Diglossia refers to a socio-linguistic situation in which two varieties of the same language are used for distinct purposes in everyday life. In Arabic, Spoken Arabic (SA) is the firstly acquired dialect used for oral and informal communication in everyday conversations. Literary Arabic (LA), acquired later in life through formal education, is used for reading and writing and by literate individuals in formal settings such as the media and official speeches. Because of the linguistic distance between the two Arabic varieties, some authors have suggested that SA and LA might cognitively function as first (L1) and second language (L2). Up to now, very few studies using imaging techniques had addressed the question of the neural basis of diglossia in Arabic native speakers. In this study, we sought to test whether or not the visual processing of high (LA-HF), low frequency LA words (LA-LF), and SA-HF words induce detectable differences in the brain responses collected by functional Near-Infra Red Spectroscopy (fNIRS). For this aim, a semantic categorization task, previously assessed in fMRI studies. Based on previous observations, it was predicted that LA words, will be processed faster and more accurately than SA ones. Furthermore, it was predicted that a modulation of the responses (by language conditions) will be found in the left frontal areas. At the behavioral level, the analysis of RTs revealed an effect of language variety in individual response variance and accuracy showed a clear advantage for LA-HF words over LA-LF and SA-HF ones. The analysis of oxygenation level revealed a significant response modulation in frontal and posterior areas by language variety. These results are discussed in the context of diglossia and the advantages/limitations of this new imaging methodology and its use to assess language processing in the brain.