EDITORS: Ball, Martin J.; Crystal, David; Fletcher, Paul
TITLE: Assessing Grammar
SUBTITLE: The Languages of LARSP
SERIES: Communication Disorders Across Languages
PUBLISHER: Multilingual Matters
Leah R. Paltiel-Gedalyovich, “Hedim” Institutes of Audiology, Ber Sheva, Israel
In Martin J. Ball’s introduction he describes the purposes of this volume as including 1) composing a source book for clinicians working with various languages for which versions of the Language Assessment Remediation and Screening Procedure (LARSP) exist and 2) introducing new versions of LARSP for languages which had not yet had LARSP adapted to them. The chapters of the volume are of a (loosely) uniform format including: a description of the target language and acquisition of the target language grammar, information on the database used to develop the LARSP charts which plot grammatical acquisition, and the charts themselves. Differences between the charts result from cross-linguistic differences in grammar as well as from methodological issues. The volume reflects the cross-linguistic history of LARSP while clearly indicating that this history is not complete as LARSP continues to be modified and adapted to new languages.
Chapter 1: On the origin of LARSPecies (David Crystal)
This chapter begins with a history of the motivation for the original LARSP. LARSP came into being in an environment where speech-language clinicians had little or no training in child language development or in grammar. LARSP presented a structured way to view children’s grammatical typical and atypical performance and a way to plan remediation goals. Crystal argues that “grammar is the key to understanding language disability” (p. 6), despite the frequent (co-)occurrence of impaired vocabulary skills. The LARSP chart concentrated on development up to age 5. Although grammatical development certainly continues after this age, the rationale was that the majority of children seeking intervention would be functioning within this age range.
Practical considerations guided the development of a chart that would fit on a single page and be applicable given caseload and time constraints of clinicians. The tool aimed to cover four areas: screening, assessment, diagnosis and therapy. The organization of the chart allowed the clinician to refrain from analyzing problematic (and often unanalyzable) utterances. The final version of the chart included four sections: A. Time saving, B. Interaction, C. Types of sentence structure and function (seven stages of grammatical acquisition), and D. Summary. The chart aims to improve control and monitor treatment efficacy.
Chapter 2: LARSP thirty years on (Paul Fletcher, Thomas Klee and William Gavin)
This chapter looks at implications of developmental variability, the gradual nature of development, and grammatical ‘faultlines’ in the clinical application of LARSP.
Studies using the LARSP found stability in the order of development but variability in the ages of acquisition of various structures. The gradual nature of acquisition is seen in the use of certain structures with limited vocabulary in early stages and more productive use emerging with increased age.
The authors report data suggesting that an upwards revision of the ages of the various stages of the original LARSP based on developmental research would be appropriate for the stages up to Stage 5 (3 years).
The authors further point out that although linguistic theory has undergone many changes in the past thirty years, since LARSP provides a comprehensive description of the grammar that a child needs to acquire, and allows the mapping of where s/he stands in terms of the acquisition of these grammatical structures, LARSP remains a relevant and appropriate clinical (and research) tool.
Chapter 3: ‘Computerized Profiling’ of Clinical Language Samples and the Issue of Time (Steven Long)
In this chapter, Long deals with practical considerations in the clinical application of LARSP. Specifically, the linguistic knowledge required to use the LARSP profile is great and often not mastered by clinicians. Also, the time required to profile may be so great as to be prohibitive in many clinical settings. The computerized profile aims to support clinical knowledge and perform non ‘expert’ tasks such as quantification of results and search functions.
LARSP analysis involves creating a transcript file in the CORPUS module and then a tentative parse of each sentence is provided by the LARSP module. The parse must be checked manually. The program works hierarchically from clause to word structure. Incorrect parsings at higher levels affect lower levels but once these are changed manually the subsequent levels are automatically adjusted. The program assigns the structures to the appropriate stages and these are then quantified. A chart identical to the manual chart is generated. In a study comparing manual and computerized profiling, computerized profiling was found to be more accurate and significantly faster. Computerized analysis can cut the time needed significantly so that even the most complex sample in this study was analyzed in under an hour. Even so the procedure is still relatively time consuming. The justification for its use comes from the clinical benefits of a comprehensive profile of a child’s grammatical performance.
Chapter 4: HARSP: A developmental Language Profile for Hebrew (Ruth A. Berman and Lyle Lustigman)
This is the first chapter of the volume dealing with applications of the LARSP methodology to languages other than English. The chapter begins with an account of the database, which served as the material for the original HARSP (Hebrew Assessment, Remediation and Screening Procedure) analyses. The final database included cross-sectional data from 100 children aged 1;0-5;11 and longitudinal data from four more children aged 1;3-3;6. In addition the ages specified in the revised HARSP take into account developmental Hebrew language research. The version of the HARSP presented in this volume is considerably changed from an original draft version developed by Berman, Rom and Hirsch (1982).
The chapter continues with an outline of Hebrew grammar. The most obvious feature of Hebrew grammar which differs significantly from English is the rich morphology which affects verb, noun and adjective systems and results in complex agreement. Word order in Hebrew is mainly subject initial, except for existential, possessive and VS constructions. It is also flexible.
The HARSP chart follows the English version for its first four sections, which deal with ‘Types of Utterances’. The ‘Grammatical Analyses’ section has been adapted to suit Hebrew grammar. Six stages of language development are described from 0;9-4;0. The ages correspond to the LARSP ages. The HARSP differs from LARSP in that instead of a deviant category and the Stage 6 error box, an error line allows recording of errors at each of the stages 3-6. For Stages 1-2, utterances are analyzed at clause, phrase and word levels. For Stages 3-6 utterances are also analyzed for connectivity. Utterances are considered grammatical if they occur in adult colloquial Hebrew, regardless of their grammaticality in prescriptive terms. Errors are only marked if the child also demonstrates adult usage of the structure.
Chapter 5: Profiling linguistic disability in German-speaking children (Harald Clahsen and Detlef Hansen)
The LARSP was adapted to German in 1986 and COPROF, a computer-assisted analysis was published in 1991. The German version follows the LARSP procedure with several differences. Five developmental stages are described. The authors discuss the oral language sample itself, emphasizing the importance of collecting a spontaneous speech sample, and supplementing, but not relying solely on, elicited productions.
A summary of the German linguistic phenomena analyzed is given. The areas addressed by the chart include word and constituent structure, inflectional morphology and sentence structure. This section deals with the word order and combination of major constituents within a sentence. Word/phrase structure analysis includes nominal, adverbial and verbal elements, coordinating and subordinating conjunctions and case markings. The developmental ages suggested in the analysis are based on empirical studies of children acquiring German as their first language. After the general discussion of the chart, an illustrative case profile is presented.
The great amount of time required to complete a comprehensive analysis of a language sample has restricted clinical application of the COPROF. Attempts at shortcuts result in profiles that may provide screening but do not result in specific treatment goals. The computer-assisted profiling program is available at no cost, but even with the time saved with this, the authors point out that clinical feasibility of the tool is limited.
Chapter 6. GRAMAT: A Dutch adaptation of LARSP (Gerard W. Bol)
The Dutch version of LARSP resulted from an MA thesis and later doctoral dissertation by Bol and Kuiken (1988). The data for the original chart was based on data collected on 12 children at each 6 month interval between the ages of 1;0 and 4;0. Six stages of development were described.
The first four sections of the original LARSP chart are represented in two sections in the GRAMAT chart. Mean Length of Utterance (MLU) for the entire sample and also MLU of the five longest utterances are noted. The criterion for including a structure at a given stage was having that structure appear in utterances of 50% of the children at that age. Frequency charts for the structures of the chart are given. Diagnosis is based on the pattern of frequencies the child displays. Treatment goals relate to structures appearing above 75% or below 25%.
The use of the chart with a language disordered child is used as illustration. The child’s performance is compared to that of typically developing children of the same MLU. Their stage is determined by this measure and then the frequencies of their structures at this stage are compared to the frequencies for these structures of typically developing children, with structures below the 25th percentile and above the 75th percentile targets for therapy. In addition all structures are mapped on a complete chart. Therapy goals involve moving the child through the stages for those structure areas where there is infrequent use of structures.
Chapter 7: LLARSP: A grammatical profile for Welsh (Martin J. Ball and Enli Môn Thomas)
This chapter begins with a survey of bilingual Welsh-English acquisition in Wales. Great differences are found in the age of acquisition of various Welsh structures depending on the degree of exposure and use.
A review of some structures of Welsh grammar is given including gender assignment to nouns, adjective gender agreement and modification. Marking of gender in distant elements is variable even in adult speakers. Profiling soft mutation and plural morphology may be useful in diagnosing deviant language.
The greatest difficulty in the development of LLARSP was lack of developmental data. Removing ages of development makes the chart applicable to children from varied linguistic backgrounds. Language universals are good candidates for the profile in the absence of detailed language specific data.
Three charts result from the application of LARSP to Welsh: syntax (LLARSP-C), morphology (LLARSP-M) and mutations (LLARSP-T).
LLARSP-C: The LLARSP-C mirrors the LARSP up to Stage I, except for Section A which records code-switching. Different levels of code-switching are noted. Stages II-IV include changes related to differences between Welsh and English, including word-order differences. Transfer stages are retained. Stages V-VI are very similar to English. The chart is geared to spoken Welsh and structures which appear primarily in written Welsh are ignored.
LLARSP-M profiles the verb phrase, noun phrase, and prepositional and adjectival morphology. A summary includes a type:token ratio of types of morphemes and types of errors.
LLARSP-T profiles initial consonant mutations, three types of phonological change which are syntactically motivated. Errors are difficult to describe because of sociolinguistic and dialect differences but the profiler can adjust the profile accordingly.
The point is raised that the chart should possibly be in Welsh.
Chapter 8: An investigation of syntax in children of Bengali (Sylheti)-speaking families (Jane Stokes)
The Sylheti database was collected in the 1980’s and includes 30 children aged 1;6-4;0 years. As well as the syntactic analysis, MLU was calculated.
A short survey of Bengali is given. Bengali is the written version and Sylheti is the oral, rather different version. Basic word order is SOV. Verbal morphology is rich. Changes made to the chart to accommodate Bengali included changes in clause and phrase constituent order. At word level morphology was expanded to suit Bengali. At the two and three word levels similar categories to LARSP were used, but the stages were not preserved.
A strong correlation was found between age and MLU so ‘key features’ were described for MLU groups. The list of key features for a certain group included structures of varying length. At each stage negated utterances were the first long utterances to appear. Lexical differences in the utterances of the same length at each level are noted.
The author suggests that LARSP was the basis of a first framework for describing Bengali syntactic development, although the result of this description was not actually a Bengali version of LARSP.
Chapter 9. ILARSP: A grammatical profile of Irish (Tina Hickey)
The database comes from four typically developing, Irish speaking children ages 1;4-3;6. An outline of Irish grammar is provided. Irish is a VSX language, where the verb precedes the subject which is then followed by the remaining parts of the sentence. Phonological mutations are morphologically motivated. Age ranges are not discussed because of the limited sample. A main difference is the separate ‘Negative’ column. Transfer stages are preserved. Here, too, a code-switching section is included. Subscripts and superscripts and combinations are used to mark expansions unique to Irish. Allophonic mutations are also noted here. Mutation errors are noted separately.
Chapter 10: Persian: Devising the P-LARSP (Habibeh Samadi and Mick Perkins)
The data comes from the language of three children studied longitudinally. The ages covered are 1;8-3;4. A brief outline of Persian is provided. It is a pro-drop verb final language. Word order differs from English. Persian also differs from English as it has rich inflection. The analysis of inflection is related to the phrase level analysis. Transitional stages of LARSP are maintained as are the first four sections of the profile, as the authors consider these to be non-language specific.
Due to lack of data, ages are not assigned to the various stages. Generally, structures are assigned to the different stages based on number of constituents. Word level is depicted similarly to how it is done on the English chart, but inflections include nouns and verbs as well as aspectual inflections.
Chapter 11: Frisian TARSP. Based on the methodology of the Dutch TARSP (Jelske Dijkstra and Liesbeth Schlichting)
Frisian is a minority language spoken in the Netherlands. Most of its speakers also speak Dutch. The chart parallels the previous version of a Dutch chart. A survey of Frisian grammar is given. Word order differs from English and there is a richer inflectional system.
The database consisted of samples from 100 children aged 1;9-4;2. The children were divided into groups by 6 month age intervals, covering 2 SES levels and both sexes. The children came mainly from Frisian speaking families.
Up to three main clauses could be included in one utterance, with their accompanying
subordinate clauses. Developmental stages are determined by number of constituents while questions and commands have fewer constituents than statements. Language samples were indexed on the basis of the longest utterance (maximum number of clause elements).
Sections B and D of the chart have been excluded since the authors claim that these are not usually used clinically.
The chart includes the original sentence level columns. The other columns are connectivity, phrase, pronoun and word. Pronouns were included due to clinician demand. Within each stage the order of the structures represents the relative frequency of production. By counting the number of structures a quantitative score is achieved.
A Frisian language version of the chart is also provided.
Chapter 12: C-LARSP: Developing a Chinese grammatical profile (Lixian Jin, Bee Lim Oh and Rogayah A. Razak)
There is a current lack of (linguistically and socially) appropriate assessment tools for Chinese. Lack of developmental data and lack of agreement on Chinese grammar are obstacles.
The database comes from Mandarin children speaking Chinese in Malaysia. There were 130 children, aged 1;0-6;, speaking Mandarin as the dominant home language. Data included: free conversation, story-(re)telling + self generated narratives for children over 4 yrs.
A (partial) survey of Mandarin grammar is given. Terms are used from LARSP (clause, phrase and word levels) and an effort is made to use terms common to Chinese speakers. Chinese sentences can be classified as in English (statement, question, command, and exclamation). In addition they can be classified as Subject-Predicate or Non-Subject-Predicate. For Subject-Predicate sentences, if the subject is omitted it can be recovered from the context. For Non-Subject-Predicate, this is not possible. A further distinction is made between single clause sentences and double/multiple clause sentences. Clause and phrase level elements are similar to English. Word-level elements are bound morphemes.
In the C-LARSP the basic structures are the same as LARSP but there is less adherence to the number of structures at each stage.
The chart does not deal with bilingual issues.
Chapter 13: F-LARSP: A computerized tool for measuring morphosyntactic abilities in French (Christophe Parisse, Christelle Maillart and Jodi Tommerdahl)
A description of French morphosyntax is given. French has basic SVO order, and no null subjects except for imperatives (which need an impersonal subject). There is moderate inflection for person, gender, and number; and inflection for nouns, verbs and adjectives. Verbal tense is explicit with different morphology for past events.
The adaptation of the LARSP involved omission of English structures not present in French and identification of French specific target structures (e.g. dislocations). The software was developed based on CLAN (Computerized Language Analysis, MacWhinney, 2000): possible parts of speech analyses are suggested and the best candidate is chosen based on statistical information. Recursively elements are grouped by the program and then the output requires manual correction. The final stage is a print-out summarizing results on a chart. The authors found difficulty with coding later stages.
The database consisted of 36 recordings, each 20 minutes long, of children aged 2;0-4;0. Structures were assigned to a stage when they occurred for 50% of the children at that age. A chart of the analysis is provided. Accuracy checks compared the computer and hand analyses. The software had difficulty with stages from Stage V and on.
Chapter 14: Spanish acquisition and the development of PERSL (Ana Isabel Codesido-García, Carmen Julia Coloma, Elena Garayzăbal-Heinze, Victoria Marrero, Elvira Mendoza and Ma Mercedes Pavez)
This chapter starts with a summary of Spanish grammar. Although SVO is the most common order, various word orders are acceptable. Spanish has a rich inflectional system. There are ‘variable’ elements which change for number, gender, conjugation, time, mood, aspect, voice and form, while ‘invariable’ elements (adverbs, conjunctions, prepositions and interjections) do not change. Gender and number agreement exists between determiners and adjectives and nouns. Possible clause and phrase structures are given. Specific predicted difficulties for children acquiring Spanish include: pronouns, det-N agreement, irregular verbs and complex sentences.
This is a (preliminary) adaptation of the LARSP with expanded word-level analysis and selection of structures appropriate for Spanish.
The database used is the CHILDES Spanish corpus (MacWhinney, 1995). The latest stage was modified based on a further corpus of Chilean Spanish, collected for ten children aged 5-6 years. The most frequent structures used by children in the corpus are listed. Charts of the relative frequency of the various structures are provided. Word level development is only noted in Stage VI.
Chapter 15: LARSP for Turkish (TR-LARSP) (Seyhun Topbaş, Özlem Cangökçe-Yaşar and Martin J. Ball)
The database for Turkish was based on 70 children, aged 0;9-3;6.
An outline of Turkish grammar is given. Turkish is described as an agglutinative language, primarily suffixing. Derivational and inflectional morphology is rich. There is flexible word order. Branching is to the left of the head.
For the TR_LARSP (Turkish LARSP), Sections A-D are maintained. There is a separate word-level chart and LARSP Stage I is divided into 2 levels. In describing clause level structures, scoring takes into account that verbal inflection marks subordination. Word-level derivations and inflections are complex as vowel harmony results in variation in these affixes.
This volume achieves its goals of being both a source book for clinicians working in various languages and introducing new adaptations of LARSP. Although initial chapters relate to the history of LARSP, there is no overview of the original LARSP chart and I felt that in reading it, a familiarity with the original LARSP is assumed.
Considering that this volume contains fifteen chapters by different authors, it retains coherency, largely because of the closeness of the topics covered and preservation of several main areas of discussion: a grammatical survey, database and the adapted chart itself. There is some variety in the formatting and order of presentation of these areas in different chapters and I found it easiest to follow those chapters where the three areas were clearly separated.
The short surveys of the grammars of the various languages presented vary in detail and complexity. Although this is not a primary goal of this volume, they serve to give the reader a brief glimpse of grammatical diversity.
This volume reports the adaptation (as opposed to translation) of LARSP to twelve different languages from different language families. The emphasis on adaptation suggests great differences in language acquisition cross-linguistically. Yet, the adaptability of the LARSP to the various languages, and the ultimate similarity of the various charts in their overall structure, highlight the universal pattern of first language acquisition. Differences are primarily in details and not in the general pattern.
Some of the charts retain ages while other do not. This is largely based on the size of the relevant database. One of the issues that arises is the choice of structures to be included at each age, especially considering that many early structures are grammatical and continue to be used at later stages and even through to adulthood. The criteria chosen were usually related to the percent of children using the structure at a given age/stage or the frequency of occurrence of the structure. These criteria are related to the more general issue in language acquisition as to what constitutes acquisition of a grammatical structure.
Three of the chapters (3, 5 and 13) discuss computerization of the profiling process for English, German and French, respectively. The clinical advantages of computerization in terms of time savings are great, while not detracting from clinician knowledge and involvement in the assessment process.
Clinical examples are included for Dutch (chapter 6) and German (chapter 5), while for some other languages the development and use of the charts is in earlier stages and wide-spread clinical use is not yet a reality.
Steven Long (chapter 2) suggests that LARSP is still relevant thirty years on. The chapters in this volume support that argument, showing how a detailed and organized description of developing grammar can lead to better understanding of normative grammatical acquisition as well as identify atypical development and indicate appropriate treatment goals. Improved clinician training to facilitate knowledge of morphosyntax in the target language, greater knowledge of normative developmental patterns and computerization (programs being freely downloadable) aid in promoting LARSP and its multi-linguistic adaptations as important and efficient clinical tools.
Bol, G.W. and Kuiken, F. (1988) Grammaticale Analyse van Taalonwikkelings stoornissen. Dissertation: University of Amsterdam.
Berman, R.A., Rom, A. and Hirsch, M. (1982) Working with HARSP: Hebrew adaptation of the LARSP Language Assessment Remediation and Screening Procedure. Ms., Tel Aviv University.
MacWhinney, B. (2000) The CHILDES Project: Tools for Analyzing Talk. Third Edition.
Mahwah, NJ: Lawrence Erlbaum Associates.
MacWhinney, B. (1995) The CHILDES Project: Computational Tools for Analyzing Talk (2nd edn.). Hillsdale, NJ: Lawrence Erlbaum.
ABOUT THE REVIEWER