Morphology and Reranking for the Statistical Parsing of Spanish

Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing (HLT/EMNLP), pages 795–802, Vancouver, October 2005. c⃝2005 Association for Computational Linguistics Morphology and Reranking for the Statistical Parsing of Spanish Brooke Cowan MIT CSAIL brooke@csail.mit.edu Michael Collins MIT CSAIL mcollins@csail.mit.edu Abstract We present two methods for incorporat- ing detailed features in a Spanish parser, building on a baseline model that is a lex- icalized PCFG. The first method exploits Spanish morphology, and achieves an F1 constituency score of 83.6%. This is an improvement over 81.2% accuracy for the baseline, which makes little or no use of morphological information. The second model uses a reranking approach to add arbitrary global features of parse trees to the morphological model. The reranking model reaches 85.1% F1 accuracy on the Spanish parsing task. The resulting model for Spanish parsing combines an approach that specifically targets morphological in- formation with an approach that makes use of general structural features. 1 Introduction Initial methods for statistical parsing were mainly developed through experimentation on English data sets. Subsequent research has focused on apply- ing these methods to other languages. There has been widespread evidence that new languages ex- hibit linguistic phenomena that pose considerable challenges to techniques originally developed for English; because of this, an important area of cur- rent research concerns how to model these phenom- ena more accurately within statistical approaches. In this paper, we investigate this question within the context of parsing Spanish. We describe two meth- ods for incorporating detailed features in a Spanish parser, building on a baseline model that is a lexical- ized PCFG originally developed for English. Our first model uses morphology to improve the performance of the baseline model. English is a morphologically-impoverished language, while most of the world’s languages exhibit far richer mor- phologies. Spanish is one of these languages. For instance, the forms of Spanish nouns, determiners, and adjectives reflect both number and gender; pro- nouns reflect gender, number, person, and case. Fur- thermore, morphological constraints may be mani- fested at the syntactic level: certain constituents of a noun phrase are constrained to agree in number and gender, and a verb is constrained to agree in num- ber and person with its subject. Hence, morphol- ogy gives us important structural cues about how the words in a Spanish sentence relate to one another. The mechanism we employ for incorporating mor- phology into the PCFG model (the Model 1 parser in (Collins, 1999)) is the modification of its part-of- speech (POS) tagset; in this paper, we explain how this mechanism allows the parser to better capture morphological constraints. All of the experiments in this paper are carried out using a freely-available Spanish treebank pro- duced by the 3LB project (Navarro et al., 2003). This resource contains around 3,500 hand-annotated trees encoding ample morphological information. We could not use all of this information and ade- quately train the resulting parameters due to lim- ited training data. Hence, we used development data to test the …

Content truncated to 500 words