Using Discourse Information for Education with a Spanish-Chinese Parallel Corpus

Using Discourse Information for Education with a Spanish-Chinese Parallel Corpus Shuyuan Cao ▲＊, Harritxu Gete ＊ ▲Universitat Pompeu Fabra (UPF) ＊University of Basque Country (UPV/EHU) shuyuan.cao@hotmail, harritxu.gete@gmail.com Abstract Nowadays, with the fruitful achievements in Natural Language Processing (NLP) studies, the concern of using NLP technologies for education has called much attention. As two of the most spoken languages in the world, Spanish and Chinese occupy important positions in both NLP studies and bilingual education. In this paper, we present a Spanish-Chinese parallel corpus with annotated discourse information that aims to serve for bilingual language education. The theoretical framework of this work is Rhetorical Structure Theory (RST). The corpus is composed of 100 Spanish-Chinese parallel texts, and all the discourse markers (DM) have been annotated to form the education source. With pedagogical aim, we also present two programs that generate automatic exercises for both Spanish and Chinese students using our corpus. The reliability of this work has been evaluated using Kappa coefficient. Keywords: discourse analysis, education, corpus, bilingual language learning 1. Introduction Using natural language processing (NLP) for educational applications starts from the early history (Burstein, 2009). Different NLP studies make a great advance in different educational areas, for instance, translation studies, text retrieval, text mining or speech recognition. Among different NLP studies, the emphasis on the idea that discourse information may be useful for Natural Language Processing (NLP) has become increasingly popular. Discourse analysis is an unsolved problem in this field, although discourse information is crucial for many NLP tasks (Zhou et al., 2014). As two of the most spoken languages in the world, Spanish and Chinese occupy important positions in NLP development. Due to the great linguistic distance that between this pair of languages, the number of differences in their discourse structure is also considerable. The following examples show some of the discourse differences between Spanish and Chinese. Ex.1: 1.1 Sp: Aunque aún no contamos con resultados, intuimos que el modelo será más amplio que el del sintagma nominal. [Aunque aún no contamos con resultados,]Unit1 [intuimos que el modelo será más amplio que el del sintagma nominal.]Unit2 [DM1 still no get still no get results,]Unit1 [we consider that the model will more extensive than the sentence group nominal2.]Unit2 1.2 Sp: Intuimos que el modelo será más amplio que el del sintagma nominal, aunque aún no contamos con resultados. [Intuimos que el modelo será más amplio que el del 1 DM means discourse marker. We will give the specific definition of discourse marker in the methodology section. 2 In this work, for all the presented examples, we will give an English literature translation for each example. sintagma nominal,]Unit1 [aunque aún no contamos con resultados.]Unit2 [We consider that the model will more extensive than the sentence group nominal.]Unit1 [DM still no get still no get results]Unit2 1.3 Ch: 尽管还没有取得最终结果，但是我们认为该模 型已囊括了语段模型涉及的内容。 [尽管还没有取得最终结果，]Unit1 [但是我们认为该模 型已囊括了语段模型涉及的内容。]Unit2 [DM1 still no get results,]Unit1 [DM2 we consider that the model contains the sentence group nominal.]Unit2 1.4 Eng: Although we haven’t got the results yet, we consider that …

Content truncated to 500 words