The Establishment Of A System For Language Disorder Assessment:Research Based On Mandarin-Speaking Children's Corpus

Research on the assessment of children with language disorders had become a key area of international children's language development research.In recent years,the international children's corpus system CHILDES had begun to use corpus analysis,a non-standardized approach to construct a language disorder assessment system with norm-referenced database;however,mandarin-speaking children's corpus still remained at the sharing level.It had become a very necessary research task to create a language disorder assessment system that meet the characteristics of chinese language and culture and had a norm-referenced database.This study mainly used corpus research methods to analyze 341 3-6-year-old normal developing children's conversational samples and the conclusions were as below.Firstly,through the analysis of vocabulary and grammar development of normal developing children,this study constructed a corpus-based language disorder assessment system,which is mainly composed of vocabulary assessment system and grammar assessment system.The vocabulary assessment system includes first-level indicators: voc D and secondary indicators: the type and token of nouns,verbs,quantifiers,prepositions,adverbs,and conjunctions.The grammar assessment system includes first-level indicators: MLGU,MLGU5,and grammatical structures quantity values and secondary indicators: quantitative structures,object phrases,object clauses,and serial verb constructions.When applying the above assessment system,we need first focus on the primary indicators: voc D,MLGU,MLGU5,and grammatical structure quantity values.By comparing with the reference data of the above indicators,the basic problem of children's language development could be predicted and found.Second,the language disorder assessment system could further discover problems in vocabulary and grammar development.When a child was significantly behind the reference data in voc D,the first-level indicator of vocabulary assessment system,we could refer to the secondary indicators of the vocabulary assessment system,the type and token of nouns,verbs,quantifiers,prepositions,adverbs,and conjunctions to further judge the status and existing problems of children's vocabulary development;when a child was significantly behind the reference data in MLGU,MLGU5 and grammatical structure quantity values,the first-level indicator of grammar assessment system,we could refer to the secondary indicators of the grammar assessment system : quantitative structure,object phrases,object clauses and serial verb construcions to further judge the status and problems of children's grammar development.Third,the corpus-based language assessment diagnosis system established by this study was used to assess four kinds of children with language disorder(three mentally handicapped children,four hearing impaired children,four autistic children,and four children with specific language impairment(SLI)).Analysis on those cases' longitudinal corpus showed those children all showed an "abnormal state" of language development in different aspects,but the language problems of different types of children were different.For mentally handicapped children,mildly mentally handicapped children were lagging behind in the development of adverbs,quantifiers,prepositions,and object clauses,while moderately mentally handicapped children were lagging behind in the development of conjunctions,prepositions,object phrases,object clauses,and serial verb constructions.For children with hearing impairment,when MLGU was equal to the nreference database,they did not show vocabulary and grammar development problems;other hearing-impaired children generally had insufficient quantifiers production,and insufficient output of object phrases.For children with autism,children with high-functioning autism had limited production of conjunctions,psychological verbs,and insufficient sentence length.Low-functioning autistic children had low vocabulary diversity,limited production of conjunctions and psychological verbs,insufficient sentence length,and insufficient production of grammatical structures.For SLI children,when MLGU lagged behind normal children's reference data,they would have problems with insufficient sentence length and low vocabulary diversity in 4 years old.The assessment and discovery of the above four types of children with language disorder provided basic data and directions for educational intervention.The innovations of this research were,that we constructed a language disorder assessment system based on computer,and confirmed that the non-standardized method,the corpus analysis could be used for assessing mandarin-speaking children with language disorders.Moreover,during the construction process of the assessment system,we built up two new indicators,MLGU and grammatical structure quantity values,which could specifically be used for mandarin-speaking children's language sample analysis based on CHILDES system.Finally,based on the Java program,the efficiency of word segmentation was greatly improved and based on machine learning,the grammatical tier %gra used for automatically grammatical structure analyzing was constructed,which helped us get the characteristics of children's grammar development and enriched our understanding of children's production of grammatical constructions based on computer.