N-Gram Statistical Grammar Checker for an Indian Language

Abstract

This study aims to illustrate the author’s findings about the design and the advances that have occurred in the Statistical Grammar Checker, specifically for a robust, vibrant and a rich morphological language, that is, Punjabi. Along with this, several other submissions for Grammar Checking have also been discussed. How in a given sentence, at first the Parts-Of-Speech tags of particular words combining together to build a new sentence are analyzed, and how then the N-gram based probabilistic strategies are used to conclude whether the sentence stands vague or conveys a desired meaning, forms the basis of this research article. The system is, however, tested with definite types of corpus. Additionally, Bigram and Trigram probabilities have also been considered and calculated. As a result of which, The Recall of Statistical Grammar Checker (92.29) and Precision (93.14) have successfully been achieved.