A Rule Based Approach for Detection and Correction of Grammar Errors in Written Active Voice Sinhala Sentences
Uresha Sewwandi, L. Ranathunga, Sumudu Wijethunge · 2021
Sinhala is the official language and major spoken language in Sri Lanka. The Sinhala language is one of the rare languages having diglossia, in which the written variety differs from the spoken variety. The written variety of Sinhala is very complex because it is defined by many grammar rules. Due to this complexity, there is not a specific tool to detect grammar errors in typed sentences as a web-based solution for grade five students to practice answering essay questions. This paper proposes a grammar checker module that helps primary school students who learn in Sinhala medium to identify grammar errors in input sentences. The proposed module also defines the type of errors and provides suggestions to correct errors in subject or verb or both. Therefore, this system enables to improve written competency in Sinhala learning and minimize grammar errors. A rule-based approach has been used for implementing the grammar module and the system is evaluated by the recall, precision, and accuracy, based on grammar error detection, error type identification, and error suggestion aspects. Precision, recall, F1 score for the grammar detection is 1, error type identification accuracy is 93.81% and suggestion accuracy is 83.85% within the defined scope.