How to deal with languages that do not fit our limited compositional view of parsing? I was just reminded of the wonderful existence of Reut Tsarfaty and the beautiful little video that she made together with Noa Tsarfaty for the PhD.
Just watch it, it's very good.
Full reference: Tsarfaty, Reut. 2010 "Relational-Realizational Parsing". PhD Dissertation, University of Amsterdam. ILLC Dissertation Series DS-2010-01 ISBN:978-90-5776-205-5
Links: txt-file of abstract here, entire fulltext here
Extract from abstract: Statistical parsing models aim to assign accurate syntactic analyses to natural language sentences based on the patterns and frequencies observed in human-annotated training data. State-of-the-art statistical parsers to date demonstrate excellent performance in parsing English, but when the same models are applied to languages different than English, they hardly ever obtain comparable results. The grammar of English is quite unusual in that it is fairly configurational. This means that the order of words inside sentences in English is relatively rigid and that the morphology of words is rather impoverished. The main challenge associated with parsing languages that are less configurational than English, such as German, Arabic, Hebrew or Warlpiri, is the need to model and to statistically learn complex correspondence patterns between functions, i.e., sets of abstract grammatical relations, and their morphological and syntactic forms of realization. This thesis proposes a new model, called the Relational-Realizational (RR) model, that can effectively cope with parsing languages that allow for flexible word-order patterns and rich morphological marking.