Un corpus de bigramas utilizado como corrector ortográfico y gramatical destinado a hablantes nativos de Español

Abstract

This paper describes the performance of an algorithm of spelling and grammar checker for texts written in Spanish by proficient native speakers during proof-reading. Possible mistakes are not detected by tagging and parsing but by statistical analysis, comparing combinations of two words used in the text to a hundred-million-word corpus. Those pairs of words (bigrams) which do not occur or which are suspiciously infrequent in the corpus are highlighted; and such pairs often contain errors. The main limitation is that mistakes that arise from non-adjacent words are not detected. Nevertheless, as we have seen in this study with some different texts, the algorithm detects many errors that other grammar checkers are not able to identify.