University of Birmingham

Graduate Student, School of Psychology

Psychology

Thesis Title: Computational Modelling of Syllable Structure within the Mental Lexicon

About

I am exploring the place of syllable structure during speech production. Most current speech production models consider syllabification to be post-lexical. However, a substantial amount of the literature suggests that syllable structure may be present within the mental lexicon. Since the syllable is a predictable unit of speech, linguists and psychologists alike tend to assign syllables to the articulatory stages of speech production rather than as central representations.

I am implementing computational models based in patient data to investigate whether lexically represented syllabic structures could better account for the data collected from speech errors and chronometric experiments.

A small part of my research involves applying my skills in investigating languages that have not been studied in terms of psycholinguistics and computational linguistics. I am primarily interested in the Indian languages families and believe that data from such sources will prove valuable in arriving at a universal model for speech production as well as in isolating language specific mechanisms that may be present in human speech production. As an aside, one of the algorithms I wrote for reading and recording the frequency of phonological units can be adapted to analyse transcripts of almost all Indian languages (except Urdu) which will be valuable in creating psycholinguistic experiments in the future.

The initial stages of my research was primarily concerned with writing syllabification algorithms to quantify the amount of resyllabification that occurs in English speech as well as applying concept of information theory to phonological units. I used entropy calculations to quantify the syllable-related units in various speech production models as well as a new hypothesis to illustrate the comparison between their information content. The results show that there are substantial savings in terms of computational cost versus storage with regard to syllabic representation in the lexicon.

I am currently pursuing experiments in English and Hindi to determine if phonological patients from these languages show the same preservation of syllabic-structural information as previously studied Italian patients. I have recently completed two projects in Delhi with Hindi Aphasic patients funded by the Universitas 21 program and intend to extend the project further in the future. I am also conducting experiments in the UK with funding from the Experimental Psychological Society.

Contact Information