Treebank Annotation in the Light of the Meaning-Text Theory

Abstract

A treebank may contain the annotation of different phenomena such as word order, morphological features, syntactic and semantic relations, etc., which are rather different in their nature. Quite often, the annotation of these phenomena is combined in a single structure, which leads to low-quality training results and is verifiably deficient from a theoretical (linguistic) perspective. We argue that the annotation of corpora requires a well-defined linguistic model which supports multi-level annotation, with one type of phenomenon per level. Our experience with dependency treebanks created or adjusted for surface-oriented natural language generation and based on the Meaning-Text Theory, a multi-level linguistic model, supports this argumentation.