Peter Chew's doctoral dissertation

---

Title: A Computational Phonology of Russian

Objective:

The aim of this research project, which is scheduled for completion by 30 September 1999, is to build a computational model of Russian phonology.

This will be done by stating the rules governing the Russian sound system in the form of a context-free grammar implemented as a Prolog computer program. Prolog is a declarative programming language which has been extensively used in natural language processing. (For some introductory information on Prolog, see http://burks.bton.ac.uk/burks/language/prolog/amzi/advtop.htm.)

Significance of the project:

Computer programming offers the opportunity to test phonological theories in a rigorous and exhaustive manner. When the program has been written, it will be tested on a large corpus of Russian words. The results of this testing may lead to one or both of the following interim conclusions:

• that the theory (as stated in the program) should be modified;
• that items which do not behave as expected should be listed separately as lexical exceptions.

Clearly the rigour and truth of the grammar, as measured by its performance when real Russian words are introduced to it, will improve as the process of trial and error indicates where the grammar should be modified. It is envisaged that new insights will be gained into Russian phonology (particularly areas which have traditionally been regarded as controversial) as the end result should be a maximally true and simple statement of the rules.

Russian is a language with certain characteristics of particular interest to the phonologist. These include the following:

• apparently unpredictable and mobile stress;
• impact of consonant clusters on syllabification;
• the "fleeting vowel" which appears in many paradigms.

For an outline of my dissertation, click here.

Return to Peter Chew's Home Page

Contact Peter Chew