Representation Of Arabic Pharyngeal Sounds Among Tamahaq Speakers In Al-hoggar : A Comparative Study Using Chatgpt

Abstract

This study investigates the role of artificial intelligence (AI), particularly ChatGPT, in the phonetic representation of Arabic guttural consonants produced by Tamahaq-speaking learners of Arabic in Al-Hoggar region. As a pilot exploratory study, it combines contrastive analysis, error analysis, International Phonetic Alphabet (IPA)-based transcription, and acoustic analysis using Praat software to evaluate the relative accuracy of AI-generated phonetic representations. The acoustic analysis focused on formant frequencies (F1/F2), duration measurements, and spectrographic observation. The findings indicate that ChatGPT provides relatively accurate phonetic approximations for several consonants; however, important limitations remain, especially with pharyngeal consonants that are phonetically marked in Arabic. The study concludes that AI can support phonetic analysis and IPA-based transcription, although expert verification remains necessary.