A Dependency Treebank of Classical Chinese Poems

2012 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 191–199, Montr´eal, Canada, June 3-8, 2012. c⃝2012 Association for Computational Linguistics A Dependency Treebank of Classical Chinese Poems John Lee and Yin Hei Kong The Halliday Centre for Intelligent Applications of Language Studies Department of Chinese, Translation and Linguistics City University of Hong Kong {jsylee,yhkong}@cityu.edu.hk Abstract As interest grows in the use of linguistically annotated corpora in research and teaching of foreign languages and literature, treebanks of various historical texts have been developed. We introduce the first large-scale dependency treebank for Classical Chinese literature. De- rived from the Stanford dependency types, it consists of over 32K characters drawn from a collection of poems written in the 8th century CE. We report on the design of new depend- ency relations, discuss aspects of the annota- tion process and evaluation, and illustrate its use in a study of parallelism in Classical Chi- nese poetry. 1 Introduction Recent efforts in creating linguistically annotated text corpora have overwhelmingly focused on modern languages. Among the earliest and most well-known are the part-of-speech (POS) tagged Brown Corpus (Francis & Kučera, 1982), and the syntactically analyzed Penn Treebank (Marcus et al., 1993). However, the first digital corpus, which emerged soon after the invention of computers, had as its subject matter a collection of 13th-century texts --- in 1949, Roberto Busa initiated the POS tagging of the complete works of Thomas Aquinas, written in Latin. In the past decade, Humanities scholars have begun to use digital corpora for the study of an- cient languages and historical texts. They come in a variety of languages and genres, from Old Eng- lish (Taylor et al., 2003) to Early New High Ger- man (Demske et al., 2004) and Medieval Portu- guese (Rocio et al. 2000); and from poetry (Pintzuk & Leendert, 2001) to religious texts such as the New Testament (Haug & Jøhndal, 2008) and the Quran (Dukes & Buckwalter, 2010). They are increasingly being leveraged in teaching (Crane et al., 2009) and in research (Lancaster, 2010). This paper describes the first large-scale de- pendency treebank for Classical Chinese. The treebank consists of poems from the Tang Dynasty (618 – 907 CE), considered one of the crowning achievements in traditional Chinese literature. The first half of the paper reviews related work (section 2), then describes the design of the treebank (sec- tion 3), its text and evaluation (section 4). The second half shows the research potentials of this treebank with a study on parallelism in (section 5). 2 Previous Work Existing linguistic resources for Chinese is pre- dominantly for the modern language. This section first describes the major Modern Chinese treebanks on which we based our work (section 2.1), then summarizes previous research in word segmenta- tion and POS tagging, two pre-requisites for build- ing a Classical Chinese treebank (section 2.2). 2.1 Modern Chinese Most treebanks have been annotated under one of two grammatical theories, the phrase structure grammar, which is adopted by the …

Content truncated to 500 words