Correcting Semantic Collocation Errors with L1-induced Paraphrases

Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, pages 107–117, Edinburgh, Scotland, UK, July 27–31, 2011. c⃝2011 Association for Computational Linguistics Correcting Semantic Collocation Errors with L1-induced Paraphrases Daniel Dahlmeier1 and Hwee Tou Ng1,2 1NUS Graduate School for Integrative Sciences and Engineering 2Department of Computer Science, National University of Singapore {danielhe,nght}@comp.nus.edu.sg Abstract We present a novel approach for automatic collocation error correction in learner English which is based on paraphrases extracted from parallel corpora. Our key assumption is that collocation errors are often caused by se- mantic similarity in the first language (L1- language) of the writer. An analysis of a large corpus of annotated learner English con- firms this assumption. We evaluate our ap- proach on real-world learner data and show that L1-induced paraphrases outperform tradi- tional approaches based on edit distance, ho- mophones, and WordNet synonyms. 1 Introduction Grammatical error correction (GEC) is emerging as a commercially attractive application of natural lan- guage processing (NLP) for the booming market of English as foreign or second language (EFL/ESL1). The de facto standard approach to GEC is to build a statistical model that can choose the most likely correction from a confusion set of possible correc- tion choices. The way the confusion set is defined depends on the type of error. Work in context- sensitive spelling error correction (Golding and Roth, 1999) has traditionally focused on confusion sets with similar spelling (e.g., {dessert, desert}) or similar pronunciation (e.g., {there, their}). In other words, the words in a confusion set are deemed con- fusable because of orthographic or phonetic simi- larity. Other work in GEC has defined the confu- 1For simplicity, we will collectively refer to both terms as English as a foreign language (EFL) sion sets based on syntactic similarity, for exam- ple all English articles or the most frequent English prepositions form a confusion set (see for example (Tetreault et al., 2010; Rozovskaya and Roth, 2010; Gamon, 2010; Dahlmeier and Ng, 2011) among oth- ers). In contrast, we investigate in this paper a class of grammatical errors where the source of confusion is the similar semantics of the words, rather than or- thography, phonetics, or syntax. In particular, we focus on collocation errors in EFL writing. The term collocation (Firth, 1957) describes a sequence of words that is conventionally used together in a particular way by native speakers and appears more often together than one would expect by chance. The correct use of collocations is a major difficulty for EFL students (Farghal and Obiedat, 1995). In this work, we present a novel approach for au- tomatic correction of collocation errors in EFL writ- ing. Our key observation is that words are poten- tially confusable for an EFL student if they have similar translations in the writer’s first language (L1- language), or in other words if they have the same semantics in the L1-language of the writer. The Chinese word 看 (kàn), for example, has over a dozen translations in English, including the words see, look, …

Content truncated to 500 words