Abstract:
A grammar checker checks each sentence in a text for conformity to grammar rules. If conflicts are detected, it suggests corrections. This paper proposes an approach to detecting and correcting Tamil grammatical errors related to subject-verb disagreement (in terms of person, number, and gender) and tense aspect conflicts. Hierarchical POS tags of words are used for detection, and two sets of Tamil grammar rules are applied for correction. The first set handles subject-verb disagreement, while the second set addresses tense aspect mismatches. These rules are implemented as Python conditional statements based on standard Tamil grammar. Suggestions for corrections are provided for third-person personal pronouns and finite-verbs across all subjects. Additionally, a tense-aspect mapping is used to improve correction accuracy. Tests on 100 sentences produced 136 suggestions, of which a Tamil language scholar approved 132 as optimal and 4 as acceptable.
