Tagging the Dutch PAROLE Corpus
Jesse de Does, John van der Voort van der Kleij · 2002
We discuss the annotation with part of speech and lemma of the Dutch PAROLE Internet Corpus. The PAROLE PoS tagger is a combination of statistical taggers. It includes the Markov tagger TnT and 3 taggers developed at the INL with the purpose of using other information besides the training data. Lemma is assigned by a deterministic procedure, based on an extensive lexicon. The output is in some respects not entirely satisfactory; we discuss what can be done about this without having to manually correct the complete corpus.