Analisis Kesalahan Pelafalan Fonem Bahasa Arab Mahasiswa Institut Agama Islam Pemalang Menggunakan Speech Recognition: Perspektif Ilmu Ashwat

Abstract

Accurate pronunciation of Arabic phonemes according to their correct makhraj is essential for Arabic learners, especially in institutions preparing graduates to read and teach the Qur’an. Students of the Arabic Education Study Program at Institut Agama Islam Pemalang often mispronounce phonemes that do not exist in Indonesian, particularly pharyngeal, emphatic, and interdental sounds. Conventional analysis based only on lecturers’ hearing is time-consuming, subjective, and difficult to apply consistently, creating a need for an objective technology-assisted method. This study aimed to identify the patterns and frequency of Arabic phoneme pronunciation errors and explain their causes from the perspective of Ilmu Ashwat. Using a mixed-methods explanatory sequential design, quantitative data were collected through a phoneme-comprehensive reading test involving 52 students selected by total sampling. The recordings were processed using an automatic speech recognition application and validated by two Ilmu Ashwat lecturers. Qualitative validator notes were analyzed using the interactive model of Miles, Huberman, and Saldaña. The results showed that pharyngeal and interdental phonemes were the most frequently mispronounced. The speech recognition results also demonstrated high agreement with expert judgments, indicating its reliability as an initial screening tool. The main causes were Indonesian phonological interference and limited makhraj articulation practice. The study offers a replicable diagnostic procedure combining speech recognition efficiency with expert linguistic validation in Indonesian Islamic higher education.