Humans express their emotions mostly through speech. People can speak effectively through the active coordination of facial as well as vocal muscles. The proposed study attempts to analyze the muscle activation during the articulation of phonemes, particularly ‘അ’ (a) and ‘ആ’ (aa), which are commonly used in most Indian languages. In this work, the Malayalam language is selected for the study. Surface Electromyography signals are recorded from three facial muscles (Zygomaticus Major, Depressor Anguli Oris, and Mentalis) and one neck muscle (Anterior Belly of Digastric) of healthy female subjects using standard protocols. Root Mean Square (RMS) and mean frequency are extracted from the recorded signals. Further, classifiers namely Decision Tree, Random Forest, and KNN are applied to differentiate the phonemes ‘അ’ (a) and ‘ആ’ (aa). The results show that the mean RMS of the signals from Mentalis is greater during the articulation of phonemes ‘അ and ‘ആ’, compared to other muscles. In addition, KNN showed a maximum classification accuracy of 87 % while differentiating the phonemes with the extracted features. This study may be helpful for developing assistive devices for people suffering from speech disorders.