Ambiguity Resolution by Reordering Rules in Text Containing Errors

Writing aids such as spelling and grammar checkers are often based on texts by adult writers and are not sufficiently targeted to support children in their writing process. This paper reports on the development of a writing tool based on a corpus of Swedish text written by children and on the parsing methods developed to handle text containing errors. The system uses finite state techniques for finding grammar errors without actually specifying the error. The ‘broadness’ of the grammar and the lexical ambiguity in words, necessary for parsing text containing errors, also yields ambiguous and/or alternative phrase annotations. We block some of the (erroneous) alternative parses by the order in which phrase segments are selected, which causes bleeding of some rules and more ‘correct’ parsing results are achieved. The technique shows good coverage results for agreement and verb selection phenomena.