Adapting a parser to historical English
暂无分享,去创建一个
The automatic syntactic analysis of natural language texts has made impressive progess. But when applied to historical linguistic texts, parser performance drops considerably. We first report parsing performance on several time periods in the Archer corpus. Second, we implement and evalutate a variety of adaptations to historical texts. We address spelling issues, adapt local grammar rules, and conduct an analysis of errors. Parser performance increases due to our adaptations. This pilot study also tests a number of more global extensions to address freer word order and different punctuation in earlier texts. We give an outlook on possible applications.