Dialects

The variety of Arabic dialects reflect the ethnic and social diversity of its speakers. There are two main classes of Arabic dialects: the Eastern dialects of Egypt, Sudan, and the middle East, and the Western dialects of North Africa. These dialect classes are distinguished by the reduction of the vowel system in the Eastern dialects and a contrast in the stress system [10]. The dialect classes can be further broken down into Gulf, Levantine, Egyptian/Sudanese, and Maghrebi dialect groups. Between these groups we see major phonological differences in realization of specific phonemes, such as the uvular stop qaf, the palatal fricative jim, and the interdental fricatives tha and dha. Morphological differences are also significant enough that spoken Moroccan and Yemeni are mutually unintelligible.

Dialects are a concern for speech synthesis for several reasons. What dialect is to be generated? One must decide between generating MSA, which is a mother tongue for no one, and one of the dialects, risking markedness and possibly a narrower potential listener base. MSA is widely understood only by communities with formal education in it, however, so its listener base is also limited.

Our system generates Egyptian Arabic, specifically Cairene, which enjoys broad exposure because of the prominence of Egyptian entertainment media across the Arab world.