Modeling ESL Word Choice Similarities By Representing Word Intensions and Extensions

Proceedings of COLING 2012: Technical Papers, pages 2915–2930, COLING 2012, Mumbai, December 2012. Modeling ESL Word Choice Similarities By Representing Word Intensions and Extensions Huichao Xue,Rebecca Hwa Department of Computer Science, University of Pittsburgh, 210 S Bouquet St, Pittsburgh, PA 15260 USA hux10@cs.pitt.edu, hwa@cs.pitt.edu ABSTRACT Automatic error correction systems for English as a Second Language(ESL) speakers often rely on the use of a confusion set to limit the choices of possible correction candidates. Typically, the confusion sets are either manually constructed or extracted from a corpus of manually corrected ESL writings. Both options require the involvement of English teachers. This paper proposes a method to automatically construct confusion sets for commonly used prepositions from non-ESL corpus without manual intervention. The proposed method simulates how ESL learners learn both the intensions and extensions of English words from standard English text. Our experimental results suggest that the automatically constructed confusion sets based on the similarities between the learned words’ intensions is competitive with those directly learned from an ESL corpus containing about 150K preposition usages. TITLE AND ABSTRACT IN ANOTHER LANGUAGE, L2 (OPTIONAL, AND ON SAME PAGE) 通 通 通过 过 过分 分 分析 析 析单 单 单词 词 词的 的 的内 内 内涵 涵 涵和 和 和外 外 外延 延 延来 来 来对 对 对用 用 用词 词 词混 混 混淆 淆 淆建 建 建模 模 模 针对把英语作为第二语言的人群的自动语法纠错系统，通常会需要使用“混淆集” 来限制系统纠错的种类。一般来说，这些混淆集或者是由专家总结经验得出，或者是从被 专家纠错过的文字当中提取的。这两种方法都需要英语专家的介入。在这篇论文当中，我 们提出了一种无需专家介入，自动建立常用介词混淆集的方法。在此方法中，我们对英语 单词的内涵和外延建模，并且模拟了英语学习者们学习单词内涵和外延的过程。实验表 明，使用单词内涵之间的相似度来创建的混淆集，与从含15万介词的标注语料当中提取的 混淆集质量是相当的。 KEYWORDS: Confusion Sets, Lexical Semantics, Grammatical Error Correction, Distance Metric Learning, English as Second Language, Second Language Acquisition. KEYWORDS IN L2: 选词混淆、混淆集、词汇语义学、语法纠错、距离度量学习、英语作 为第二语言、第二语言学习. 2915 1 Introduction A large portion of the English text (e.g., on the web) is written by people whose native language is not English. Many English as a Second Language (ESL) writers, even those with a high level of proficiency, make common grammatical mistakes. Researchers working on Grammar Error Correction (GEC) try to analyze the patterns of these mistakes in order to understand the underlying reasons for their occurrence and to build tools that help ESL writers to correct their errors (Leacock et al., 2010). Many recently developed GEC systems (Chodorow et al., 2007; J. R. Tetreault & Chodorow, 2008; Gamon et al., 2009; Liu et al., 2010; Rozovskaya & Roth, 2011; Dahlmeier & Ng, 2011a) share a similar infrastructure: first, they isolate some specific types of errors (e.g., preposition errors, article errors, or word choice errors); then, they propose a correction for each instance by treating it as a classification problem. To cast the correction problem as a classification problem, the system has to know, a priori, what are the set of possible corrections for an error. That is, the system needs to pre-define a confusion set for each error type. Previous work has shown the importance of the role of confusion sets. However, the construction of confusion sets requires a great deal of human involvement. English teachers are involved in Liu et al. (2010) to manually filter the initial large verb confusion sets; Rozovskaya & Roth (2010a) used annotated ESL corpus to …

Content truncated to 500 words