A Syntax-first Approach to High-quality Morphological Analysis and Lemma Disambiguation for the TüBa-D/Z Treebank

摘要: Morphological analyses and lemma information are an important auxiliary resource for any treebank, especially morphologically rich languages since such is a useful precondition task that needs to link surface forms semantic interpretation (either through wordnets or distributional measures). In contrast common practice in parsing, the method used TuBaD/Z treebank uses syntactic morphological disambiguation. We argue this approach has advantage context of treebanking many ambiguities morphology lemmas can be eliminated given context.