Diagnosing primary progressive aphasia (PPA) is difficult to diagnose even for experienced physicians and PPA leads to loss of speech and language, which are crucial functions for daily life. Participants were placed into groups of 10 based on different subtypes of progressive aphasia and different segments from two audiobooks were played in segments. While the subjects were listening to the audiobooks, linguistic features were derived and processed through a temporal response function using electroencephalography (EEG) and an ML-based classifier. An active electrode system was used at 25,000 Hz to generate an EEG, especially for theta and delta bands, which are important for speech processing and temporal response function (TRF) beta weights were used to analyze how the brain responds to stimuli (sound). The model was preprocessed and classified based on recall, precision, and F1 scores. Using these areas of interest, the EEG bands of frequency did not have an effect on the classification accuracy of PPA. However, TRF beta weights were proven to have significant contributions in diagnosing PPA and this proved to be more accurate than solely EEG data. The TRF biomarkers were successfully able to differentiate between subtypes in PPA, however, a diagnosis of PPA prior to differentiating between the subtypes has to have been made. Although a TRF model was able to diagnose between different subtypes of PPA, there needs to be more studies done not only using this data, but determining the best uses for TRF.