Analyzing syntactic structure is the most complicated task for Indian Languages. In this paper, a probabilistic parser is proposed for Hindi language comprising the empirical and rationalist approaches. The task of tagging is accomplished with the help of TnT POS tagger. In this research work, along with the development and evaluation of probabilistic parser, evaluation of rule based and conditional random fields (CRF) based shallow parser is also done by using a test dataset of 100 tagged sentences of Hindi. The generation of probabilistic parser is formulated mainly by using rule based shallow parser, constructing grammar rules and assigning probabilities. The proposed probabilistic parser has shown the accuracy of 66%.