This thesis presents a subject matter of improvements of the English Syntactic Analysis System which was developed as a part of MAchine Translation Environment for English to Korean(MATE-EK). The improvements can be seen mainly in two respects, one for parsing scope and the other for processing efficiency.
Former system has a critical weak point of incapability of parsing complex phrases and the complex sentence depth of more than 2 which occurs frequently in written English. It can be resolved by the employment of revised algorithm that reflects structural recursiveness. Regularizing the syntactic structures make subsequent processing concern itself with a much smaller number of structures and hence simplified. Also, applying some rejection rules positively results in reduced trials of meaningless parsing by eliminating some trees as early as possible.
The newly designed system was implemented using GWL, a grammar writing language based on tree-to-tree transformation. GWL was found to be easy to represent, modify and extend linguistic grammar, which is essential property of a grammar writing language.
