A Topic-aware Comparable Corpus of Chinese Variations

TL;DR

This paper constructs a regularly updated, topic-aware comparable corpus of Mainland Chinese and Taiwanese Mandarin from social media sources, facilitating cross-strait linguistic analysis.

Contribution

It introduces a novel, topic-aware corpus of Chinese dialects from social media, reflecting contemporary language use and enabling comparative linguistic studies.

Findings

Corpus is regularly updated and reflects current social media language.

Provides a resource for cross-strait linguistic comparison.

Enhances understanding of language variation in modern Chinese.

Abstract

This study aims to fill the gap by constructing a topic-aware comparable corpus of Mainland Chinese Mandarin and Taiwanese Mandarin from the social media in Mainland China and Taiwan, respectively. Using Dcard for Taiwanese Mandarin and Sina Weibo for Mainland Chinese, we create a comparable corpus that updates regularly and reflects modern language use on social media.