Final Report Summary - EAAS (Early Acquisition of Argument Structure)
This fellowship was devoted to (1) investigation of the early acquisition of morphosyntax of transitive sentences in Estonian, and (2) training the fellow in methods for language acquisition research. This involved a three-pronged approach typical of studies of acquisition, particularly for less studied languages, including (a) analysis of a corpus of child speech, (b) comparison of child and caregiver speech, and (c) testing children’s early linguistic knowledge experimentally.
Summary of project objectives:
Both the Research and Training Objectives of the project have been met.
Research Objectives (RO) of the fellowship were: (RO1) analysis of the acquisition of argument structure in Estonian and crosslinguistic comparison of findings, aiming to identify cognitive and typological factors affecting the various paths to adult grammatical knowledge; (RO2) intralinguistic comparison of the acquisition of various constructions and predicates, and the role played by variability in the input; and (RO3) consideration of contextual cues to acquisition, including semantic, syntactic and pragmatic context.
Training Objectives (TO) of the project were: (TO1) Training in language acquisition theory, cross-cutting all three RO’s, supported especially by monthly seminars on a range of associated topics in the ESRC International Centre for Language and Communicative Development (LuCiD); (TO2) training in corpus data analysis, using CLAN software and R packages for naturalistic data analysis and measures of entropy and productivity developed by members of LuCiD at Manchester and Liverpool, particularly relevant for RO1 and RO3; and (TO3) training in experimental methods and analytical tools used in psycholinguistic approaches to language acquisition, including experimental design, specifically applied to structured elicitation, sentence completion, forced-choice pointing and referential communication paradigms, for testing (a) early grammatical knowledge, and (b) the level of abstraction young children operate with.
Work performed since the beginning of the project:
The dense corpus analysed in the project consists of naturalistic data recordings at two age points. In Year One of the project, transitive utterances were culled from the naturalistic corpus to form four datsets: caregiver and child, at two time points, 2 and 3 years of age. This involved manual coding of 6752 utterances according to potentially relevant predictors. Analysis of the naturalistic corpus data provided the basis for four experiments designed to test a broader cohort of children (aged 3 to 6) than is possible with naturalistic data on three domains of linguistic knowledge which in Estonian involve variation in the input.
Research Assistants were trained to run four experiments in Estonia:
Experiment 1 used structured elicitation to test 3 to 5-year-old children’s knowledge of noun inflections and their ability to generalise patterns to novel nouns. As Estonian noun declension involves multiple frequent and productive declension classes which are not fully predictable from phonology or semantics, this is a particularly challenging aspect of acquisition in Estonian.
Experiments 2 and 3 tested 3 to 5-year-old children’s knowledge of object case-marking in Estonian, which involves variation between three cases and depends on the interaction of lexical, syntactic and semantic factors. Production of object forms was tested with a sentence completion task (exp. 2), and comprehension was tested with a forced choice pointing paradigm (exp. 3).
Finally, Experiment 4 tested 4 to 6-year-old children’s pragmatic abilities with a referential communication task to determine what affects children’s choice of referential form in a short discourse. The corpus analysis revealed exceptionally high proportions of omitted subjects and objects, raising questions about how children make the choice of overt expression (pronoun or lexical noun) or non-expression of arguments, as related to information structure status on one hand, and grammatical role on the other.
Experiments were conducted by Research Assistants in Estonia, overseen by the fellow. Altogether, 223 children aged 3 to 6 were tested, as well as control groups with a total of 49 adults. The fellow has revisited half of the preschools in Tartu, Estonia, to give feedback on the studies and conduct workshops on linguistic development with teachers and interested parents. Results were analysed by the fellow in collaboration with colleagues in Manchester.
Description of the main results achieved so far:
Results demonstrate that 3-year-olds have a basic understanding of the noun declension system (the morphological form of noun cases) in Estonian, and the ability to generalise the paradigms to new nouns, indicating more consolidated and abstract knowledge, develops significantly between 3 and 5 years of age. Both children and adults generalise to novel nouns based on analogy with forms they know rather than on abstract rules.
Genitive forms are easier to generalise than partitive case forms, despite the high frequency of both in the input. Regarding case functions, both the corpus analysis and experimental results demonstrate that children aged 3 first approach the production and comprehension of object case categorically, whereas 5-year-old children are more attuned to the complexity of determining the appropriate form-meaning mapping and show a sensitivity to both lexical and semantic factors, unlike the pattern we see in either 3-year-olds or adults.
Expected final results and their potential impact and use:
Final results speak to theoretical questions regarding linguistic and cognitive development and typology, including: how children cope with variability in the input; the ability of young children to generalise grammatical knowledge to new forms and structures; the developmental trajectory of children learning a rich inflectional system; the illusory dichotomy between regular and irregular morphology; and the differing timescales required for acquisition of semantics, syntax and pragmatics.
On a more localised level, potential for practical application of the findings can be foreseen primarily in Estonia, for supplementing knowledge of typical language development, relevant especially for speech and language therapists and teachers of children with special needs. Results are also of interest to parents with either typically or atypically developing children, wishing to compare their child’s development to norms in Estonia.
A planned further research programme will aim to compare the results for Estonian children with Russian children, who also face a complex inflectional system, and in the longer term to draw a picture of bilingual development with the two languages which have equally challenging morphosyntax but differently structured inflectional systems.
Project website:
In ENGLISH: http://lucid.ac.uk/what-we-do/research/affiliated-projects/virve-anneli-vihman-marie-curie-research-fellow/
In ESTONIAN: http://lucid.ac.uk/what-we-do/research/affiliated-projects/virve-anneli-vihman-marie-curie-teadur/
Summary of project objectives:
Both the Research and Training Objectives of the project have been met.
Research Objectives (RO) of the fellowship were: (RO1) analysis of the acquisition of argument structure in Estonian and crosslinguistic comparison of findings, aiming to identify cognitive and typological factors affecting the various paths to adult grammatical knowledge; (RO2) intralinguistic comparison of the acquisition of various constructions and predicates, and the role played by variability in the input; and (RO3) consideration of contextual cues to acquisition, including semantic, syntactic and pragmatic context.
Training Objectives (TO) of the project were: (TO1) Training in language acquisition theory, cross-cutting all three RO’s, supported especially by monthly seminars on a range of associated topics in the ESRC International Centre for Language and Communicative Development (LuCiD); (TO2) training in corpus data analysis, using CLAN software and R packages for naturalistic data analysis and measures of entropy and productivity developed by members of LuCiD at Manchester and Liverpool, particularly relevant for RO1 and RO3; and (TO3) training in experimental methods and analytical tools used in psycholinguistic approaches to language acquisition, including experimental design, specifically applied to structured elicitation, sentence completion, forced-choice pointing and referential communication paradigms, for testing (a) early grammatical knowledge, and (b) the level of abstraction young children operate with.
Work performed since the beginning of the project:
The dense corpus analysed in the project consists of naturalistic data recordings at two age points. In Year One of the project, transitive utterances were culled from the naturalistic corpus to form four datsets: caregiver and child, at two time points, 2 and 3 years of age. This involved manual coding of 6752 utterances according to potentially relevant predictors. Analysis of the naturalistic corpus data provided the basis for four experiments designed to test a broader cohort of children (aged 3 to 6) than is possible with naturalistic data on three domains of linguistic knowledge which in Estonian involve variation in the input.
Research Assistants were trained to run four experiments in Estonia:
Experiment 1 used structured elicitation to test 3 to 5-year-old children’s knowledge of noun inflections and their ability to generalise patterns to novel nouns. As Estonian noun declension involves multiple frequent and productive declension classes which are not fully predictable from phonology or semantics, this is a particularly challenging aspect of acquisition in Estonian.
Experiments 2 and 3 tested 3 to 5-year-old children’s knowledge of object case-marking in Estonian, which involves variation between three cases and depends on the interaction of lexical, syntactic and semantic factors. Production of object forms was tested with a sentence completion task (exp. 2), and comprehension was tested with a forced choice pointing paradigm (exp. 3).
Finally, Experiment 4 tested 4 to 6-year-old children’s pragmatic abilities with a referential communication task to determine what affects children’s choice of referential form in a short discourse. The corpus analysis revealed exceptionally high proportions of omitted subjects and objects, raising questions about how children make the choice of overt expression (pronoun or lexical noun) or non-expression of arguments, as related to information structure status on one hand, and grammatical role on the other.
Experiments were conducted by Research Assistants in Estonia, overseen by the fellow. Altogether, 223 children aged 3 to 6 were tested, as well as control groups with a total of 49 adults. The fellow has revisited half of the preschools in Tartu, Estonia, to give feedback on the studies and conduct workshops on linguistic development with teachers and interested parents. Results were analysed by the fellow in collaboration with colleagues in Manchester.
Description of the main results achieved so far:
Results demonstrate that 3-year-olds have a basic understanding of the noun declension system (the morphological form of noun cases) in Estonian, and the ability to generalise the paradigms to new nouns, indicating more consolidated and abstract knowledge, develops significantly between 3 and 5 years of age. Both children and adults generalise to novel nouns based on analogy with forms they know rather than on abstract rules.
Genitive forms are easier to generalise than partitive case forms, despite the high frequency of both in the input. Regarding case functions, both the corpus analysis and experimental results demonstrate that children aged 3 first approach the production and comprehension of object case categorically, whereas 5-year-old children are more attuned to the complexity of determining the appropriate form-meaning mapping and show a sensitivity to both lexical and semantic factors, unlike the pattern we see in either 3-year-olds or adults.
Expected final results and their potential impact and use:
Final results speak to theoretical questions regarding linguistic and cognitive development and typology, including: how children cope with variability in the input; the ability of young children to generalise grammatical knowledge to new forms and structures; the developmental trajectory of children learning a rich inflectional system; the illusory dichotomy between regular and irregular morphology; and the differing timescales required for acquisition of semantics, syntax and pragmatics.
On a more localised level, potential for practical application of the findings can be foreseen primarily in Estonia, for supplementing knowledge of typical language development, relevant especially for speech and language therapists and teachers of children with special needs. Results are also of interest to parents with either typically or atypically developing children, wishing to compare their child’s development to norms in Estonia.
A planned further research programme will aim to compare the results for Estonian children with Russian children, who also face a complex inflectional system, and in the longer term to draw a picture of bilingual development with the two languages which have equally challenging morphosyntax but differently structured inflectional systems.
Project website:
In ENGLISH: http://lucid.ac.uk/what-we-do/research/affiliated-projects/virve-anneli-vihman-marie-curie-research-fellow/
In ESTONIAN: http://lucid.ac.uk/what-we-do/research/affiliated-projects/virve-anneli-vihman-marie-curie-teadur/