Locating and Reusing Sundry NLP Flotsam in an e-Learning Application

Abstract

We describe the background and motivation for an e-learning project—IT-based Collaborative Learning in Grammar—where NLP resource reuse has become an important issue. The resources are of several kinds: POS-tagged and syntactically annotated corpora (treebanks), parsing systems and grammar writer’s workbenches, and visulization and manipulation tools for linguistically annotated corpora. Our experience thus far has been that although there are a number of such resources available e.g. on the Web, as a rule, numerous incompatibilities and lack of standardization at all levels—markup formats, linguistic annotation schemes, grammatical framework, software APIs, etc.—make the reuse of these resources into a non-trivial endeavor. 0. Preamble: the Setting It is generally acknowledged that the goal of teaching grammar—especially at the university level—should not primarily be that students memorize definitions of concepts and grammatical constructions, but rather that they understand and learn to recognize different structural patterns. This can hardly be achieved without giving students practical training in the skill of grammatical analysis. Research