8
Clinical Forum 184 so. Rather clinicians now have a much more sophisticated way of thinking about the needs of these children and look for assessments which allow them to target speci c skills such as the Phonological Assessment Battery (Frederickson et al. 1997). The UK nature of the materials and the standardization should be real selling points although I suspect that the age range and the use of the subtests oVered by the CELF (Semel et al. 1987, Wiig et al. 1992) in its pre-school and school forms may oVer some stiV competition for those looking to monitor change in their clients across the school range. None the less this revision of the Reynell Developmental Language Scales remains another milestone and will, I am sure, become a part of the repertoire of clinicians and researchers in this area. Of particular interest is the potential of the scales as an outcome measure for educational or therapeutic intervention. The early versions frequently appeared as such an outcome but there was some concern that, like many standardized procedures, they proved to be relatively static and insensitive to change. It will be interesting to see if the increased diV erentiation of the scales does indeed provide a more sensitive indicator of change. References Bishop, D. V. M. and Edmundson, A., 1987, Language impaired 4 year olds: distinguishing transient from persistent impairment. Journal of Speech and Hearing Disorders, 53, 156–173. Dockrell, J., George, R., Lindsay, G. and Roux, J., 1997, Problems in the identi cation and assessment of children with speci c speech and language impairment. Educational Psychology in Practice, 13, 29–38. Edwards, S. and Letts, C., 1997, Reynell Scales [Letter]. Bulletin of the Royal College of Speech and Language Therapists, 548, 18. Frederickson, N., Frith, U. and Reason, R., 1997, The Phonological Assessment Battery (Windsor: NFER Nelson). Knowles, W. and Masidlover, M., 1987, The Derbyshire Language Scheme (Derbyshire: Education Testing Service). Renfrew, C. E., 1997, Bus Story Test (4th edn) (Bicester: Winslow Press). Semel, E., Wiig, E. H. and Secord, W., 1987, Clinical Evaluation of Language Fundamentals—Revised (Sidcup: Psychological Corporation). Wiig, E. H., Secord, W. and Semel, E., 1992, Pre-school CELF (Sidcup: Psychological Corporation). Address correspondence to: James Law, Department of Clinical Communication Studies, City University, Northampton Square, London EC1V 0HB, UK. Reply Moving on Susan Edwards, Michael Garman, Arthur Hughes, Carolyn Letts and Indra Sinka The University of Reading, Reading, UK Introduction We are pleased that the commentators have been so positive about the new Reynell Developmental Scales, and that the points they make allow us to address a number

Reply Moving on

Embed Size (px)

Citation preview

Page 1: Reply Moving on

Clinical Forum184

so Rather clinicians now have a much more sophisticated way of thinking aboutthe needs of these children and look for assessments which allow them to targetspeci c skills such as the Phonological Assessment Battery (Frederickson et al 1997)The UK nature of the materials and the standardization should be real selling pointsalthough I suspect that the age range and the use of the subtests oVered by theCELF (Semel et al 1987 Wiig et al 1992) in its pre-school and school forms mayoVer some stiV competition for those looking to monitor change in their clientsacross the school range None the less this revision of the Reynell DevelopmentalLanguage Scales remains another milestone and will I am sure become a part ofthe repertoire of clinicians and researchers in this area Of particular interest is thepotential of the scales as an outcome measure for educational or therapeuticintervention The early versions frequently appeared as such an outcome but therewas some concern that like many standardized procedures they proved to berelatively static and insensitive to change It will be interesting to see if the increaseddiVerentiation of the scales does indeed provide a more sensitive indicator of change

References

Bishop D V M and Edmundson A 1987 Language impaired 4 year olds distinguishing transientfrom persistent impairment Journal of Speech and Hearing Disorders 53 156ndash173

Dockrell J George R Lindsay G and Roux J 1997 Problems in the identication andassessment of children with speci c speech and language impairment Educational Psychology inPractice 13 29ndash38

Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech andLanguage Therapists 548 18

Frederickson N Frith U and Reason R 1997 The Phonological Assessment Battery (WindsorNFER Nelson)

Knowles W and Masidlover M 1987 The Derbyshire Language Scheme (Derbyshire EducationTesting Service)

Renfrew C E 1997 Bus Story Test (4th edn) (Bicester Winslow Press)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(Sidcup Psychological Corporation)Wiig E H Secord W and Semel E 1992 Pre-school CELF (Sidcup Psychological Corporation)

Address correspondence to James Law Department of Clinical Communication Studies CityUniversity Northampton Square London EC1V 0HB UK

Reply

Moving on

Susan Edwards Michael Garman Arthur Hughes Carolyn Lettsand Indra SinkaThe University of Reading Reading UK

Introduction

We are pleased that the commentators have been so positive about the new ReynellDevelopmental Scales and that the points they make allow us to address a number

Clinical Forum 185

of interesting issues All three commentators raise queries about our methodologywhich are important in the current climate of lsquoevidence-based practicersquo Moregenerally we see Leesrsquo commentary as focusing on clinical issues Lawrsquos on method-ological issues and Ballrsquos on the historical background We will address each ofthese areas in turn We will start by responding to points made by Lees concerningthe target population of our test Our target population is all children whose rstlanguage is English but our commentators also raise the issue of whether the testcan be used with children who speak regional varieties of English Our reply issimply lsquoyesrsquo but we will of course provide justi cation for this response We thenmove on to points raised by Law about our sample A section dealing with thecontent of the test follows and nally we review the theoretical motivationsunderpinning the test and include a consideration here of Ballrsquos view of the test asa more advanced version of LARSP we feel able to reassure him that it is not

Special populations

The earlier versions of the Reynell Developmental Language Scales (RDLS) haddirections for modifying the test for use with special populations Given that manyspeech and language therapists work with children who have motor and sensoryimpairments the non-inclusion of such guidance in this new edition might seem agrave omission Lees expresses regret that we did not include a version of RDLSIII suitable for use with physically impaired children especially as earlier versionsof the test included a Comprehension Scale B for use with children who hadphysical diYculties manipulating the small toys Instructions in that section weretherefore modi ed so that children could use lsquogross arm movementsrsquo or lsquoeye-pointingrsquo (Reynell 1977) Instructions were given to space the toys so that thechildrsquos response could be interpreted clearly Additional modi cations were sug-gested for section 8 No information was given in the manual on trialling orstandardizing the test using these modi cations However in a separate chapter inthe 1985 version (Reynell and Huntley 1985) which oVers directions for using thetest with hearing-impaired children the authors state that there was no attempt tostandardize the test on deaf children

While we accept that clinicians have found these suggested modi cations helpfuland will inevitably be resourceful in adapting tests to use with multi-handicappedchildren we felt uneasy about including such advice in our revised version Therewere two main reasons for this reluctance First there is growing awareness that itis very diYcult to interpret non-verbal gestures such as eye-pointing in a consistentand reliable way especially in the absence of specialist recording equipment Secondwe opined that if such a scale was to be constructed then it would be important totrial items on the target population as well as a normal population This extensionto our work would be a worthwhile additional project A project of this kind couldalso incorporate technological developments that have taken place since the con-struction of the B scale notably computer-based systems that may compensate fordiYculty with ne motor control

What about the concerns raised by the commentators concerning regionalvarieties of English What approaches are open to those who seek to develop astandardized assessment procedure One possibility is to develop as many variety-speci c forms of the assessment items as there are varieties in the targeted usercommunities and standardize each on its proper variety We know of no assessment

Clinical Forum186

procedure that has attempted this certainly we would not have been able to do soin terms of our brief and it may simply be impossible in practice given the rangeof varieties in potential user communities Another possibility is to ignore the issueentirely and standardize on one standard variety We opted for a third approachwhich was to standardize on the standard variety and suggest guidelines foraccommodating other forms

Among these we may distinguish between cases where the local non-standardvariety lacks a contrast that the standard has such as the lack of third personsingular present -s in East Anglian English and those where the local variety eitherhas a further contrast or a diVerent form In the former case we see no optionbut to score the unmarked form as appropriate that is interpret it as a good localtoken The danger of this is a tendency to overestimate the development of thechildrsquos language abilitymdashbut we suggest that this eVect will be so slight as to beswamped by the other items contributing to the score at least as far as mostvarieties are concerned

Overseas in situations such as are found in Singapore we are aware thatlsquoSinglishrsquo is characterized by very widespread non-occurrence of determiners copulado-support pronominal case marking and so on (as well as by a host of occurringfeatures which are not found in British English eg utterance particles such as-lah) What we have in such situations is a varietal continuum among the samespeakers from least formal (most diVerent) to most formal (most similar to standardBritishAmerican English) Assessment in such a situation may appropriately tapthe childrsquos ability to move to the formal end of the continuum even though thisis not what would be used in more colloquial settings

Closer to home Ball gives us the example of Northern Ireland (NI) Englishtags as in The boy wants to go out so he does (cp doesnrsquot he) So how would our guidancebe acted upon in this case In respect of pronominal form and auxiliary selectionthere is no diVerence but the NI form shows no inversion or feature of oppositepolarity (assuming the corresponding negative would be something like The boydoesnrsquot want to go out nor he doesso he doesnrsquot) On the other hand the NI form doesrequire the pro-form so We would expect the local clinician to score such NI tagsas appropriate and we feel in the absence of speci c developmental informationthat this would not lead to a distortion of the resulting global score If the childcould not produce such a tag form the interpretation would proceed as for thestandard case

The standardization population

Regarding the appropriateness of the standardization sample Law notes that thesocioeconomic status of children in our standardization sample is not addressedand considers the issue of using a lsquotruncatedrsquo sample To consider socioeconomicstatus rst in our remarks above we have already trespassed on the borders ofsocial variation but here the point at issue is that of social class eVects on languagedevelopment Our view on this is that it is likely that social class aVects the rate ifnot the forms (lsquocodersquo) of language development The problem lies in knowingprecisely what these eVects are in terms of which aspects of language developmentare involved at what points in the developmental curve and in relation to whichpoints on the social class continuum Even if we had precise information the

Clinical Forum 187

problem of how to standardize in the face of it would remain (much as for theregional varieties above)

We take some comfort from the ndings of the Wells Bristol Project thatlsquoFamily Background accounts for only 16 of the variance in language developmentusing the language development scale score at the age of 42 monthsrsquo (Wells 1985p 347) We should recall also that the bulk of the eVect in this as in other studiesthat have tried to reveal it derives from the extreme points of the social scale Inview of this and the consideration that the Bristol Project population substantiallyover-represented social classes A and D Wells concludes that lsquoit can be argued thatthe obtained correlation almost certainly over-estimates the relationship betweenFamily Background and rate of language development in the population as a wholersquo(1985 pp 349ndash50) We also note that when parental occupation data available fora subset of the standardization sample for TROG was analysed the followingconclusion was drawn (note that no attempt was made to stratify the TROG samplein terms of social class) (Bishop 1989 p 6)

although social class had a statistically signi cant eVect on TROG performanceits absolute magnitude was small and one could be con dent that any bias in thesocial class distribution of the TROG sample would not have any serious eVecton the normative data

We would hazard a guess that other things being equal children from the extremesof social class variation may show up diVerently from each other and from themid-range population on the revised Reynell assessment Where individual childrenare placed in the lsquoat riskrsquo category as a result perhaps our view diVers from that inthe commentary we would not be averse to such children receiving speech andlanguage therapy input on this basis even if in the current conception of suchcategories they are not perceived as having some autonomous lsquolanguageimpairmentrsquo

Lawrsquos point about the truncated sample also needs to be addressed althoughas he observes we have dealt with this issue elsewhere (Edwards and Letts 1997)The issue is whether we want to compare the performance of the child in our clinicwith the lsquonormal childrsquo or whether we want to plot the performance of that childon a scale which represents performance of the total population If the former isthe aim then a truncated sample is needed if the latter then the second optionmust be preferred One could envisage the need for both Comparing the childrsquosperformance with norms arrived at from a truncated sample will inform the therapisthow far behind or advanced a child is compared with hisher normally developingpeers However if we are concerned with prioritizing case loads a case could bemade for the second option

We were persuaded to use a truncated sample for two reasons First there arethe practical diYculties of ensuring that one obtains a fully representative sampleof the total population (for example how does one ensure that all levels and typesof impairment are represented) Second we followed practice and custom oftesting procedure in this area (for example TROG and CELF Semel et al 1987)We therefore assume that clinicians tend to make judgements about a childrsquosperformance and thereafter decisions about management of that case based oncomparisons with the norm However this may not be the best way to proceedand Law has raised an interesting and important methodological point

Clinical Forum188

Content of the test and measuring eY cacy

What about the content of the test Lees is concerned that the test gives littleprominence to developmental pragmatics The issue here is not so much thatpragmatic functioning is not considered lsquoequally essentialrsquo as grammatical and lexicalskills but that despite much recent research in the area the assessment of pragmaticskill is still problematical In any case appropriate assessment of pragmatic skillnecessitates the collecting of data in naturalistic situations a diVerent assessmentregime In RDLS III we have not ignored pragmatic function altogether but haveconcentrated on aspects that can be looked at by means of a formal assessmentand about which enough is known to ensure that we are assessing these areasappropriately There is work on inferencing skills in children (see for exampleLeinonen and Letts 1997) and also on the use of pronouns as opposed to explicitreference in accord with given and new information in an utterance (see Hickman1995) The former is looked at in section J of the Comprehension Scale while thelatter is involved in successful performance on section D of the Expressive Scale

Lees is also concerned that the binary scoring used within the test may not besensitive enough for clinical purposes especially when it comes to measuring clinicaleYcacy While we would argue that RDLS III is more than a simple screeningdevice in that its use should give an indication of speci c areas of breakdown themanual makes clear that more detailed assessment of such areas may well benecessary in order to plan appropriate remediation To adopt a qualitative gradedscoring system may have been useful here but we were reluctant to do this sincewe anticipated potential severe problems with reliability of scoring across testersIn order to ensure such reliability experience with graded scoring systems in the eld of acquired aphasia suggests that testers would require a considerable numberof training hours For example the manual for the Porch Index of Communicative Ability(Porch 1973) states that lsquotraining can usually be completed in forty hours twentyhours of which are spent in testing ten patientsrsquo (p 13) The result for us wouldhave been a much less accessible test at present we believe that any individual witha speech and language therapy quali cation should be able to use RDLS III

In cases where further more focused assessment will be required we recommendthat formal specialized test procedures are used or that it may be necessary forthe therapist to design individualized probes To take one example of the latterperformance on section G of the Comprehension Scale and section Bii of theExpressive Scale may suggest that the child has not acquired prepositions In orderto establish the exact nature of the problem the therapist will need to look at atleast some of the following issues

E Which out of the set of English prepositions does the child nd diYcult tocomprehend or to produce Several trials may be needed with each type toestablish this

E Can the child cope with prepositions where minimal processing loads arerequired (ie relating two objects) as opposed to where a prepositionalphrase is embedded in a longer construction (again applying to both compre-hension and production)

E Is there variation in ability depending on the real-world correlates of therelationships expressed through prepositions for example an array of objectsas opposed to picture material

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 2: Reply Moving on

Clinical Forum 185

of interesting issues All three commentators raise queries about our methodologywhich are important in the current climate of lsquoevidence-based practicersquo Moregenerally we see Leesrsquo commentary as focusing on clinical issues Lawrsquos on method-ological issues and Ballrsquos on the historical background We will address each ofthese areas in turn We will start by responding to points made by Lees concerningthe target population of our test Our target population is all children whose rstlanguage is English but our commentators also raise the issue of whether the testcan be used with children who speak regional varieties of English Our reply issimply lsquoyesrsquo but we will of course provide justi cation for this response We thenmove on to points raised by Law about our sample A section dealing with thecontent of the test follows and nally we review the theoretical motivationsunderpinning the test and include a consideration here of Ballrsquos view of the test asa more advanced version of LARSP we feel able to reassure him that it is not

Special populations

The earlier versions of the Reynell Developmental Language Scales (RDLS) haddirections for modifying the test for use with special populations Given that manyspeech and language therapists work with children who have motor and sensoryimpairments the non-inclusion of such guidance in this new edition might seem agrave omission Lees expresses regret that we did not include a version of RDLSIII suitable for use with physically impaired children especially as earlier versionsof the test included a Comprehension Scale B for use with children who hadphysical diYculties manipulating the small toys Instructions in that section weretherefore modi ed so that children could use lsquogross arm movementsrsquo or lsquoeye-pointingrsquo (Reynell 1977) Instructions were given to space the toys so that thechildrsquos response could be interpreted clearly Additional modi cations were sug-gested for section 8 No information was given in the manual on trialling orstandardizing the test using these modi cations However in a separate chapter inthe 1985 version (Reynell and Huntley 1985) which oVers directions for using thetest with hearing-impaired children the authors state that there was no attempt tostandardize the test on deaf children

While we accept that clinicians have found these suggested modi cations helpfuland will inevitably be resourceful in adapting tests to use with multi-handicappedchildren we felt uneasy about including such advice in our revised version Therewere two main reasons for this reluctance First there is growing awareness that itis very diYcult to interpret non-verbal gestures such as eye-pointing in a consistentand reliable way especially in the absence of specialist recording equipment Secondwe opined that if such a scale was to be constructed then it would be important totrial items on the target population as well as a normal population This extensionto our work would be a worthwhile additional project A project of this kind couldalso incorporate technological developments that have taken place since the con-struction of the B scale notably computer-based systems that may compensate fordiYculty with ne motor control

What about the concerns raised by the commentators concerning regionalvarieties of English What approaches are open to those who seek to develop astandardized assessment procedure One possibility is to develop as many variety-speci c forms of the assessment items as there are varieties in the targeted usercommunities and standardize each on its proper variety We know of no assessment

Clinical Forum186

procedure that has attempted this certainly we would not have been able to do soin terms of our brief and it may simply be impossible in practice given the rangeof varieties in potential user communities Another possibility is to ignore the issueentirely and standardize on one standard variety We opted for a third approachwhich was to standardize on the standard variety and suggest guidelines foraccommodating other forms

Among these we may distinguish between cases where the local non-standardvariety lacks a contrast that the standard has such as the lack of third personsingular present -s in East Anglian English and those where the local variety eitherhas a further contrast or a diVerent form In the former case we see no optionbut to score the unmarked form as appropriate that is interpret it as a good localtoken The danger of this is a tendency to overestimate the development of thechildrsquos language abilitymdashbut we suggest that this eVect will be so slight as to beswamped by the other items contributing to the score at least as far as mostvarieties are concerned

Overseas in situations such as are found in Singapore we are aware thatlsquoSinglishrsquo is characterized by very widespread non-occurrence of determiners copulado-support pronominal case marking and so on (as well as by a host of occurringfeatures which are not found in British English eg utterance particles such as-lah) What we have in such situations is a varietal continuum among the samespeakers from least formal (most diVerent) to most formal (most similar to standardBritishAmerican English) Assessment in such a situation may appropriately tapthe childrsquos ability to move to the formal end of the continuum even though thisis not what would be used in more colloquial settings

Closer to home Ball gives us the example of Northern Ireland (NI) Englishtags as in The boy wants to go out so he does (cp doesnrsquot he) So how would our guidancebe acted upon in this case In respect of pronominal form and auxiliary selectionthere is no diVerence but the NI form shows no inversion or feature of oppositepolarity (assuming the corresponding negative would be something like The boydoesnrsquot want to go out nor he doesso he doesnrsquot) On the other hand the NI form doesrequire the pro-form so We would expect the local clinician to score such NI tagsas appropriate and we feel in the absence of speci c developmental informationthat this would not lead to a distortion of the resulting global score If the childcould not produce such a tag form the interpretation would proceed as for thestandard case

The standardization population

Regarding the appropriateness of the standardization sample Law notes that thesocioeconomic status of children in our standardization sample is not addressedand considers the issue of using a lsquotruncatedrsquo sample To consider socioeconomicstatus rst in our remarks above we have already trespassed on the borders ofsocial variation but here the point at issue is that of social class eVects on languagedevelopment Our view on this is that it is likely that social class aVects the rate ifnot the forms (lsquocodersquo) of language development The problem lies in knowingprecisely what these eVects are in terms of which aspects of language developmentare involved at what points in the developmental curve and in relation to whichpoints on the social class continuum Even if we had precise information the

Clinical Forum 187

problem of how to standardize in the face of it would remain (much as for theregional varieties above)

We take some comfort from the ndings of the Wells Bristol Project thatlsquoFamily Background accounts for only 16 of the variance in language developmentusing the language development scale score at the age of 42 monthsrsquo (Wells 1985p 347) We should recall also that the bulk of the eVect in this as in other studiesthat have tried to reveal it derives from the extreme points of the social scale Inview of this and the consideration that the Bristol Project population substantiallyover-represented social classes A and D Wells concludes that lsquoit can be argued thatthe obtained correlation almost certainly over-estimates the relationship betweenFamily Background and rate of language development in the population as a wholersquo(1985 pp 349ndash50) We also note that when parental occupation data available fora subset of the standardization sample for TROG was analysed the followingconclusion was drawn (note that no attempt was made to stratify the TROG samplein terms of social class) (Bishop 1989 p 6)

although social class had a statistically signi cant eVect on TROG performanceits absolute magnitude was small and one could be con dent that any bias in thesocial class distribution of the TROG sample would not have any serious eVecton the normative data

We would hazard a guess that other things being equal children from the extremesof social class variation may show up diVerently from each other and from themid-range population on the revised Reynell assessment Where individual childrenare placed in the lsquoat riskrsquo category as a result perhaps our view diVers from that inthe commentary we would not be averse to such children receiving speech andlanguage therapy input on this basis even if in the current conception of suchcategories they are not perceived as having some autonomous lsquolanguageimpairmentrsquo

Lawrsquos point about the truncated sample also needs to be addressed althoughas he observes we have dealt with this issue elsewhere (Edwards and Letts 1997)The issue is whether we want to compare the performance of the child in our clinicwith the lsquonormal childrsquo or whether we want to plot the performance of that childon a scale which represents performance of the total population If the former isthe aim then a truncated sample is needed if the latter then the second optionmust be preferred One could envisage the need for both Comparing the childrsquosperformance with norms arrived at from a truncated sample will inform the therapisthow far behind or advanced a child is compared with hisher normally developingpeers However if we are concerned with prioritizing case loads a case could bemade for the second option

We were persuaded to use a truncated sample for two reasons First there arethe practical diYculties of ensuring that one obtains a fully representative sampleof the total population (for example how does one ensure that all levels and typesof impairment are represented) Second we followed practice and custom oftesting procedure in this area (for example TROG and CELF Semel et al 1987)We therefore assume that clinicians tend to make judgements about a childrsquosperformance and thereafter decisions about management of that case based oncomparisons with the norm However this may not be the best way to proceedand Law has raised an interesting and important methodological point

Clinical Forum188

Content of the test and measuring eY cacy

What about the content of the test Lees is concerned that the test gives littleprominence to developmental pragmatics The issue here is not so much thatpragmatic functioning is not considered lsquoequally essentialrsquo as grammatical and lexicalskills but that despite much recent research in the area the assessment of pragmaticskill is still problematical In any case appropriate assessment of pragmatic skillnecessitates the collecting of data in naturalistic situations a diVerent assessmentregime In RDLS III we have not ignored pragmatic function altogether but haveconcentrated on aspects that can be looked at by means of a formal assessmentand about which enough is known to ensure that we are assessing these areasappropriately There is work on inferencing skills in children (see for exampleLeinonen and Letts 1997) and also on the use of pronouns as opposed to explicitreference in accord with given and new information in an utterance (see Hickman1995) The former is looked at in section J of the Comprehension Scale while thelatter is involved in successful performance on section D of the Expressive Scale

Lees is also concerned that the binary scoring used within the test may not besensitive enough for clinical purposes especially when it comes to measuring clinicaleYcacy While we would argue that RDLS III is more than a simple screeningdevice in that its use should give an indication of speci c areas of breakdown themanual makes clear that more detailed assessment of such areas may well benecessary in order to plan appropriate remediation To adopt a qualitative gradedscoring system may have been useful here but we were reluctant to do this sincewe anticipated potential severe problems with reliability of scoring across testersIn order to ensure such reliability experience with graded scoring systems in the eld of acquired aphasia suggests that testers would require a considerable numberof training hours For example the manual for the Porch Index of Communicative Ability(Porch 1973) states that lsquotraining can usually be completed in forty hours twentyhours of which are spent in testing ten patientsrsquo (p 13) The result for us wouldhave been a much less accessible test at present we believe that any individual witha speech and language therapy quali cation should be able to use RDLS III

In cases where further more focused assessment will be required we recommendthat formal specialized test procedures are used or that it may be necessary forthe therapist to design individualized probes To take one example of the latterperformance on section G of the Comprehension Scale and section Bii of theExpressive Scale may suggest that the child has not acquired prepositions In orderto establish the exact nature of the problem the therapist will need to look at atleast some of the following issues

E Which out of the set of English prepositions does the child nd diYcult tocomprehend or to produce Several trials may be needed with each type toestablish this

E Can the child cope with prepositions where minimal processing loads arerequired (ie relating two objects) as opposed to where a prepositionalphrase is embedded in a longer construction (again applying to both compre-hension and production)

E Is there variation in ability depending on the real-world correlates of therelationships expressed through prepositions for example an array of objectsas opposed to picture material

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 3: Reply Moving on

Clinical Forum186

procedure that has attempted this certainly we would not have been able to do soin terms of our brief and it may simply be impossible in practice given the rangeof varieties in potential user communities Another possibility is to ignore the issueentirely and standardize on one standard variety We opted for a third approachwhich was to standardize on the standard variety and suggest guidelines foraccommodating other forms

Among these we may distinguish between cases where the local non-standardvariety lacks a contrast that the standard has such as the lack of third personsingular present -s in East Anglian English and those where the local variety eitherhas a further contrast or a diVerent form In the former case we see no optionbut to score the unmarked form as appropriate that is interpret it as a good localtoken The danger of this is a tendency to overestimate the development of thechildrsquos language abilitymdashbut we suggest that this eVect will be so slight as to beswamped by the other items contributing to the score at least as far as mostvarieties are concerned

Overseas in situations such as are found in Singapore we are aware thatlsquoSinglishrsquo is characterized by very widespread non-occurrence of determiners copulado-support pronominal case marking and so on (as well as by a host of occurringfeatures which are not found in British English eg utterance particles such as-lah) What we have in such situations is a varietal continuum among the samespeakers from least formal (most diVerent) to most formal (most similar to standardBritishAmerican English) Assessment in such a situation may appropriately tapthe childrsquos ability to move to the formal end of the continuum even though thisis not what would be used in more colloquial settings

Closer to home Ball gives us the example of Northern Ireland (NI) Englishtags as in The boy wants to go out so he does (cp doesnrsquot he) So how would our guidancebe acted upon in this case In respect of pronominal form and auxiliary selectionthere is no diVerence but the NI form shows no inversion or feature of oppositepolarity (assuming the corresponding negative would be something like The boydoesnrsquot want to go out nor he doesso he doesnrsquot) On the other hand the NI form doesrequire the pro-form so We would expect the local clinician to score such NI tagsas appropriate and we feel in the absence of speci c developmental informationthat this would not lead to a distortion of the resulting global score If the childcould not produce such a tag form the interpretation would proceed as for thestandard case

The standardization population

Regarding the appropriateness of the standardization sample Law notes that thesocioeconomic status of children in our standardization sample is not addressedand considers the issue of using a lsquotruncatedrsquo sample To consider socioeconomicstatus rst in our remarks above we have already trespassed on the borders ofsocial variation but here the point at issue is that of social class eVects on languagedevelopment Our view on this is that it is likely that social class aVects the rate ifnot the forms (lsquocodersquo) of language development The problem lies in knowingprecisely what these eVects are in terms of which aspects of language developmentare involved at what points in the developmental curve and in relation to whichpoints on the social class continuum Even if we had precise information the

Clinical Forum 187

problem of how to standardize in the face of it would remain (much as for theregional varieties above)

We take some comfort from the ndings of the Wells Bristol Project thatlsquoFamily Background accounts for only 16 of the variance in language developmentusing the language development scale score at the age of 42 monthsrsquo (Wells 1985p 347) We should recall also that the bulk of the eVect in this as in other studiesthat have tried to reveal it derives from the extreme points of the social scale Inview of this and the consideration that the Bristol Project population substantiallyover-represented social classes A and D Wells concludes that lsquoit can be argued thatthe obtained correlation almost certainly over-estimates the relationship betweenFamily Background and rate of language development in the population as a wholersquo(1985 pp 349ndash50) We also note that when parental occupation data available fora subset of the standardization sample for TROG was analysed the followingconclusion was drawn (note that no attempt was made to stratify the TROG samplein terms of social class) (Bishop 1989 p 6)

although social class had a statistically signi cant eVect on TROG performanceits absolute magnitude was small and one could be con dent that any bias in thesocial class distribution of the TROG sample would not have any serious eVecton the normative data

We would hazard a guess that other things being equal children from the extremesof social class variation may show up diVerently from each other and from themid-range population on the revised Reynell assessment Where individual childrenare placed in the lsquoat riskrsquo category as a result perhaps our view diVers from that inthe commentary we would not be averse to such children receiving speech andlanguage therapy input on this basis even if in the current conception of suchcategories they are not perceived as having some autonomous lsquolanguageimpairmentrsquo

Lawrsquos point about the truncated sample also needs to be addressed althoughas he observes we have dealt with this issue elsewhere (Edwards and Letts 1997)The issue is whether we want to compare the performance of the child in our clinicwith the lsquonormal childrsquo or whether we want to plot the performance of that childon a scale which represents performance of the total population If the former isthe aim then a truncated sample is needed if the latter then the second optionmust be preferred One could envisage the need for both Comparing the childrsquosperformance with norms arrived at from a truncated sample will inform the therapisthow far behind or advanced a child is compared with hisher normally developingpeers However if we are concerned with prioritizing case loads a case could bemade for the second option

We were persuaded to use a truncated sample for two reasons First there arethe practical diYculties of ensuring that one obtains a fully representative sampleof the total population (for example how does one ensure that all levels and typesof impairment are represented) Second we followed practice and custom oftesting procedure in this area (for example TROG and CELF Semel et al 1987)We therefore assume that clinicians tend to make judgements about a childrsquosperformance and thereafter decisions about management of that case based oncomparisons with the norm However this may not be the best way to proceedand Law has raised an interesting and important methodological point

Clinical Forum188

Content of the test and measuring eY cacy

What about the content of the test Lees is concerned that the test gives littleprominence to developmental pragmatics The issue here is not so much thatpragmatic functioning is not considered lsquoequally essentialrsquo as grammatical and lexicalskills but that despite much recent research in the area the assessment of pragmaticskill is still problematical In any case appropriate assessment of pragmatic skillnecessitates the collecting of data in naturalistic situations a diVerent assessmentregime In RDLS III we have not ignored pragmatic function altogether but haveconcentrated on aspects that can be looked at by means of a formal assessmentand about which enough is known to ensure that we are assessing these areasappropriately There is work on inferencing skills in children (see for exampleLeinonen and Letts 1997) and also on the use of pronouns as opposed to explicitreference in accord with given and new information in an utterance (see Hickman1995) The former is looked at in section J of the Comprehension Scale while thelatter is involved in successful performance on section D of the Expressive Scale

Lees is also concerned that the binary scoring used within the test may not besensitive enough for clinical purposes especially when it comes to measuring clinicaleYcacy While we would argue that RDLS III is more than a simple screeningdevice in that its use should give an indication of speci c areas of breakdown themanual makes clear that more detailed assessment of such areas may well benecessary in order to plan appropriate remediation To adopt a qualitative gradedscoring system may have been useful here but we were reluctant to do this sincewe anticipated potential severe problems with reliability of scoring across testersIn order to ensure such reliability experience with graded scoring systems in the eld of acquired aphasia suggests that testers would require a considerable numberof training hours For example the manual for the Porch Index of Communicative Ability(Porch 1973) states that lsquotraining can usually be completed in forty hours twentyhours of which are spent in testing ten patientsrsquo (p 13) The result for us wouldhave been a much less accessible test at present we believe that any individual witha speech and language therapy quali cation should be able to use RDLS III

In cases where further more focused assessment will be required we recommendthat formal specialized test procedures are used or that it may be necessary forthe therapist to design individualized probes To take one example of the latterperformance on section G of the Comprehension Scale and section Bii of theExpressive Scale may suggest that the child has not acquired prepositions In orderto establish the exact nature of the problem the therapist will need to look at atleast some of the following issues

E Which out of the set of English prepositions does the child nd diYcult tocomprehend or to produce Several trials may be needed with each type toestablish this

E Can the child cope with prepositions where minimal processing loads arerequired (ie relating two objects) as opposed to where a prepositionalphrase is embedded in a longer construction (again applying to both compre-hension and production)

E Is there variation in ability depending on the real-world correlates of therelationships expressed through prepositions for example an array of objectsas opposed to picture material

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 4: Reply Moving on

Clinical Forum 187

problem of how to standardize in the face of it would remain (much as for theregional varieties above)

We take some comfort from the ndings of the Wells Bristol Project thatlsquoFamily Background accounts for only 16 of the variance in language developmentusing the language development scale score at the age of 42 monthsrsquo (Wells 1985p 347) We should recall also that the bulk of the eVect in this as in other studiesthat have tried to reveal it derives from the extreme points of the social scale Inview of this and the consideration that the Bristol Project population substantiallyover-represented social classes A and D Wells concludes that lsquoit can be argued thatthe obtained correlation almost certainly over-estimates the relationship betweenFamily Background and rate of language development in the population as a wholersquo(1985 pp 349ndash50) We also note that when parental occupation data available fora subset of the standardization sample for TROG was analysed the followingconclusion was drawn (note that no attempt was made to stratify the TROG samplein terms of social class) (Bishop 1989 p 6)

although social class had a statistically signi cant eVect on TROG performanceits absolute magnitude was small and one could be con dent that any bias in thesocial class distribution of the TROG sample would not have any serious eVecton the normative data

We would hazard a guess that other things being equal children from the extremesof social class variation may show up diVerently from each other and from themid-range population on the revised Reynell assessment Where individual childrenare placed in the lsquoat riskrsquo category as a result perhaps our view diVers from that inthe commentary we would not be averse to such children receiving speech andlanguage therapy input on this basis even if in the current conception of suchcategories they are not perceived as having some autonomous lsquolanguageimpairmentrsquo

Lawrsquos point about the truncated sample also needs to be addressed althoughas he observes we have dealt with this issue elsewhere (Edwards and Letts 1997)The issue is whether we want to compare the performance of the child in our clinicwith the lsquonormal childrsquo or whether we want to plot the performance of that childon a scale which represents performance of the total population If the former isthe aim then a truncated sample is needed if the latter then the second optionmust be preferred One could envisage the need for both Comparing the childrsquosperformance with norms arrived at from a truncated sample will inform the therapisthow far behind or advanced a child is compared with hisher normally developingpeers However if we are concerned with prioritizing case loads a case could bemade for the second option

We were persuaded to use a truncated sample for two reasons First there arethe practical diYculties of ensuring that one obtains a fully representative sampleof the total population (for example how does one ensure that all levels and typesof impairment are represented) Second we followed practice and custom oftesting procedure in this area (for example TROG and CELF Semel et al 1987)We therefore assume that clinicians tend to make judgements about a childrsquosperformance and thereafter decisions about management of that case based oncomparisons with the norm However this may not be the best way to proceedand Law has raised an interesting and important methodological point

Clinical Forum188

Content of the test and measuring eY cacy

What about the content of the test Lees is concerned that the test gives littleprominence to developmental pragmatics The issue here is not so much thatpragmatic functioning is not considered lsquoequally essentialrsquo as grammatical and lexicalskills but that despite much recent research in the area the assessment of pragmaticskill is still problematical In any case appropriate assessment of pragmatic skillnecessitates the collecting of data in naturalistic situations a diVerent assessmentregime In RDLS III we have not ignored pragmatic function altogether but haveconcentrated on aspects that can be looked at by means of a formal assessmentand about which enough is known to ensure that we are assessing these areasappropriately There is work on inferencing skills in children (see for exampleLeinonen and Letts 1997) and also on the use of pronouns as opposed to explicitreference in accord with given and new information in an utterance (see Hickman1995) The former is looked at in section J of the Comprehension Scale while thelatter is involved in successful performance on section D of the Expressive Scale

Lees is also concerned that the binary scoring used within the test may not besensitive enough for clinical purposes especially when it comes to measuring clinicaleYcacy While we would argue that RDLS III is more than a simple screeningdevice in that its use should give an indication of speci c areas of breakdown themanual makes clear that more detailed assessment of such areas may well benecessary in order to plan appropriate remediation To adopt a qualitative gradedscoring system may have been useful here but we were reluctant to do this sincewe anticipated potential severe problems with reliability of scoring across testersIn order to ensure such reliability experience with graded scoring systems in the eld of acquired aphasia suggests that testers would require a considerable numberof training hours For example the manual for the Porch Index of Communicative Ability(Porch 1973) states that lsquotraining can usually be completed in forty hours twentyhours of which are spent in testing ten patientsrsquo (p 13) The result for us wouldhave been a much less accessible test at present we believe that any individual witha speech and language therapy quali cation should be able to use RDLS III

In cases where further more focused assessment will be required we recommendthat formal specialized test procedures are used or that it may be necessary forthe therapist to design individualized probes To take one example of the latterperformance on section G of the Comprehension Scale and section Bii of theExpressive Scale may suggest that the child has not acquired prepositions In orderto establish the exact nature of the problem the therapist will need to look at atleast some of the following issues

E Which out of the set of English prepositions does the child nd diYcult tocomprehend or to produce Several trials may be needed with each type toestablish this

E Can the child cope with prepositions where minimal processing loads arerequired (ie relating two objects) as opposed to where a prepositionalphrase is embedded in a longer construction (again applying to both compre-hension and production)

E Is there variation in ability depending on the real-world correlates of therelationships expressed through prepositions for example an array of objectsas opposed to picture material

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 5: Reply Moving on

Clinical Forum188

Content of the test and measuring eY cacy

What about the content of the test Lees is concerned that the test gives littleprominence to developmental pragmatics The issue here is not so much thatpragmatic functioning is not considered lsquoequally essentialrsquo as grammatical and lexicalskills but that despite much recent research in the area the assessment of pragmaticskill is still problematical In any case appropriate assessment of pragmatic skillnecessitates the collecting of data in naturalistic situations a diVerent assessmentregime In RDLS III we have not ignored pragmatic function altogether but haveconcentrated on aspects that can be looked at by means of a formal assessmentand about which enough is known to ensure that we are assessing these areasappropriately There is work on inferencing skills in children (see for exampleLeinonen and Letts 1997) and also on the use of pronouns as opposed to explicitreference in accord with given and new information in an utterance (see Hickman1995) The former is looked at in section J of the Comprehension Scale while thelatter is involved in successful performance on section D of the Expressive Scale

Lees is also concerned that the binary scoring used within the test may not besensitive enough for clinical purposes especially when it comes to measuring clinicaleYcacy While we would argue that RDLS III is more than a simple screeningdevice in that its use should give an indication of speci c areas of breakdown themanual makes clear that more detailed assessment of such areas may well benecessary in order to plan appropriate remediation To adopt a qualitative gradedscoring system may have been useful here but we were reluctant to do this sincewe anticipated potential severe problems with reliability of scoring across testersIn order to ensure such reliability experience with graded scoring systems in the eld of acquired aphasia suggests that testers would require a considerable numberof training hours For example the manual for the Porch Index of Communicative Ability(Porch 1973) states that lsquotraining can usually be completed in forty hours twentyhours of which are spent in testing ten patientsrsquo (p 13) The result for us wouldhave been a much less accessible test at present we believe that any individual witha speech and language therapy quali cation should be able to use RDLS III

In cases where further more focused assessment will be required we recommendthat formal specialized test procedures are used or that it may be necessary forthe therapist to design individualized probes To take one example of the latterperformance on section G of the Comprehension Scale and section Bii of theExpressive Scale may suggest that the child has not acquired prepositions In orderto establish the exact nature of the problem the therapist will need to look at atleast some of the following issues

E Which out of the set of English prepositions does the child nd diYcult tocomprehend or to produce Several trials may be needed with each type toestablish this

E Can the child cope with prepositions where minimal processing loads arerequired (ie relating two objects) as opposed to where a prepositionalphrase is embedded in a longer construction (again applying to both compre-hension and production)

E Is there variation in ability depending on the real-world correlates of therelationships expressed through prepositions for example an array of objectsas opposed to picture material

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 6: Reply Moving on

Clinical Forum 189

E How does the child cope with conventional use of prepositions in Englishwhich are idiosyncratic and do not conform to the canonical semanticrelationship (eg you hang a picture on the wall whereas it is in fact lsquoagainstrsquoor lsquonext torsquo the wall)

E Can the child deal with prepositions that relate to time rather than location(eg lsquoJack arrived at work before Tomrsquo)

Obviously RDLS-III can only serve to alert the clinician to such problems (especi-ally as there is only one prepositional phrase elicited in Expressive Scale Bii) Inpractice many therapists will use a diagnostic therapy approach in order to locatethe level of breakdown at which to concentrate their eVorts

Lees also comments on issues of eYcacy We only claim that RDLS III lsquocancontributersquo to eYcacy measurement (p 30) and certainly not that it can do thewhole lsquomeasuring treatment eYcacy jobrsquo We recommend that the test should notbe repeated at less than 6 month intervals if this recommendation is followed thenobviously the test is unsuitable for the measurement of short-term gains Where anattempt is made to measure the eYcacy of a speci c treatment regime targeting aspeci c aspect of language baselines of performance need to be established forthat particular aspect (and preferably for another aspect as well which can act as alsquocontrolrsquo) Clearly more qualitative approaches are needed here arguably the sortof individually designed probes discussed above in order to show whether acceler-ated progress in the treatment area has been made However the longer termeVectiveness of remediation programmes also needs to be considered and it is herethat RDLS III should be most useful If this is what Lees means by lsquooutcomemeasures in a more general sensersquo then we are in agreement here although obviouslyfurther measures are also relevant such as parental satisfaction childrsquos con denceprogress in school and so forth EYcacy studies require a robust measure as astarting point with children this needs to be norm-referenced so that overall gainscan be shown to be additional to those resulting from normal maturation Enderbyand Emerson (1995 p 52) argue for the need for a lsquocore information set so thatresearchers in the eld can gather similar information on clientsrsquo in order toestablish when a child can bene t from intervention We would argue that testingwith RDLS III can contribute to such a set of information

Theoretical motivation

Ball speculates about the in uence of LARSP on this new test We have consideredthe development of RDLS III to be a separate and quite diVerent project from thedevelopment of LARSP originally published over 20 years ago (Crystal et al 1976)Ball asks how the work on linguistic assessment of some members of the develop-ment team has been informative in the revision of the Reynell Scales We haveexplicitly stated in the manual (Edwards et al 1997) and in the lead paper for thisforum that our brief for RDLS III was to provide a test which considered a childrsquoslanguage from a developmental perspective In doing this we have referred to alarge body of research and list in the lead paper and in the manual a number ofkey references The theoretical background then is much broader than Ballrsquos papermight suggest but given the popularity of the pro ling approach it is worth spendingsome time responding to some of the points that Ball raises concerning LARSP

The weaknesses of previous versions of RDLS lay in a number of areas butnowhere more strikingly than in the expressive scale so here was an obvious point

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 7: Reply Moving on

Clinical Forum190

of contact between our experience and the task in hand It is perhaps worthemphasizing that part of our experience is that we are acutely conscious of theweaknesses of LARSP A productive challenge therefore faced us in this respectSome of the unresolved diYculties with LARSP as a pro le of language abilitieshave to do with the way that structures are counted for the purpose of quantifyingpro le patterns the lack of coding for certain aspects of syntactic informationespecially adverbials and complex sentences and the lack of coding for vocabularyin general These points have been addressed in a number of places ( ie Fletcheret al 1986 Fletcher and Garman 1988 French 1988) Subsequently we havedeveloped utterance segmentation heuristics further laying the basis for a morere ned approach to complex grammatical constructions and the structure of dis-course generally (Garman 1989a Garman and Edwards 1995) Developments suchas STASS have also been useful in providing a practical elicitation format forLARSP-based structures but it should be noted that these do not go beyond stageIV of the pro le

But what of the scoring of grammar One of our criticisms of the older Reynellis of the selective approach taken to syntactic analysis A selective approach iswrong if there is no basis for the selection This is what we criticized at the timethat LARSP was developed for example lsquowhen Reynell (1969) isolates pronounsprepositions conjunctions and questions (other than intonational) as four categoriesto be scored (item 11 on the Expressive Developmental Language Scale) one wantsto know on what grounds this particular selection of features was madersquo (Crystalet al 1976 p 15) If we can point to the selection criteria in terms of child languagedevelopment research ndings then our point can be satis ed

This leads to the other issue of scoring as opposed to pro ling grammaticalabilities We see pro le and score systems as having diVerent and complementarypurposes All scores are simpli cations in as much as they allow for the same scoreto be arrived at by diVerent means But data reduction is no less a proper procedurefor assessment than it is for research what is important is the rationale by whichthis data reduction is achieved As we have pointed out elsewhere there areassessment procedures that emphasize their nal score but which have usefulimplicit pro les to aid the interpretation of the score (eg Leersquos 1974 DevelopmentalSentence Analysis see Garman 1989b) The important point is to have a pro le logicbehind the items in the assessment that contribute to the score The developmentalrationale of the Reynell Scales provided for this in a way that for example theBPVS does not (since it lacks a semantic logic to the selection of items) and thiswas certainly an aspect which we wanted to retain and if possible strengthen inour revision

Thus we saw an opportunity to look again at the issue of devising an eVectivestructured and quanti able assessment procedure that would t the aims of theoriginal Reynell in terms of level of detail and scope of age range but which wouldlook in a more balanced way at expression alongside comprehension and bring itmore into line with current work on child language development

Conclusion

To conclude this response has tried to show the following First we have beenaware of the potential importance of issues such as socioeconomic status of thestandardization sample dialectal variation the dangers of using a truncated sample

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK

Page 8: Reply Moving on

Clinical Forum 191

and the usefulness of a test like RDLS III for physically impaired children andthose with pragmatic disorders We have been conscious of the need however toproduce a test which is of practical use to a large number of clinicians and wetrust that the decisions we have made in these areas are well motivated theoreticallyand empirically Second we are making no claim that RDLS III could function asthe only tool for use in eYcacy studies only that it can make a contribution to thewhole eld of measurement of clinical eVectiveness and outcomes of therapyFinally RDLS III is a new departure in the eld of child language testing buildingon traditions already established in the earlier RDLS versions and through otherprocedures such as LARSP but incorporating more recent research ndings andapproaches as well

References

Bishop D V M 1989 Test for Reception of Grammar 2nd edn (London MRC)Crystal D Fletcher P and Garman M 1976 The Grammatical Analysis of Language Disability

(London Edward Arnold)Edwards S and Letts C 1997 Reynell Scales [Letter] Bulletin of the Royal College of Speech and

Language Therapists 548 18Edwards S Fletcher P Garman M Hughes A Letts C and Sinka I 1997 Manual for the

Reynell Developmental Language Scales III (Windsor NFER-Nelson)Enderby P and Emerson J 1995 Does Speech and Language Therapy Work (London Whurr)Fletcher P and Garman M 1988 LARSPing by numbers a response to French British Journal of

Disorders of Communication 23 309ndash322Fletcher P Garman M Johnson M Schelletter C and Stodel L 1986 Characterizing

language impairment in terms of normal language development advantages and limitationsProceedings of the Seventh Annual Wisconsin Symposium on Research in Child Language Disorders (MadisonUniversity of Wisconsin)

French A 1988 The LARSP pro le of the normal 5 year old with special reference to phrasestructure British Journal of Disorders of Communication 23 293ndash308

Garman M 1989a The role of linguistics in speech therapy assessment and interpretation In PGrunwell and A James (Eds) The Functional Evaluation of Language Disorders (London CroomHelm) pp 29ndash57

Garman M 1989b Syntactic assessment of expressive language In K Grundy (Ed) Linguistics inClinical Practice (London Taylor amp Francis) pp 92ndash117

Garman M and Edwards S 1995 Syntactic assessment of expressive language In K Grundy (Ed)Linguistics in Clinical Practice 2nd edn (London Taylor amp Francis) pp 134ndash166

Hickman M 1995 Discourse organization and the development of reference to person space andtime In P Fletcher and B MacWhinney (Eds) The Handbook of Child Language (OxfordBlackwell) pp 194ndash218

Lee L 1974 Developmental Sentence Analysis a Grammatical Assessment Procedure for Speech and LanguageDisorders (Evanston IL Northwestern University)

Leinonen E and Letts C 1997 Why pragmatic impairment A case study in the comprehensionof inferential meaning European Journal of Disorders of Communication 32 35ndash52

Porch B 1973 Administration Scoring and Interpretation Porch Index of Communicative Ability (CaliforniaConsulting Psychological Press)

Reynell J 1969 Developmental Language Scale (Slough Bucks NFER)Reynell J K 1977 Manual for the Reynell Developmental Language Scale (revised) (Windsor NFER-

Nelson)Reynell J K and Huntley M 1985 Manual for the Reynell Developmental Language Scale (2nd revision)

(Windsor NFER-Nelson)Semel E Wiig E H and Secord W 1987 Clinical Evaluation of Language FundamentalsmdashRevised

(NY Psychological Corporation)Wells G 1985 Language Development in the Pre-school Years (Cambridge Cambridge University Press)

Address correspondence to Carolyn Letts Department of Linguistic Science The University ofReading Reading RG6 6AA UK