Annotation and Issues in Building an English Dependency Treebank

The Paninian Grammar framework, given by Panini for his analysis of Sanskrit Language, is finding its extensive application on languages other than Sanskrit, about two thousand five hundred years after its formulation. The work presented in this paper is one such application that extends Paninian Grammar (PG or CPG: Computational Paninian Grammar) to English, a fixed word order language. It presents how CPG can account for English and makes available, a linguistically rich resource in the form of an English Dependency Treebank. At present, 2000 sentences have been annotated as part of this effort, using the Hyderabad Dependency Treebank (HyDT) Annotation Scheme for Indian languages, (modelled on CPG). In the course of this paper we talk about CPG and the annotation scheme used for this work. We then talk about the task of annotation of the English language data per the scheme and how its application to English varies from Hindi. Further, we discuss our handling of some constructions of English, and some anomalies in the language that pose a challenge to the application of this annotation scheme to English, as is.