reply to ConnoUy. BlDC 19, 72-7. - 1986: The diagnosis of language disorders in children. Medical Research Cou ncil Project Report (G 8306096 NA). - 1987a: Towards a 'bucket' theory of language disability: taking account of interaction The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. • REFERENCES 231 between linguistic levels. Clinical Linguistics & Phonetics 1, 7-22. 1987b: Meeting the need for case studies. Child Language Teaching and Therapy 3, 305-10. - 1988: Rediscover grammar. London: Longman. CRYSTAL. D. and CRAIG. E. 1978: Contrived sign language. In 1. Schlesinger and L. Namir (eds.), Current trends in the study of sign language of the deaf New York: Academic Press, 141-68. CRYSTAL. D. and DAVY, D. 1969: Investigating English style. London: Longman. - 1975: Advanced conversational English. London: Longman. CRYSTAL D. and FOSTER. J. 1979-83: Databank. London: Edward Arnold. DALE, P.S. 1972: Language development: structure and function. Hinsdale, Ill.: Dryden Press. DARLEY, F and MOLL, K. 1960: Reliability of language measures and sizes of language sample. J,\llR 3, 166-73. DAVIS, EA 1938: Developmental changes in the distribution of parts of speech. Ch. Dev. 9, 309-17. DAvlS, GA 1973: Linguistics and language therapy: the language construction board. JSllD 38,205-14. DENNIS, D.B. 1981: Structure in language teaching. In A Jackson (ed.), Ways and means, Vol. 3. Basingstoke: Globe Education. DEPARTMENT OF EDUCATION AND SCIENCE. 1972: Speech therapy services. London: HMSO. DERWIN, B.L. 1973: Transformational grammar as a theory of language acquisition. Cambridge: Cambridge University Press. DEVER. R.B. 1972a: TALK (Teaching the American Language to Kids). Experimental Materials. Final Report 27.3. Bloomington, Indiana University: Center for Innovation in Teaching the Handicapped. - 1972b: A comparison ofthe results of a revised version ofBerko's test of morphology with the free speech of mentally retarded children. JSHA 15, 169-78. - 1974: The deve]opmentofthe clause: toward a unified theory oflanguage development in children. Mimeo. DEVER, R.B. and BAUMAN, P.M 1971: Scale of children's clausal development. Mimeo. Reprinted in Longhurst 1974. DEVER, R.B. and GARDNER, W. 1971: Performance of normals and retardates on Berko's test of morphology. Lg and Sp. 13 162-81. DONALDSON. MC and WALES. RJ. 1970: On the acquisition of some relational terms. In Hayes 1970, 235-68. DORE, J. 1975: Holophrases. speech acts and language universals. J.ChLg 2,21-40. EASTERBROOK, A., BYERS BROWN, B. & PERERA. K 1982: A comparison of the speech of adult aphasic subjects in spontaneous and structured interactions. BJDC 17, 93-107. EISENSOJ'.i,1. and INGRAM, D. 1972: Childhood aphasia - an updated concept based on recent research. Acta Svmbolica 3, 108-16. ELLSWORTH, R.B. 1951: Th~ repression of schizophrenic language. J.Consu/t. Psychol 15, 387-91. ELSON. B. and PIC KElT, v. 1965: An introduction to morphology and syntax. Santa Ana, Calif.: SIL. ENGLER. LF., HANNAH. E.P. and LONGHURST. TM. 1973: Linguistic analysis of speech samples: a practical guide for clinicians. JSHD 38, 192-204. FERGUSON,CA andSLOBIN. OJ. (eds.) 1973: Studies of child language development. New York: Holt, Rinehart and Winston. FILLENBAUM,S.,JOl"ES.LY. and \VEPMAN.1.M. 1961: Some linguistic features ofspeech from aphasic patients. Lg. and Sp. 4,91-108. FLETCHER, P 1985: A child's learning of English. Oxford: Blackwell. 1986: Semantic-pragmatic disorders: an approach from syntax. In Ad~ances in The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 232 THE GRAMMATICAL ANALYSIS OF LANGUAGE DlSABILIlY working with language-disordered children. London: ICAN. FLETCHER. P. and GARMAN. M. 1988: Normal language development and language impairment: syntax and beyond. Clinical linguistics & Phonetics 2.97-114. f FLETCHER, p, and PETERS.] 1984: Characterizing language impairment in children: an exploratory study. l.l1nguage Testing 1. 33-49. FLETCHER. Po, PETERS. ] and HIXSON, P. 1982: Sampling situations and language assessment. Paper presented at the Annual Convention of the American Speech Language-Hearing Association, Toronto. FLETCHER, p,. GARMAN, M" JOHNSON, M., SCHELLETTER. C. and STODEL, L. 1986: Characterizing language impairment in terms of normal language development: advantages and limitations. In Proceedings of the Seventh Annual Wisconsin Symposium on Research in Child l.l1nguage Disorders. Madison: University of Wisconsin-Madison. FODOR, 1. and GARRETT, M. 1966: Some reflections on competence and performance. In Lyons and Wales 1966, 135-62. FOSTER, C.R. GIDDAN, lJ. and STARK, J, (n.d.): ACLC: Assessment of Children:~ l.l1nguage Comprehension. Palo Alto: Consulting Psychologists Press. FRASER, G.M. and BLOCKLEY, 1. 1973: The language-disordered child. Slough, Bucks.: NFER. FYGETAK1S. LJ. and INGRAM, D. 1972: Language rehabilitation and programmed conditioning: a case study. PRCLD 4, 169-78. GARMAN, M. 1988: The role oflinguistics in speech therapy: assessment and interpretation. In P. Grunwell and A James (eds.), The functional evaluation of language disorders. Beckenham: Croom Helm, GILLIES, M. and HUIT, E. 1985: l.l1nguage through reading. London: ICAN. GIMSON, AC. 1980: An introduction to the pronunciation of English, 3rd edn. London: Edward Arnold. GLEASON,l.B. 1973: Code switching in children's language. In Moore 1973, 159-67. GODA, S. 1964: Spoken syntax of normal, deaf, and retarded adolescents. JJlLVB 3, 401-5. GOODGLASS, H. 1968: Studies on the grammar of aphasics. In S. Rosenberg and J. Koplin (eds.),Developments in applied psycholinguistics research. New York: Macmillan, 177-208. GOODGLASS, H., FODOR, 1.0. and SCHULHOFF. C. 1967: Prosodic factors in grammar - evidence from aphasia. JSHR 10, 5-20. GOODGLASS. H, and HUNT.J. 1958: Grammatical complexity and aphasic speech. Word 14, 197-207. GOTTS LEBEN. RH., TYACK, D. and BUSCHINI, 0. 1974: Three case studies in language training: applied linguistics. JSHD 39. 213-23. GRAHAM. IT and GRAHAM, L.W 1971: Language behavior of the mentally retarded: syntactic characteristics. AJ.Ment.Def 75, 623-9. GREENFlELD, P .• SMITH. 1. and LAUFER, P. 1976: Communication and the beginnings of language: the development of semantic structure in one-word speech and beyond. New York: Academic Press. GRIFFITH,I, and MINER, L.E.1969: LCI reliability and size oflanguage sample.!. Comm. Dis. 2, 264-7. HADEN, R and PENN. C 1985: The twin situation and its effects on syntax and interactional language over time. BJOC 20, 19-29. HAHN. E. 1948: Analyses ofthe content and form of the speech of first-grade children. QJ.Sp. 34,361-6. HART, B. 1975: The use of adult cues to test the language competence of young children. !.Chl.l1ng. 2, 105-24. HASS. WA and WEPMAN,1.M. 1969: Surface structure, deep structure, and transformations: a model for syntactic development. JSHD 34, 303-1 L HATFlELD. EM. 1972: Looking for help from linguistics. BJOC 7,64-81. HAWKINS, p, and SPENCER, H. 1985: Imitative versus spontaneous language assessment: a comparison of CELl and LARSP. BJOC 20, 191-200. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. REFERENCES 233 HAYDEN,C. 1984: An adaptation to Irish ofLARSP. Unpublished MA thesis, University of Reading. HAYES,1.R. (ed.) 1970: Cognition and the development of language. New York: Wiley. HEIDER. ER and HEIDER, G.M. 1940: A comparison of sentence structure of deaf and hearing children. Psychol. Monog. 52, 42-103. HERRIOT, P. 1969: - e comprehension of tense by young children. Ch. Dev. 40,103-10. HICKEY. T. 1987: 1 ~ early acquisition of Irish: grammatical patterns and the role of fonnulas. Unpublished PhD thesis, University of Reading. HOCKEIT, C.F. 1958: A course in modem linguistics. New York: Macmillan. HOCKEIT, C.E and ALTMANN. SA 1968: A note on design features. In T.A Sebeok (ed.), Animal communication. Bloomington: Indiana University Press, 61-72. HOWES, D. 1967: Some experimental investigations oflanguage in aphasia. In K. Salzinger and S. Salzinger (eds.), Research in verbal behavior and some neurophysiological implications. New York: Academic Press, 181-96. HUNT, K. 1964: Differences in grammatical structures written at three grade levels. the structures to be analysed by transfonnational methods. Report to US Office of education, cooperative Research Project 1998. Tallahassee, Florida. - 1970: Syntactic maturity in schoolchildren and adults. Monogr. Soc. Res. Ch. Dev. 35. HUTCHINSON. M.KF. 1972: An experiment in applying linguistics to speech therapy.BlDC 7, 49-53. HurT. E. 1986: Teaching language-disordered children: a structured curriculum. London: Edward Arnold. HurTENLOCHER. 1. 1974: The origins of language comprehension. In RL. Solso (ed.), Theories in cognitive psychology: the Loyola Symposium. New York: Wiley. HUXLEY, R. 1970: The development of the correct use of subject personal pronouns in two children. In G.B. Hores d'Arcais and W J,M. Levelt(eds.),Advances in psycholinguistics. London: North-Holland, 141-65. INGRAM. D. 1972a: The acquisition of the English verbal auxiliary and copula in nonnal and linguistically deviant children. PRCW 4, 79-91. - 1972b: Language program for linguistically deviant children. Stanford University, School of Medicine, unpublished. - 1972c: The development of phrase-structure rules. Long. Learning 22, 65-77. - 1972d: The acquisition of questions and its relation to cognitive development in normal and linguistically deviant children: a pilot study. PRCLD 4, 13-18. JESPERSEN, O. 1922: lAnguage: its nature. development. and origin. London: Allen and Unwin. JOHNSON, M. 1986: A computer-based approach to the analysis of child language data. Unpublished PhD thesis, University of Reading. JOHNSON, W., DARLEY, F.L., and SPRIESTERSBACH, D.C. 1963: Diagnostic methods in speech pathology. New York: Harper and Row. JONES. L.Y. and WEPMAN,J.M. 1967: Grammatical indicants of speaking style in normal and aphasic speakers. In K.. Salzinger and S. Salzinger (eds.), Research in verbal behavior and some neurophysiological implications. New York: Academic Press, 169-80. JONES, L.Y., GOODMAN, M.F. and WEPMAN, J,M. 1963: The classification of parts of speech for the characterization of aphasia. Lg. and Sp. 6, 94-108. lOOS, M. (ed.) 1957: Readings in linguistics I. London: University of Chicago Press. JORDAN. C.M. and ROBINSON, W.P. 1972: The grammar of working and middle-class children using elicited imitations. Lg. and Sp. 15, 122-40. KEENEY. TJ. and WOLFE. 1. 1972: The acquisition of agreement in English.JVLVB 11, 698- 705. KILLING LEY, S-Y. 1981: A note on certain difficulties in using GALD.Nottingham Linguistics Circular 10, 193-7. KIRK. SA 1966: The diagnosis and remediation of psycholinguistic disabilities. University of Illinois: Institute for Research on Exceptional Children. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 234 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY KlRK SA and KIRK W.D. 1971: Psycholinguistic learning disabilities: diagnosis and remediation. Chicago: University of IlIinois Press. KlRK SA, McCARTHY, 1. and KlRK w.n 1968: The Rlinois Test of Psycholinguistic Abilities. Urbana: University of Illinois Press. KLEE, T. and FITZGERALD, M.D. 1985: The relation between grammatical development and mean length of utterance in morphemes. J Ch.Lang 12, 251-69. KLEE. T and SAHLIE. E 1986: Review of Automated LARSP. Child Language Teaching and Therapy 2, 97-100. 1987: Review of Computerized Profiling. Child Language Teaching and Therapy 3, 87- 94. KLIMA, ES. and BELLUGI, U. 1966: Syntactic regularities in the speech of children. In Lyons and Wales. 1966. LACKNER. J. 1968: A developmental study of language behavior in retarded children. Neuropsychol. 6,301-20. LEE, L. 1966: Developmental sentence types. JSHD 31, 311-30. - 1969: The Northwestern Syntax Screening Test. Evanston, Ill.: Northwestern University. - 1974: Developmental sentence analysis: a grammatical assessment procedure for speech and language disorders. Evanston, Ill.: Northwestern University. LEE. L. and CANTER. S.M. 1971: Developmental sentence scoring: a clinical procedure for estimating syntactic development in children's spontaneous speech. JSHD 36,315-40. LEE, L., KOENIGSKNECHT, RA and MULHERN, S.T. 1975: Interactive language development teaching: the clinical presentation of grammatical structure. Evanston, Ill.: Northwestern University. LEECH, G .. GARSIDE, R. and ATWELL, E. 1983: Recent developments in the use of computer corpora in English language research. Transactions of the Philological Society 23-40. LENNEBERG, EH. 1%7: Biological foundations of language. New York: Wiley. LENNEBERG. EH. NICHOLS.IA and ROSENBERGER, E.F. 1964: Primitive stages oflanguage development in mongolism. In Disorders of Communication 42: Research Publications. Baltimore, Maryland: Williams and Wilkins. LEONARD. L.B 1972: What is deviant language? JSHD 37, 427-46. LEREA, L 1958: Assessing language development. JSHR 1, 75-85. LIMBER. 1 1973: The genesis of complex sentences. In Moore 1973, 169-85. LONG, S,H. 1986: Computerized profiling. PO Box 740, Arcata, LA 95521, USA - 1987: 'Computerized profiling' of clinical language samples. Clinical Linguistics & Phonetics 1. 97 -105. LONGHURST, TM. (ed.) 1974: Linguistic analysis of children s speech: readings. New York: MSS Infonnation Corporation. LONGHURST, TM, and SCHRANDT. TAM. 1973: Linguistic analysis of children's speech: a comparison of four procedures. JSHD 38, 240-49. LOZAR, 9., WEPMAN, JM and HASS, W, 1973: Syntactic indices of language use of mentally retarded and nonnal children, Lg. and Sp. 16, 22-33. LYONS. 1. 1963: Structural semantics. Publications of the Philological Society 20. Oxford: Basil Blackwell. - 1968: Introduction to theoretical linguistics, Cambridge: Cambridge University Press. - ]970: Chomsky. London: Fontana, - 1973: Human language. In RA Hinde (ed.), Nonverbal communication, London: Cambridge University Press, 49-85. LYONS, 1. and WALES, R (eds.) 1966: Psycholinguislics papers. Edinburgh University Press, MACOONALDJD. and NICKOLS, M. 1974: Environmental language inventory: a semantic-based assessment for training generalized communication. Ohio State University: The Nisonger Center. MACKAY, DD, THOMPSON, 9. and SCHAUB, 1. 1970: Breakthrough to literacy. Teacher's ManuaL London: Longam, MACKAY. WF 1965: Language teaching analysis. London: Longman. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. REFERENCES 235 MATTHEWS. P.H. 1974: Inflectional morphology. Cambridge: Cambridge University Press. MAXIM. J. 1985: A grammatical analysis of the language of the senescent. Unpublished PhD dissertation, University of Reading.. MAXIM, 1. and THOMPSON, I. forthcoming: Elderly speakers. London: Cole and Whurr. McCARTHY. D. 1930: The language development of the pre-school child. Minneapolis: University of Minnesota. - 1954: Language development in children. In L. Carmichael (ed.), Manual of child psychology. New York: Wiley, 492-630. McCAWLEY, J.D. 1968: The role of semantics in a grammar. In E. Bach and R. Harms (eds.), Universals in linguistic theory. New York: Holt, Rinehart and Winston. McGRATH. C.O. and KUNZE, L.H. 1973: Development of phrase structure rules involved in tag questions elicited from children. JSHR 16,498-512. McNEILL, D. 1966: Developmental psycholinguistics. In F. Smith and GA Miller (eds.), The genesis of language. Cambridge, Mass.: MIT Press, 15-84. - 1970: The acquisition of language. New York: Harper and Row. MENYUK.. P. 1963: Syntactic structures in the language of children. Ch. Dev. 34,407-22. - 1964: Comparison of grammar of children with functionally deviant and normal speech. JSHR 7, 109-21. - 1969: Sentences children use. Cambridge, Mass.: MIT Press. - 1971: The acquisition and development of language. Englewood Cliffs: Prentice-Hall. MENYUK, P. and LOONEY, P.L. 1972: A problem oflanguage disorder: length versus structure. JSHR 15, 264-79. MILLER,1.F. 1973: Sentence imitation in preschool children. Lg. and Sp. 16,1-14. MILLER, J.F., KLEE, T.M., PAUL, R and CHAPMAN, RS. 1981: Assessing language production in children. London: Edward Arnold. MILLER, J.F. and YODER. D. 1974: Teaching language to retardates. In R.L. Schiefelbusch and L.L. Lloyd (eds.), Language perspectives - acquisition, retardation and intervention. Baltimore, Maryland: University Park Press. MINER. L.E. 1969: Scoring procedures for the length-complexity index: a preliminary report. 1. Comm. Dis. 2, 224-40. MlNIFlE, F., DARLEY. F. and SHERMAN, D. 1963: Temporal reliability of seven language measures. JSHR 6, 139-48. MITTLER, P .. JEFFREE, D., WHELDALL, K. and BERRY, P. 1974: Assessment and remediation of language comprehension and production in severely subnormal children. University of Manchester: Hester Adrian Research Centre. MOORE, T.E. 1973; Cognitive development and the acquisition of language. New York: Academic Press. MOREHEAD, D.M. 1972: Early grammatical and semantic relations: some implications for a general representational deficit in linguistically deviant children. PRCLD 4, 1-12. MOREHEAD, D.M. and INGRAM, D. 1973: The development of base syntax in normal and linguistically deviant children. JSHR 16, 330-52. MOREHEAD, DM. and JOHNSON, M. 1972: Piaget's theory of intelligence applied to the assessment and treatment of linguistically deviant children. PRCLD 4. 143-61. MORLEY. HJ. 1960: Applying linguistics to speech and language therapy for aphasics. Lang. Learn. 10, 135-49. MOLLER, D.1., MUNRO, S.M. and CODE. C. 1981: Language assessment for remediation. London: Croom Helm. MUMA. J.R 1971: Syntax of preschool fluent and disfluent speech: a transformational analysis. JSHR 14. 428-41. - 1973a: Language assessment: the Co-occurring and Restricted Structure procedure (CORS). Acta Syrnbolica 4, 12-29. - 1973b: Language assessment: some underlying assumptions. ASJL4 15,331-8. MYERSON. Rand GOODGLASS, H. 1972: Transformational grammars of aphasic patients. Lg. and Sp. 15, 40-50. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 236 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY MYKLEBUST, H. 1965: Development and disorders of written language. New York: Grune and Stratton. NELSON, KE., CARSKADDON, G. and BONVILLIAN, 1.0. 1973; Syntax acquisition: impact of experimental variation in adult verbal interaction with the child. Ch. Dev. 44,497-504. NEWF1ELD, M.u. and SCHLANGER. B.B. 1968: The acquisition of English morphology by normal and educable mentally retarded children. JSHR 11, 693-706. NICE, M.M. 1925: Length of sentences as a criterion of a child's progress in speech. J Educ. Psych. 16, 370-79. O'CONNOR. J.D. 1973: Phonetics. Harmondsworth: Penguin. O'DONNELL, R., GRlFFlN, W. and NORRIS, R. 1967: Syntax of kindergarten and elementary schoolchildren: a trans/onnational analysis. NCTE 8. 'Ill PENN. C. and BEHRMANN, M, 1986: Towards a classification scheme for aphasic syntax. BlOC 21,21-38. PERERA, K. 1984: Children's wn'ting and reading. Oxford: BlackwelL PIAGET, J ]952: The origins of intelligence in children. New York: Norton. - 1970: Piaget's theory. In P.H. Mussen (ed.), Cannichael's manual of child psychology 1, New York: Wiley. PRESSNELL, L.M. 1973: Hearing-impaired children's comprehension and production of syntax in oral language. JSHRI6, 12-21. QUIRK, R., GREENBAUM, S., LEECH, G. and SVARTVlK, 1. 1972: A grammar of contemporary English. London: Longman. - 1985: A comprehensive grammar of the English language. London: Longman. QUIRK, R. and GREENBAUM. S. 1973: A university grammar of English. London: Longman. QUIRK, R. and SVARTVIK, J. 1966: Investigating linguistic acceptability. The Hague: Mouton. RADFORD, A 1989: Syntactic theory and the acqUisition of syntax (working title). Oxford: Blackwell. REES, N.S. ]971: Bases of decision in language training. JSHD 36, 283-304. - 1973: Auditory processing factors in language disorders: the view from Procrustes' bed. JSHD 38,304-15. REID, 1. and LOW,1. 1973: Link-up. Edinburgh: Holmes McDougall. RENFREW, C. (n.d.): The bus story. Published by the author. ROBINS, R. 1971: General linguistics: an introductory survey, 2nd edn. London: Longman. ROOD, Ll. and BRAINE, M.o.S. 1970: Children's imitations of syntactic constructions as a measure of linguistic competence. JVLVB 10, 430-43. RONDAL, lA, GHIOTTO, M., BREDART, S. and BACHELET, J-F. 1981: Age-relation, reliability and grammatical validity of measures of utterance length. JCh.Lang 14, 433-46. ROSENBAUM, P.S. 1967: Specification and utilization of a transformational grammar. In Project Report 4641, IBM Watson Research Center. ROSENBERG, S. 1973: Problems of language development in children: a discussion of Olson's review. In Uc. Haywood (ed.), Sociocultural aspects of mental retardation. New York: Appleton Century Crofts. ROSENTHAL, W.S .. EISENSON. 1. and LUCKAU. J.M. 1972: A statistical test of the validity of diagnostic categories used in childhood language disorders: implications for assessment procedures. PRCLD 4, 121-41. SCHNEIDERMAN. N. 1955: A study of the relationship between articulatory disability and language ability. JSHD 20, 359-64. SCHUELt, H. 1965: The Minnesota test for differential diagnosis of aphasia: administrative manual and card materials. Minneapolis: University of Minnesota Press. SCOTT. C.M. and TAYLOR. AE. 1978: A comparison of home and clinic gathered language samples. JSHD 43, 482-95. SEFER. J.W. and SHAW, R. 1972: The use of psycholinguistic principles in the treatment of aphasia. BlOC 7, 87-9. SHARF, OJ. 1972: Some relationships between measures of early language development. JSHD 37. 64-74. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. REFERENCES 237 SHERMAN, D .• SHRINER. T.H. and SILVERMAN, F. 1965: Psychological scaling of language development of children. Proc. Iowa Acad. Sc. 72,366-71. SHIPLEY. E.F .. SMITH, C.S. and GLEITMAN, LR I %9: A study in the acquisition oflanguage. Lg. 45. 322-42. SHRINER. T.H. 1967: A comparison of selected measures with psychological scale values of language development. JSHR 10, 828-35. - 1%9: A review of mean length of response as a measure of expressive language development in children. JSHD 34. 61-8. SHRINER, T.H. and MINER. L. 1%8: Morphological structures in the language of disadvantaged and advantaged children. JSHR 11,605-10. SHRINER, T.H. and SHERMAN, D. 1967: An equation for assessing language development. JSHR 10,41-8. SHULTZ, T.R and PILON, R 1973: Development of the ability to detect linguistic ambiguity. Ch. Dev. 44, 728-33. SINCLAIR-DE-ZWART. H. 1969: Developmental psychological linguistics. In D. Elkind and J. Flavell (eds.), Studies in cognitive development. London: Oxford University Press. SLOBIN, D. (ed.) 1985: The cross-linguistic study of language acquisition. Hillsdale, NJ: Erlbaum. SLOBIN, 0.1. and WELSH, CA 1968: Elicited imitation as a research tool in developmental psycholinguistics. Working Paper 10, Language Behavior Research Laboratory, Berkeley. Reprinted in CA Ferguson and D.1. Slobin (eds.) 1973, Studies in child language development. New York: Harper and Row, 485-97. SMITH, M. 1933: The influence of age, sex and situation on the frequency, form, and function of questions asked by preschool children. Ch. Dev. 4, 201-13. SPREEN, O. and WACHAL. RS. 1973: Psycholinguistic analysis of aphasic language: theoretical formulations and procedures. Lg. and Sp. 16, 13Q-46. STACK, E.M. 1971: The language laboratory and modem language teaching. London: Oxford University Press. STARK, J .. POPPEN, R and MAY, M.Z. 1%7: Effects of alterations of prosodic features on the sequencing performance of aphasic children. JSHR 10, 844-48. STOKOE, W. 1972: Semiotics and human sign language. The Hague: Mouton. STRAUSS, A.A and McCARUS, E.N. 1953: A linguist looks at aphasia in children. JSHD 23, 54-8. SVARTVlK, J (ed.) 1973: Errata-papers in error analysis. Lund: Gleerup. SVARTVlK, J., EEG-OLOFSSON, M., FORSHEDEN 0 .• ORESTROM, B., and THAVENIUS. C. 1982: Survey of Spoken English: report on research 1975-81. Lund: Gleerup. TAYLOR. O.L and ANDERSON. c.B. 1968: Neuropsycholinguistics and language retraining. In J.W. Black and E.G. Jancosek (eds.), Proc. Can! on Language Retrainingfor Aphasics. Ohio State University, 3-18. TEMPLIN, M. 1957: Certain language skills in children. Minneapolis: University of Minnesota Press. THOMAS. Fl. 1971: Guidelines, English language course, Stage 1. Woodford Green, Essex: Woodford Educational Publications. TRUDGILL, P. 1974: Sociolinguistics. Harmondsworth, Middx.: Penguin. TYACK, D. 1972a: The use of language samples in clinical settings. PRCLD 4, 163-8. - I 972b: Some notes on Lee and Canter's 'Developmental sentence scoring: a clinical procedure for estimating syntactic development in children's spontanenous speech'. PRCLD 4, 179-84. VERHULST-SCHLICHTING. L 1982: Handleiding bi} de Voorlopige Nederlandse Versie van de LARSP Profilkaart. Institut voor Stem-en-Spraakstroomissen, Utrecht University. VERHULST-SCHLICHTING, L. and DE KONIG, T. 1988: FIT: Functionele Imitatie van Taalstructuren. Amsterdam: Swets & Zietlinger. WACHAL. R.S. and SPREEN, O. 1970: Grammatical analysis of aphasic language. Manual of instructions for classification and coding. University of Victoria. Mimeo. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 238 TIlE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY - 1973: Some measures oflexical diversity in aphasic and normal language performance, Lg. and Sp. 16, 169-81. WEBSTER. B. and INGRAM, D. 1972: The comprehension and production of the anaphoric pronouns 'he, she, him, her' in normal and linguistically deviant children. PRCLD 4, 55-77. WEIGL, E. and BIERWISCH, M. 1970: Neuropsychology and linguistics: topics of common research. FL 6, 1-18. WELLS, G. 1985: Language development in the pre-school years. Cambridge: CUP. WEPMAN. 1M. and HASS. WA 1967: Lexical and syntactic indices of children's language development. In Actes du Xeme Congres des Linguistes. Bucharest: Editions de l'Academie de la Republique Socialiste de Roumanie. WEPMAN,lM. and JONES, L.V. 1964: Five aphasias: a commentary on aphasia as a regressive linguistic phenomenon. In D.M. Rioch and EA Weinstein (eds.), Disorders of communication. Baltimore, Maryland. WEST, JJ. and WEBER. lL. 1974: A linguistic analysis of the morphemic and syntactic structures of a hard-of-hearing child. Lg. and Sp. 17,68-79. WILLIAMS, F. and NAREMORE, R.c. 1969: Social class differences in children's syntactic performance: a quantitative analysis of field study data. JSHD 34, 779-93. WILLIAMS, J.E. 1981: The Cambridgeshire Language Development Project. In A, Jackson (ed.), Ways and means, Vol. 3, Basingstoke: Global Education. - 1986: A use ofLARSP in secondary units for hearing-impaired pupils. Child Language Teaching and Therapy 2, 154-69. WILSON, M.E. 1969: A standardized method for obtaining a spoken language sample. JSHR 12,95-102. WREN. c.T. 1981: Identifying patterns of syntactic disorder in six-year-old children. BJOC 16, 101-9. - 1982: Identifying patterns of processing disorder in six-year-old children with syntax problems. BJOC 17, 83-92. - (ed.) 1983: Language learning disabilities: diagnosis and remediation. Rockville, Md.: Aspen. YNGVE, V. 1960: A model and an hypothesis for language structure. Proc. A. Phil. Soc. 108, 275-81. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Index (names) Alfonso, A 209 Altmann, SA III Amidon, A 81 Anthony, A 129 Atwell, E. 205 Bachelet, 1.F. 200 Ball, M. 207 Bamford, 1. 199 Baumann, P.M. 13, 16,26,33,36,37,44,96 Behrmann, M. 200 Bellugi, U. 69,71, 75 Bench, 1. 199 Berman, R. 207,208 Bevington. 1. 7,201 BielWisch, M. 32 Bishop, D.V.M. 206 Blasdell, R. 58 Blockley,l. 110, 112 Bloom, L.M. 17, 63, 67 Bol, G. 207 Braine, M.D.S. 86 Brannon, 1.B. 18, 87 Braun, C. 16, 33 Bredart, S. 200 Brown, H.D. 77 Brown, R 9, 11,26, 30-31, 34, 59, 61, 65, 69, 73,87, 115, 125 Buschini, G. 127 Byers Brown, B. 200 Cambon, 1. 82 Canter, S.M. 12-13,16,17-18,26,28,36,86, 87-9, 124 Carey, P. 81 Cazden, C. 73 Chapman, RS. 200 Child,1. 161 Chomsky, C. 82 Chomsky, N. 5,6, 16,34,35-6,37,38,60, III Clahsen, H. 207 Clark, R 24, 28 Code, C. 200 Conn, P. 12,26,27 Connolly. 1. 200 Cooper. RL. 12 Cowan,PA 9 Craig, E. 116 Cromer, R 86 Cruttenden, A 81 Crystal, D. 2, 7,9, 11, 12,43, 56, 57, 66, 98, 116,177,197,198,199,200,201,212 Curme, G,O, 38 Dale, P .S. 59 Darley, F. 87 Davis, EA 8 Davis, GA 33, 36 Davy, D. 12, 57, 98 Dennis, D. 201 DelWing, B.L. 60 Dever, RB. 13, 16, 17,23, 26-7, 29, 33, 36, 37,43,44,96, 115 Donaldson, M. 27 Dore,1. 63 Easterbrook, A 200 Eeg-Olofsson, M. 198 Eisenson. 1. 110 Ellsworth, RB. 8 Elson, B. 27, 44 Engler, L.F. 16, 18, 35, 37, 86, 87, 91 Feilberg, 1. 207 Ferguson, CA 59 Fitzgerald, M.D. 200 Fletcher, P. 201, 202, 204, 205, 212 Fodor, lA 26 Forsheden, 0. 198 Foster. c.R. 37, 64 Foster,1. 201 Fraser, G.M. 110, 112 Fygetakis, LJ. 124 Garman, M. 201,205 Garrett, M. 26 Garside, R 205 Ghiotto, M. 200 Gillies, M. 201 Gimson, AC. 6, 58 Gleason, 1.B. 124 Goodglass, H. 21, 31-2, 33, 58, 114, 116 Goodman. M.F. 8 Gottsleben, RH. 127 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 240 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILIlY Greenbaum, S. 38, 198,207 Greenfield. P. 63 Gregersen, K 207 Griflith. l 87 Haden, R 200 Hahn, E. 78 Halliday, MAK 37 Hanlon, C 125 Hart, B. 83 Hass, WA 33, 36 Hatfield. F.M. 14 Hawkins, P. 200 Hayden, C 207 Hayes, lR. 26 Heider, FR 12 Heider, G.M. 12 Herriot. P. 79 Hickey, M. 207,211 Hirsch, M. 207 Hishikori, Y. 207 Hixson, P. 202 Hockett. CF. 110-11 Hunt, K 81 Hutchison, M.K.F. 2. 12 Hutt, E. 201 Huttenlocher, l 120 Huxley, R 78 Ingram, D. II, 16, 19,23, 26, 29, 33, 34, 36. 37, 68,73,74,78,79,87,88,89,95, J 10,114,115, 124 Jensen, P. 58 Jespersen, O. 38, 63 Johnson. M. 37,68. 110 Johnson. M. 201,205 Jones, L.v. 8 Joos, M. 10 Jordan, CM. 86 Keeney, TJ. 79 Killingley, SoY. 200 Kirk, SA 15 Kirk, W.D. 15 Klasen, B. 16, 33 Klee, T. 200, 206 Klima, E.S. 71,75 Kruisinga, E. 38 Kuiken, F 207 Kunze, L.H. 77 Lackner, l 29 Lee. L. 5. 12-13, 16. 17-18, 23. 26. 28. 33. 36, 86, 87-9,124,197,199 Leech. G.N. 198,205,207 Lenneberg, E.H. 29. 33 Leonard, L.B. 29,33 Lerea, L. 15 Limber, l 69, 77, 79, 87 Long, S. 206 Longhurst, T.M. 5, II, 16, 17,34,37,87,91 Looney, P.L. 26, 33, 78, 86 Low,l 7 Lozar, B. 34 Lyons, J. 39,76.111,184,194 Mackay, D.D. 7 Mackey, W.F. 118 Madhani, N. 207 Matthews, P.R 10 Maxim, J. 200 McCarthy, D. 8, 9. 87 McCarthy, l 15 McCawley, J.D. 6 McGrath. Co. 77 McNeill, D. 62,63. 67·8, 78 Menyuk, P. 5, 16, 26, 29. 33, 59, 78. 86, 87 Miller, IF. 29, 33, 86, 200 Miner, L.E. 87, 115 Minifie, F 9, 87 Mittler, P. 24 Moll, K 87 Morehead, D.M. II, 16, 19.26,29,33,34,36, 37.68, 78, 79, 87, 88, 89,95, 109, 110, 114, 115 Milller, D. 200 Muma, lR 12, 33, 34, 36, 89 Munro, S. 200 Murray, T. 18,87 Myerson, R. 21. 31-2,33,114,116 Myklehurst, H. 5 Naremore, RC 37 Nelson, KE. 125 Newfield, MY. 115 Nice, M.M. 9. 87 O'Connor. J.D. 6 Orestrom, B. 198 Paul, R 200 Penn, C 200 Perera, K 200,201 Peters, J. 202. 204 Piaget, l 60, 63, 65, 66. 109, 111 Pickett, V. 27,44 Pike, KL. 37 Pilon. R 82 Pleh, C 207 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Poutsma, H. 38 Pressnell, L.M. 12,29 Quirk, R 12,29,38, 198.207 Radford, A 198 Rees, N. II, 25-8, 110 Reid.1. 7 Renfrew. C. 11,24 Reynell, J. IS. 18, 37 Robins, RH. 9 Robinson, W.P. 86 Rodd, L.J. 86 Rom.A 207 Rondal. J. 200, 207 Rosenbaum, P.S. 16,34 Rosenberg, S. 29, 33 Rosenthal, W.S. 22 Sahlie, E. 206 Sapir, E. III Schaub, I. 7 Schelletter, C. 201 Schlanger, B.B. 115 Schneiderman, N. 87 Schrandt, TAM. 11. 16, 17, 34, 37, 87.91 Schuell, H. 161 Scott, C.M. 202 Serer, JW. 32. 125 Scharf, DJ. II, 13, 18 Shaw, R. 32, 124 Sherman, D. 87 Shipley, E.F. 64, 124 Shriner, T.H. 9, 11, 87, 91, 115 Schultz, T.R 82 Sinclair (-de Zwart), H. 60, 82, III Skinner, B.F. 60 Slob in. D.I. 59,86, 119,207 Spencer, H. 200 Spreen, O. 33. 116 Stack, E.M. 118 Stark, 1. 58 Stodel, L. 201 Stokes, 1. 207 Stokoe. W. 116 Svanvik,1. 12, 29, 38. 78, 198, 207 Taylor, AE. 202 Templin, M. 11.87 Thavenius. C. 198 Thompson, B. 7 Thompson, I. 200 Trudgill, P. 73 Tyack, D. 17, 26, 87, 127 Verhulst-Schlichting, L. 207 Wachal, R.S. 33, 116 Wales, R.1. 27 Webster, B. 78 Weigl, E. 32 Wells, G. 202 Welsh, CA 86, 119 Wepman, lM. 8, 33, 36 Williams, F. 37 Williams, J.E. 200, 201 Wolfe.1. 79 Wren. c.T. 199,212 Yngve, V. 26 Yoder, D. 29, 33 INDEX 241 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Index (subjects) accent 82. 90. 201 acceptability 12-13,29.38 accidence 5 adjectival 41,53,69-70,72-5, 102-5,135-8. 143, 145,147,153.156,166-7.173.191,193,221·2 adjective 40-41,50·54.79,136,144,153-6,174, 191-4 adult language 23, 26·8, 31·2, 62, 99-108, 116, 124, 161·96 adverb 9,40-41,47,51,53·5,79, 148, 174 adverbial 32.41,46·56,69·75.79,81. 102·3, 143, 145·8, 153, 173-4, 180-89, 194-5, 205, 206,221·2 clause 42,47. 51 phrase 47,51. 71 suffix 73 affix 10, 209-11 age 30-31,61.66,70,73.75,77,84,90,113,115 agent 45 agrammatism 14. 22, 58 allomorph 115 ambiguity 69, 82,94.99, 113, 126, 163, 166 American Sign Language 116 and use of 12,76,82,101, 167.173·4, 187, 190· 91 aphasia 8, 83, 109, 114. 200 Broca 116 fluent 117, 120 syntatic continuum in 200 aphorism 49 apposition 26,41·2,48,50,80 arrangement 18 article 40, 50. 67, 72, 79. 177·8 articulation 66,89,91,115,121,126,163,166. 169·70, 178.213.215·16 artificiality 12 aspect 40,54,79, 138·9, 154.177 Aspects of the theory of syntax 5 assessment 23,31, 86-108, 113·17, 129·35, 140·45, 149·53, 156-60, 162·77, 197 attitude 56 autism 116 Automated LARSP 206 autonomous 64 auxiliary verb 40,46,51.54,72·5,79, 103·4, 116. 153·6, 167, 178. 180.202.221·3 contracted 55, 73. 167,222·3 baby talk 124-5 balance 113·15. 123 behaviourist 60 Bengali 207 blends 70-71. 74 Breakthrough to literacy 7 Bristol Survey 198 Cambridgeshire Language Development Project 201 case 40, 51, 55 case studies 199 child development 8, 18. 26, 28-32. 59·84, 128·60,167,180,196,198 clause 44 dependent 47-8, 50, 52, 53-4 sequence 75·7, 194 structure 17,27,39,43-7,64,67-75,78-80.94. 98,102·3. Ill. 114, 121-4,128,140,143·50, 166-83. 187. 190-91, 195.204.208·9 type 36,41, 81, 181·2 cognition 60,65,68,79-80, 114-15, 199 v. language 22.27. 109·111 College of Speech Therapists 6 collocation 68, 110, 138 colloquial 12·14,27, 111 command 41,45,56,63,65·6.69·74,95, 147, 163,165·6,169,173-4,180-82.184-5,189-90 comment clause 42, 49, 81 common core 38·9 communicative history 90,218·20 comparative clause 42, 49, 51, 77 comparison 41·2,55.73.81, 114-15, 167 competence 32, 34-6. 86 complement 41·2.45·56,68-77,82,100.102·3, 114,144, 153·6,169·70, 191, 222-3 complementation 45, 80 completeness 88·9. 92, 94 complexity 10-11,25·7,29·30, 43, 68, 154-5, 169, 171·2, 196 comprehension 24,79.81'2,96, Ill, 121. 126, 130. 162·3, 182·90. 192, 199 problem 23, 126, 134, 171 comprehensiveness 15,20·21, 23-5.30,39 computational studies 198, 201, 205·6 computerized profiling 206 concord 41. 79, 110, 114, 192 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. conjunction 50,76.82. 187, 191, 194-5,221·3 connectivity 41-3, 75-7, 101·2, 173-4. 189-90, 194-5 sentence 42·3,49-50,81,97, 115 contracted forms 10, 73, 129, 153. 167, 173, 179 See also auxiliary, copula, negation con traction drill 119 conversational turn 98, 104 coordination 12,41-2,47,50,74-7,80,98, 101, 128, 148. 153, 156 copula 45,55,72-3, 103, 153-4, 156,207·8 contracted 45.55,73, 115, 167,222 countability 52, 79, 173, 179-80 creativity 76" criteria 8-9, 16, 39, 65 critical elements index 64 critical period 83 cross-linguistic uses 206-12 cross-reference 50, 67, 81 Danish 207 Databank 201 deafness 7, 12, 116. 125, 199 deep structure 35, 82, III delay 113 v. deviance 28 demonstrative 52 descriptive v. theoretical 60, 198 determiner 40,52-3,69-70, 72, 75, 79, 103-5, 140,145-6.153,167,171-2,178-80,183,221-2 Developmental Language Scale (Reynell) 15, 18.129 Developmental Sentence Scoring (DSS) 17, 18, 199 deviance 28-9,94,96,99,110,115-16,125,129, 148 diagnosis 22, lOS, 197, 204-5, 212 dialect 27-8,40, 117, 125 direct speech 42 directional 185·9 disadvantage 113 discourse 42, 81 . distribution 46, 53,67, 121, 183 do 56, 145, 147-8 drills 118-20, 127, 136 duality lIo-ll Dutch 207 dynamic 185·6 dysphasia 21, 24, 25, 31, 57·8, 83, 109, 120, 161-96 echolalia II, 88-9, 96, 120-21, 134 echo-utterance 41 eclecticism 37-9, 60 INDEX 243 Edinburgh Articulation Test 129, 151 elderly speakers 200 elective mutism 116 element 63-4 ellipsis 13, 41-2, 50,63-4, 81, 95-7, 100-101, 111-12,121-2,125,127,163,165-7,182 embedding 75,77,119-20 emphatic order 81 error 13, 72, 78, 81, 98, 110, 117, 125 -analysis 78 natural 125 ESN (Educationally Sub-normal) 24,] 15 exclamation 40-4], 56, 66, 77, 95, ]69 expansion drill ] 19 explicitness ] 5-22, 161 expressive language 23, 109, 199 extralinguistic 9, 19,22,31,89-90,92, 110, 115, 214-20 finite 40,42,48, 53, 181 flexibility 25, 84, lOS, 121, 123, 139 follow-up studies 212 forced alternative (FA) question 96, 120-23, 125,135-40,145-7,172-4,179-80,184, ]91-2 foreign-language teaching 23, 78, 83, 118-20 formulaic utterance 41 French 207 frequency 27-9,69,74, S9, 98-105,113,115-17, 130, 140-41, 149, 153, 163-7,201-6 fricative 129-30, ]44, 156 gap-filler 165-6 gender 40, 52 generic 62 genitive 51, 55, 73 German 207 gesture 64-5 grading 21, 23-4, 25-32, 127 grammar definition of 5, 15,40 reference 3S-9,206-7 traditional S-9, 13, 39 Grammar of Contemporary English, A 3S-42, 52-6. So-SI, lSI, 19S, 207 graphetics 7 graphology 7 Guidelines lIS head 42, 51, 53 Hebrew 207, 208-9, 211 hemispherectomy 83 hesitation 9],94,97, 121, 163, ]79-80, 192 heterogeneity 112 hierarchy 33, 67, 114, 123 holophrase 63 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 244 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY Hugh 128-60,212 Hungarian 207 Illinois Test of Psycholinguistic Abilities 15 imitation 58,60,86,96, 120, 135, 154, 156 elicited 95-6, 119 modelled 119, 136-40, 145-8 immediate constituent analysis 37 imperative 40, 72, 88, 138-9, 145, 147-8, 222 incipient structure 116-17,126-7,180-83,187, 193 incomplete 94-5, 99, 103-4, 117, 163, 166 see also completeness incremental drill 119 indefinites 52 Indiana Scale of Clausal Develo?ment 27, 44 infinitive 48 inflections 36, 72-3, 114-15, 121, 126 initial effect 172 initiator 41, 52-3, 80 innateness 35, 60, 63 intelligibility 88-9,91-2,94, 126, 129 see also unintelligible intensifier 41, 53, 69-70, 143, 166, 222 interaction 11,56,86-7,120,124,174 interjection 49 interrogative pronoun 52 intonation 14,23,35,36,38,39,42-3,49,56-8, 63-9,81,91,93,97, 100, 102, 119, 121, 126, 147, 163-4, 166-7, 170-71, 177, 179-81, 187-8, 190-91, 194 of questions 13, 55, 155 intuition 35, 88-9 inversion 55, 71, 74, 77, 79, 110, 120, 124, 155 Irish 207,211 it 42,49, 81 Japanese 207.209-10 kernel 36, III labels (clinical) 199 language laboratory 118 Language Through Reading 201 LARSP (Language Assessment, Remediation and Screening Procedure) aims of 20-32 clinical use of 24-5, 113-95, 197-9 cross-linguistic use of 206-12 diversity in use of 198-9,204 evaluations of 200 methodology of 86-\05,200,201-5,213-20 recent uses of 197-212 syntactic framework of 35-56 LASS (Linguistic Analysis of Speech Samples) 37 LCI (length-complexity index) II. 18 length 9-11,12,27,60,68,104,122,154,156,169 levels (of language) 5-6, 126 levels (of structure) 39,43,67,94,98,112-17 lexical 35, 178-80 item 67, 121,128,133, 135, 138, 160, 178-9, 183-9, 192-3,211-2 linguistics 15-16, 109-12 Link-up 7 lip-reading 125 listening 82-3 longitudinal studies 212 look-and-say 6-7 major sentence 44-9,64,95,97.100-101,104, 163, 165-7 matched sample 19 Mean Length of Response (MLR) 9-10. 18 Mean Length of Utterance (MLU) 9-10,200 medical history 90, 128-9, 162-3 memory 120, 162, 172 mental subnormality 7, 124-5 Michigan Picture Language Inventory 15 microfiles 198,205,206 Minnesota Test for Differential Diagnosis of Aphasia 162, 169 minor sentences 41,44,49,56,66,95-7,100-101, 104, 133-4, 143, 160, 163, 165-6,221-3 mood 54, 79, 116, 156,202 morph 68 morpheme 9, 10-11,43 morphology 5.15,75,115-16,126,191,193, 209-1 I motivation 30, 39. 113. 117 Mr J 99, 161-96 multiple sentence 44,47-9 naming 121-2, 124 negation 40-41,51. 54,69-71, 74-5,77.79,102, 110,120,145,147,149,153-6,165-6,173-4. 191-4 contracted 54-5, 73, 222 neurology 110 New English Dictionary 38 non-communicating 24, 116, 124-5 nonfinite 40,42,48, 54, 77, 182 normal nonfluency 12 norms 11-14,22-3,28,30,37-8,59,84,96,113, 196 Northwestern Syntax Screening Test 13,149 Norwegian 207,209 noun 41, 53, 64-5, 68-70, 72-5, 83, 88, 102-5, The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 122, 134-8, 143, 145-8, 153, 156, 166-7, 171-3,178-80,183,201-5,221-2 classification 40, 51-2, 173 clause 42, 47, 76-7 irregular 79, 81 noun phrase 35,40,42,47, 50-54,69,72-5,78- 80, 114, 119, 140, 144-8, 153-6, 181, 183, 191, 203 nucleus 57 number 40,51,54-5,114-15,178,222 Number of Different Words (NDW) 18 object 34.41,45-56,68-77, 102, 114, 116-17, 122-3, 138, 140, 143-8. 153-5, 167,169-84, 191,221-3 direct/indirect 45,51.71, 72, 74. 145-8 order of acquisition 30-31, 59, 204-5 ordinal 40-41, 53 Other 65,69,71,74,76.79-80,97,98,100.102. 105, 116, 134, 140, 143, 160, 165 overlap 92, 100 Paget-Gorman 116 Panjabi 207 participle 41,44,48.54-5,64.73, 103-4, 115, 193,221-2 particle 54,69-70, 135, 143, 148,210.221-2 part of speech 7-9, 21. 40, 113 passive 40, 42, 54, 80. 120 patholOgy 83 pause 43, 57, 76,91,95, 100. 139. 144, 166, 177, 179, 188 perception 27.110-12 performance 27, 32, 34-6 person 54-5,73, 103, 115, 153 personal pronoun 44, 52, 102 phonation 162 phonemics 64, 115, 130 phonetics 5, 6-7 phonics 6-7 phonology 5,6-7,38,126,128-30,133,140.170, 178,180,185,212 phrase strUCTure 17,21, 34,43,67.69-75,98, 102-4, 113-17, 123, 128, 134-5, 138-40, 143-50, 166-83, 187, 190-91, 194-5,209-11 pictures. use of 87, 145-6, 148-9, 154-5. 163. 171,173-4,192,202 plural 51,55,73,114-15,129,144.148,153,180, 221-2 possessive 53, 55.65.73. 115. 129. 137 pronoun 52 postmodification 42,51.52-4,77 postpositions 209-10 pragmatics 212 predeterminer 40 INDEX 245 predicate 44, 88 premodification 41-2, 51, 53-4 preposition 40-41, 53-4,69-70, 72-5, 138-40, 143, 145-8, 153,173. 183-9,209-10,221-2 complex 41, 53, 189 prepositional phrase 4Q-1, 51,53-4,77,92,181, 183-9, 195, 203 Problems 65-6,97-\00, 102 production (of language) 24. 79, 86,96, 110, 126. 130, 182, 184 productivity 89, 128, 135. 147, 160, 189 profile 19,30,89, 113-17, 127,212 chart 61-2,72.78,80,83-5,98-108,123.129. 131,133-8, 140-5. 163-8 progressive 138-9, 154 prominence 42. 57-8 prompt 95, 119 pronoun 13,40,44,47,50-53,67,72-3,78-9,92. 102-3, 114. 117, 119, 121, 128, 134-5. 148, 153,179,221·3 types of 52-3 proper names 9, 92 prosodic features 7,35, 56-8.67,91, 101. 104. 125-6, 165 protoform 62 psychiatry 110. 116 psycholinguistics 26.31,62, 111, 119 psychology 110-12,116 puberty 81,83 pun 82 punctuation 42-3, 57 qualitative analysis 29 quantifier 40-41, 53 quantitative issues 201-5 questionnaire 90,201-7 question 40-41.46,55-6,63,65-6,69-74.95-7, 100, 110, 115. 123. 130, J 35-8, 140. 145-7. 154-6.162-3,165-6,169.171,173-4,181-4, 189-90,202 drills 120-23 open 120, 124171-3,191 open alternative 172 tag- 56, 77, 198 wh- 26,55,65,71, 74,120,130,135-7.155-6, 190,221 see al~w forced alternative Quirk Report 25. 199 rate of acquisition 30-31 reading 6-7. 14,82, 162·3.201 recall 120 recency effect 172 recursion 75-7. lIS reflexives 52, 78 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. 246 THE GRAMMATICAL ANALYSIS OF LANGUAGE DISABILITY register 38 regression 21, 31, 191, 196 relational 185·6 relative clause 26, 42, 48, 52, 77 relative pronoun 52 remediation programme 20-21, 23, 63, 109·12, 117·27, 136-40, 145-9, 153-6, 169·% repetition 96,98,100-101,115,117,163-4,166 replacement drill 119 representativeness 17·18 request 173-4, 181·2, 184, 189 response 10,49,94-7, 100-101, 103, lI5, 124, 127,130,132-7,140,145-7,155,163,165·7, 172, 182, 184, 188, 192 abnormal %·7, 100, 163 normal 95,97, 100, 163 restrictive 53 restrictiveness 42 rhythm 42, 57-8, 91, 126, 177 riddle 82 root 10 sampling 86-90, 113, 115, 200, 202-4, 213-17 scale-and-category grammar 37 scale of development 23, 30 scan 93-\05 score 18 screening 20, 26, 86 Seeing essential English 116 segmentation 177-80,190,195,201 semantic classes 65 semantics 5-6,8,38,41,50,62-5,79-82,90,116, 126,180,185, 188-9,196,212 generative III semiotics 110 sensori-motor intelligence 65-6 sentence 83, 126 count 92-8, 104, 163 function 55-6,65-6, 75, 81 identification 43, 57, 76,201 length 9-11,86,98,126, 188 novel 97, \01 -point 18 structure 8,21,35,40-41,43-9,68-9,81,98, 113-17,120-23,182 type 35,44-9, 77, 165-6 I-element 63-6,97,102. 121-2, 134-5, 143, 145,166-7,169,172 2-element 31,67-70, 102·3, 122, 143 3-element 70·73, 102-3, 122-3, 144 4-element 73-75, 102 see also kernel, major, minor, multiple, simple sequencing 58, 76, 112, 114, 162, 169, 173-4, 177-80, 194 signing 116 simple sentence 44-7 simplicity 124 singular 114 situation 136 Skylarks 7, 201 slang 91, 213, 216 social (minor sentences) 49.66,95, 102, 166 socio-economic level 84 sociolinguistic 82, 115. 196-7 special education 197 speech therapy 6, 20-21, 24-5, 155 spontaneous speech 13,81,83,86,94-7, 101, 119, 140, 147-9, 154-6, 160, 163, 166, 173-4 Stages 18, 61-83, 94, 96-8, 104-5, 112, 116, 121·3, 133, 135-6, 140-60, 166-91, 198 Stage V syndrome 212 standard English 115, 125 standardization 19, 84, 201-5 statement 41,55,63,65·6,68·74,95, 105,123, 147, 165-7, 182 statistical methods 199·200 stative 185-6 stereotype 49, 66, 89, 101-2, 117 stress 42,57,67,69,121,133,155,177 structural linguistics 34-5, 37-42 structure count 98-105 structured situation 88, 118-27, 139 style 12-14,38,40,69,81-3 SUbject 34,41. 44-56, 68-77, 82, 88, 100, 102-3, 114,117,122,135,138-40,143-8,153-6,167, 169-84, 191,221-3 subordination 41-2,47, 76-7,81,98, 101 substitution drill 119 SUMLARSP 204, 205-6 superlative 55, 73, 114 surface structure 36, 80-81, Ill, 115 Survey of English Usage 38 syllable 68, 163, 169, 173 stressed 10, 68 tonic 57, 154 symbolic noise 94, 99 syntactic expansion 67-8,70-71,74, 125 syntactic framework 20-21, 23, 33-56, 60·62 Syntactic structures 5 syntax definition of 5 quotient 5 system 50, 114 completion 77-80 systema ticness 14-15 systemic grammar 37 tagemics 16, 36-7,44 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. Teaching the American Language to Kids (TALK) 23 telegraphic 67, 114, 125 tense 40, 44, 54-5, 79 past 54-5, 73, 114-5, 202-3, 221-2 present 14,55, 114, 154, 178 sequence> 81 terminology 39, 109, 199 tests 13, 15, 86, 90, 129, 214-5 there 42,49,81,222 time factor 11,24-5, 127 tone-unit 57, 144 topic markers 209-10 transcription 89-93, 99, 201 phonetic 58, 64 prosodic 56-7 transformation 12,16,18·19,26,35-6,40-41, 60, 75-6, 78, Ill, 182 drills 120 transformational-generative (TG) 5,16,25-6, 33-8, 76, III T-unit 89 twins 200 unanalysed utterance 94, 98-9 unintelligible 94,99, 133, 140, 149, 163-4 universals 206 University Grammar of English, A (UGE) 38, 52-6 usage 38 realistic 11-14 utterances 10,86-7,91, 201 variety 40, lIS, 125 verb 34,44,64-5,68-75,82·3,88, 100, 102-4, 114,116-7,120-22,124,130,134-40,143-50, 153-6,167,169-84,191,193,205,221-3 complex 54 intransitive 45, 138-140. 193 irregular 40, 55, 79, 81 lexical 40, 55, 177 main 205 transitive 122, 140, 193 type 40, 42, 205 unmarked forms of 205 see also auxiliary, copula, mood, person, tense verb phrase 33,35,37,40,42,51,54-5,69-75, 78-80, 103, 112, 116 complex 80 verbless 42, 48, 54, 182 vocabulary 6·7, 50,68,81-3, 124, 126-7, 136, 138, 155, 178-9 vocative 41,46,49,69,221 voice 42,54 Welsh 207 West Indian English 115 word 5,63,68, 113-14,209-11 -class 7-9,64, 178·9 count 10,98, 104 'empty' 50 INDEX 247 -finding problem 117, 162-3, 170, 178·81, 183, 190, 192-3 first 66 function IS, 36, 58, 72, 114, 145, 178 identification 64 order 42,46,62,64,67, 78-81. 110-12, 120, 144, ]48 structure 42-3, 55,64,98, 103, 114-15, 144, 147, 167, 183,209·11 written language 82-3, 121, 163, 174, 181, 192 v. speech 12,38,43,49, 116 zero 96-7, 100, 130, 140, 149, 160, 165 article 114 The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand. The University of Canterbury reproduces this publication with the consent of the author David Crystal. This publication is currently out of print and all rights and ownership are retained by the author. Publication and further communication must comply with the Copyright Act of New Zealand.