Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.636603
Title: Untangling word webs : graph theory approaches to L2 lexicons
Author: Wilks, C. F.
Awarding Body: University of Wales Swansea
Current Institution: Swansea University
Date of Award: 1999
Availability of Full Text:
Access from EThOS:
Abstract:
This research sets out to examine some of the implications of metaphor of the "vocabulary network" that is widely used in lexical research. It takes a formal approach to the exploration of this metaphor by applying the principles of Graph Theory to the notion of the lexical network. It looks, in particular, at how graph theoretical principles may be applied to word association data in order to compare the relative densities of L1 and L2 lexical networks. Nine experiments were carried out. An initial series of studies followed up in detail a simple graph theoretical model, proposed by Welsh (1988) and Meara (1992). Data was collected from learners and native speakers of French, English, Spanish and German using a word association chain-building methodology. These experiments showed up significant theoretical and methodological flaws in the original model which could not be readily circumvented by the development of a more sophisticated weighted network model. In view of this, an alternative elicitation technique was devised with the aid of a computerized simulation program. This alternative methodology was employed in a final experiment which showed it to be effective as a tool for the comparison of the densities of vocabulary networks in L1 and L2. The new elicitation technique was also able to provide a means of actually quantifying lexical density levels. It is argued that it will be helpful to develop a more complex interpretation of the notion of lexical density which includes parameters such as the inclusiveness of networks as well as the number of connections they contain. Our formal investigation of the network metaphor also suggests that we need to be more precise about the type of network model that is most appropriate to vocabulary research. It is proposed that L2 word association behaviour may be best predicted by semi-random, "small-world" network models, rather than by the wholly random networks that much of the literature tacitly assumes.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.636603  DOI: Not available
Share: