Text classification is widely used in tasks such as sentiment analysis, spam detection, and hate speechdetection, but performance often suffers when models rely on only linguistic features or semantic embeddings.Single-representation methods struggle to capture both syntactic structure and deep semantic meaning, limitingrobustness and generalization across tasks. This study proposes a hybrid feature fusion framework that combinesinterpretable linguistic features with semantic embeddings from Doc2Vec and transformer-based models. Theapproach is evaluated on five benchmark datasets covering fake news detection, Bloom’s taxonomy classification, andhate speech detection. Experiments with multiple classifiers show that the fused features consistently outperformsingle-feature baselines. The best results, using a BERT-based fusion approach, achieve accuracies of 81% for fakenews detection, 67% for Bloom’s taxonomy classification, and 72% for hate speech detection, with improved precision, recall, and F1-score. Overall, the results demonstrate that integrating linguistic and semantic featuresprovides a robust, domain-agnostic solution for improved text classification.