A Comparative Analysis Of Irony Recognition Between English And Chinese

Irony is common in daily communication. As a subjective linguistic phenomenon, irony conveys the implied information beyond the semantic meaning, of which the presentation concerned with the linguistic features remains to be explored. Based on the previous studies on the criterion of irony recognition and implicit display theory by Utsumi(2000), this study attempts to verify whether the six core features can be applied to the discrimination of verbal irony in both Chinese and English by calculating the characteristic values of language materials as to find out the similarities and differences of the pragma-linguistic features in irony identification between English and Chinese in similar contexts. And the six categories of features are n-grams, POS n-grams, funny profiling, positive/negative profiling, affective profiling and pleasantness profiling. Research corpus are collected from website from 2011 to 2015, which contains the positive sets of corpus with Chinese Sina microblog comments labeled with irony or joke and English Amazon product reviews, of which the results are more than four stars. Accordingly, the negative subsets of corpus are collected from Trip Advisor and Chinese website, which contains Chinese Proverbs quotes and English comments. The whole evaluation corpus consists of 210,000 words in more than 40,000 sentences, which were retrieved and measured to make contribution to the comparative analysis. After the annotation and extraction of POS features, data are analyzed in SVM and NB classifier training method.Experimental results show that the relevance feature in verbal irony is quite satisfactory both in English and Chinese, which means that the six categories of features, n-grams, POS n-grams, funny profiling, positive/negative profiling, affective profiling and pleasantness profiling, play an important role in the discrimination of Chinese irony. What is more, feature values for each category show that the arrangement of each feature is identified differently between the Chinese and English in irony recognition. Specifically, N-gram features play a critical role in the identification of English irony in different templates, followed by the POS n-gram and pleasantness profiling features, while on the other hand, polarity and POS n-gram features play the similar role in the discrimination of Chinese irony, in which the polarity feature is more effective. In addition, affective feature has the similar effect in both English and Chinese irony recognition.Findings of this study may contribute to a more accurate processing of Chinese verbal information from the Internet and may also help intermediate and advanced learners of Chinese more effectively grasp the meaning implied under real intentions, thus may provide first-hand materials for the acquisition of pragmatic information in the process of second language learning.