Building a Spoonerism Detection System for Vietnamese

Building a Spoonerism Detection System for Vietnamese Thai-Hoang Pham Alt Inc Hanoi, Vietnam pham.thai.hoang@alt.ai Xuan-Khoai Pham City University of Hong Kong Kowloon, Hong Kong xkpham2-c@my.cityu.edu.hk Abstract This paper presents the first automatic sys- tem for Vietnamese spoonerism detection. By incorporating hand-crafted rules with Viet- namese language model, our system, although simple, achieves a promising result for this task. The proposed method achieves an over- all F1 score of 95.47% on the Vietnamese spoonerism dataset. 1 Introduction Spoonerism is a linguistic phenomenon in which parts of two spoken words including consonants, vowels, and tones can be switched to construct two other implied words. A term spoonerism is named after William Archibald Spooner, who was famous for making spoonerism. Spoonerism can be seen as one kind of wordplay used to entertain or to criti- cize. People often use spoonerism in both of written form such as poetry and in oral form such as music and folk tale. To make spoonerism, they use several rules switching parts of one word to the other word to form words with implied meaning. Figure 1 shows a well-known example of spoonerism in English. Figure 1: Spoonerism in English Spoonerism in Vietnamese has a long history. It is found in Vietnamese poetry from the 17th cen- tury and in traditional folk tale and song. Speaker of- ten uses spoonerism in mocking or sarcasm contexts where directness is avoided. They also use spooner- ism as a linguistic challenge in riddles and paral- lel sentences. Vietnamese, because of its linguistic characteristics, is very suitable for making spooner- ism. People only need to switch parts of surface words to form meaningful Vietnamese words. For this reason, spoonerism is used widely in practice by Vietnamese people. Spoonerism detection is an important step used in some natural language processing applications such as question answering and sentiment analysis. It pro- vides useful information that helps computers under- stand human conversation more deeply. However, to best of our knowledge, there is no research on build- ing an automatic system for spoonerism detection. In this paper, we present the first spoonerism de- tection system for Vietnamese language that uses hand-crafted rules associated with Vietnamese lan- guage model. Our system, although simple, achieves promising results for this task. In particular, our system obtains an F1 score of 95.47% on Viet- namese spoonerism dataset. Moreover, our approach does not require the training data when collect- ing data for this task is difficult. We also publi- cize our Vietnamese spoonerism detection system and Vietnamese spoonerism dataset for research pur- pose, which is believed to positively contributing to the long-term advancement of Vietnamese language processing1. 1https://github.com/khoaipx/Vietnamese-Spoonerism PACLIC 32 549 32nd Pacific Asia Conference on Language, Information and Computation Hong Kong, 1-3 December 2018 Copyright 2018 by the authors Input: Vietnamese sentence Output: Type 1 and Type 2 spoonerisms Step 1: Extract syllable pair set from input sentence; Step 2: for each syllable pairs do Decompose syllables pair; Generate reverse forms by using spoonerism rules; for each reverse form do …

Content truncated to 500 words