Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.536526
Title: A quantitative approach to linguistic model validation
Author: Pang, Lee Yick
Awarding Body: Institute of Education, University of London
Current Institution: UCL Institute of Education (IOE)
Date of Award: 1990
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The thesis is an attempt to identify a method of statistical analysis whereby theoretical linguistic models can be validated, to some degree, via analysis of language user perception of text structure. Such a tool of validation is indispensible but has yet to be identified. There are two areas of linguistic model validation where the proposed method of analysis can make a substantial contribution: a - in validating linguistic models, qua descriptive models, as explanatory models, and b - in establishing grounds for comparison among competing and/or conflicting linguistic models in the same area of linguistic investigation. The study has a clear methodological emphasis and explores new empirical procedures of text analysis. The statistical technique for such a validation study is repertory grid analysis (Kelly 1955, Slater 1977). This technique is widely used in psychotherapy but is used for the first time in linguistic investigation. Repertory grid analysis offers two very important contributions. It is, on the one hand, one of the most rigorous quantitative methods for the study of human perception; at the same time, it allows for qualitative analysis of the data, which is very desirable in the study proposed. The area of linguistics to be studied is the signalling approach to text analysis proposed by Winter (1977, 1982) and Hoey (1979, 1983). An informal pre-pilot was first carried out to examine broad features and potential problems of the application of repertory grid analysis to the investigation planned. A proper pilot was then carried out to investigate closely the feasibility of the study. Results from the pilot indicated that the proposed approach was usable. The main study was then performed on a representative sample of a target population (i.e. a sub-population of undergraduate students in Hong Kong). Besides analyses associated with the repertory grid technique, an ANOVA design was used for the investigation of aspects within the experimental situation that may be of relevance. The independent variables include relative English language proficiency and the major academic disciplines of the experimental subjects, different methods of grid elicitation, and variation in text structure. The data were analysed first on individual perception of text structure and then on the agreement between the theoretical model and subject perception both as individuals and as a group. In the analyses, both a quantitative and a qualitative approach were used. The results of the study indicated very clearly that repertory grid analysis was able to make interesting and informative comparisons between the theoretical model and subject perception of text structure and should be a usable technique for linguistic model validation as first hypothesized. In particular, individual characteristics of perception were uncovered; and the consensus view of the sample was captured. Furthermore, the present application of repertory grid analysis also enabled a qualitative analysis of the data which threw additional light on and provided much needed details for the research. The study has important implications for linguistics. Firstly, an objective and statistically based technique for rendering linguistic models susceptible to validation procedure, so far unavailable, has now been identified. Furthermore, the study certainly helps to establish applied linguistics as an academic discipline at once independent from and contributing to theoretical linguistics.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.536526  DOI: Not available
Share: