Rätt och fel i korpusen

Abstract

Today, language corpora are necessary requisites for the compiling of dictionaries. But to what extent should they have an impact on the lexical description? Opinions differ between lexicographers who use the corpus more or less as an inspiration for lexical description and those who rely heavily on corpus data (e.g. the wording of examples in Collins COBUILD). To what extent is a corpus reliable as a source for lexical data? This question is discussed against the background of a new Dutch spelling guide based on a 50 million word corpus.