Analysis of English Spelling Errors in a Word-Typing Game

Analysis of English Spelling Errors in a Word-Typing Game Ryuichi Tachibana, Mamoru Komachi Tokyo Metropolitan University 6-6 Asahigaoka Hino-city, Tokyo, Japan tachibana-ryuichi@ed.tmu.ac.jp, komachi@tmu.ac.jp Abstract The emergence of the web has necessitated the need to detect and correct noisy consumer-generated texts. Most of the previous studies on English spelling-error extraction collected English spelling errors from web services such as Twitter by using the edit distance or from input logs utilizing crowdsourcing. However, in the former approach, it is not clear which word corresponds to the spelling error, and the latter approach requires an annotation cost for the crowdsourcing. One notable exception is Rodrigues and Rytting (2012), who proposed to extract English spelling errors by using a word-typing game. Their approach saves the cost of crowdsourcing, and guarantees an exact alignment between the word and the spelling error. However, they did not assert whether the extracted spelling error corpora reflect the usual writing process such as writing a document. Therefore, we propose a new correctable word-typing game that is more similar to the actual writing process. Experimental results showed that we can regard typing-game logs as a source of spelling errors. Keywords: Gamification, Spelling Correction, Information Extraction 1. Introduction People mainly operate computers through a keyboard inter- face. However, they often make spelling errors when using a keyboard; e.g., hitting an adjacent key, or using phonolog- ically similar letters. In our study, we define spelling errors as (1) a typo from an incorrect keyboard operation, and (2) a spelling confusion from an incorrect identification. Baba and Suzuki (2012) introduced two types of spelling errors: (1) spelling errors that people do not notice while typing (hereafter called uncorrected spelling errors), and (2) spelling errors that people correct while typing (here- after called corrected spelling errors). The former can be a great source of online spelling corrections, while the latter have only been studied recently. Detection and correction of spelling errors have been studied over the last few decades. In particular, the emergence of the web has created a large amount of consumer-generated texts, which include countless num- bers of spelling errors. To overcome the problem of noisy texts, researchers have extracted spelling errors from web chatter by crawling through Twitter data (Aramaki et al., 2010) and from input logs by utilizing Amazon’s Mechan- ical Turk (Baba and Suzuki, 2012). However, the former study does not guarantee the correctness of the extracted spelling errors, and the latter study includes the cost of crowdsourcing. On the other hand, studies using gamification have become popular in the NLP literature (Deterding et al., 2011; Ku- maran et al., 2014; Vannella et al., 2014; Rodrigues and Rytting, 2012). Gamification allows us to obtain resources without paying users by carefully designing a game suitable for information extraction. In particular, Rodrigues and Rytting (2012) proposed to ex- tract English spelling errors by using a word-typing game. This game includes the intended word and does not require the cost of crowdsourcing. However, the writing process of a word-typing game may differ …

Content truncated to 500 words