Negation is a crucial issue in sentiment analysis and opinion extraction as it can fundamentally alter the meaning and polarity of an opinionated statement. How-ever, numerous opinion-mining tools (lexicons, models, and classifiers) which are currently in use to identify and classify sentiment in texts written in Arabic language often fall short of expectations to account for negation structures, resulting in inaccurate classifications. To address this issue, this research proposes a linguistic approach using the Nooj platform to formalize recognition rules for negation in Arabic text and apply them with an opinion lexicon to a manually collected corpus of opinionated data obtained from websites and social media posts. The method utilizes sets of local grammars specifically designed to identify negation structures in subjective Arabic texts and assign the appropriate polarity to the sentence. Through the application of these recognition rules and the use of an opinion lexicon, the proposed method yields encouraging results and demonstrates the ability to accurately identify and classify negation sentences in text written in Arabic. This research aims to improve the overall accuracy of sentiment analysis in the Arabic language by incorporating the proper handling of negation into the classification process.