VUPMC: A New Political Metaphor Corpus in Mandarin Chinese

Abstract

This article proposes the Conventional and Novel Metaphor Identification Procedure (CNMIP) for Mandarin Chinese and applies this replicable protocol to annotate the VUPMC dataset, a new Political Metaphor Corpus developed at VU University Amsterdam. The VUPMC corpus contains three Chinese political genres (Policy Documents, Remarks, News Reports) and includes over 220,000 tokens of concordance sentences for the node word 贸易 ‘trade’. The corpus analysis shows that 6.64% of lexical units in the VUPMC dataset are used as metaphor-related words (MRWs) to frame trade (e.g., using ‘war’ to frame trade as a war). Further tests show that distributions of MRWs differ significantly across genres and Parts of Speech. Similarities in MRW distributions between the VUPMC and other datasets confirm the reliability of the CNMIP procedure. The differences, however, highlight the methodological advances in manual annotation of conventional and novel MRWs as well as the distinctive features of Chinese political genres. The VUPMC dataset serves as a valuable language resource for computational detection of Chinese conventional and novel metaphors.

Resources

Details