Correcting Grammatical Verb Errors

Proceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics, pages 358–367, Gothenburg, Sweden, April 26-30 2014. c⃝2014 Association for Computational Linguistics Correcting Grammatical Verb Errors Alla Rozovskaya Columbia University New York, NY 10115 ar3366@columbia.edu Dan Roth University of Illinois Urbana, IL 61801 danr@illinois.edu Vivek Srikumar Stanford University Stanford, CA 94305 svivek@cs.stanford.edu Abstract Verb errors are some of the most com- mon mistakes made by non-native writers of English but some of the least studied. The reason is that dealing with verb er- rors requires a new paradigm; essentially all research done on correcting grammat- ical errors assumes a closed set of trig- gers – e.g., correcting the use of prepo- sitions or articles – but identifying mis- takes in verbs necessitates identifying po- tentially ambiguous triggers first, and then determining the type of mistake made and correcting it. Moreover, once the verb is identified, modeling verb errors is chal- lenging because verbs fulfill many gram- matical functions, resulting in a variety of mistakes. Consequently, the little earlier work done on verb errors assumed that the error type is known in advance. We propose a linguistically-motivated ap- proach to verb error correction that makes use of the notion of verb finiteness to iden- tify triggers and types of mistakes, before using a statistical machine learning ap- proach to correct these mistakes. We show that the linguistically-informed model sig- nificantly improves the accuracy of the verb correction approach. 1 Introduction We address the problem of correcting grammati- cal verb mistakes made by English as a Second Language (ESL) learners. Recent work in ESL er- ror correction has focused on errors in article and preposition usage (Han et al., 2006; Felice and Pulman, 2008; Gamon et al., 2008; Tetreault et al., 2010; Gamon, 2010; Rozovskaya and Roth, 2010b; Dahlmeier and Ng, 2011). While verb errors occur as often as article and preposition mistakes, with a few exceptions (Lee and Seneff, 2008; Gamon et al., 2009; Tajiri et al., 2012), there has been little work on verbs. There are two reasons for why it is difficult to deal with verb mistakes. First, in contrast to articles and prepositions, verbs are more difficult to identify in text, as they can often be confused with other parts of speech, and processing tools are known to make more errors on noisy ESL data (Nagata et al., 2011). Second, verbs are more complex linguisti- cally: they fulfill several grammatical functions, and these different roles imply different types of errors. These difficulties have led all previous work on verb mistakes to assume prior knowledge of the mistake type; however, identifying the specific category of a verb error is nontrivial, since the sur- face form of the verb may be ambiguous, espe- cially when that verb is used incorrectly. Consider the following examples of verb mistakes: 1. “We discusses*/discuss this every time.” 2. “I will be lucky if I {will find}*/find something that fits.” 3. “They wanted to visit many places without spend*/spending a …

Content truncated to 500 words