Adapting NLP and Corpus Analysis Techniques to Structured Imagery Analysis in Classical Chinese Poetry
Alex Chengyu Fang, Fengju Lo, Cheuk Kit Chinn · 2009
This paper describes some pioneering work as a joint research project between City University of Hong Kong and Yuan Ze University in Taiwan to adapt language resources and technologies in order to set up a computational framework for the study of the creative language employed in classical Chinese poetry. In particular, it will first of all describe an existing ontology of imageries found in poems written during the Tang and the Song dynasties (7th –14th century AD). It will then propose the augmentation of such imageries into primary, complex, extended and textual imageries. A rationale of such a structured approach is that while poets may use a common dichotomy of primary imageries, creative language use is to be found in the creation of complex and compound imageries. This approach will not only support analysis of inter-poets stylistic similarities and differences but will also effectively reveal intra-poet stylistic characteristics. This article will then describe a syntactic parser designed to produce parse trees that will eventually enable the automatic identification of possible imageries and their subsequent structural analysis and classification. Finally, a case study will be presented that investigated the syntactic properties found in two lyrics written by two stylistically different lyric writers in the Song Dynasty.