Correcting real-word spelling errors by restoring lexical cohesion

Spelling errors that happen to result in a real word in the lexicon cannot be detected by a conventional spelling checker. We present a method for detecting and correcting many such errors by identifying tokens that are semantically unrelated to their context and are spelling variations of words that would be related to the context. Relatedness to context is determined by a measure of semantic distance initially proposed by Jiang and Conrath (1997). We tested the method on an artificial corpus of errors; it achieved recall of 23 to 50% and precision of 18 to 25%. 1 Real-word spelling errors Conventional spelling checkers detect typing errors simply by comparing each token of a text against a dictionary of words that are known to be correctly spelled. Any token that matches an element of the dictionary, possibly after some minimal morphological analysis, is deemed to be correctly spelled; any token that matches no element is flagged as a possible error, with nearmatches displayed as suggested corrections. Typing errors that happen to result in a token that is a correctly spelled word, albeit not the one that the user intended, cannot be detected by such systems. Such errors are not uncommon; Mitton [1987, 1996] found that “real-word errors account for about a quarter to a third of all spelling errors, perhaps more if you include word-division