The aim of this study is to investigate to which extent Arabic listeners can assess some reliable perceptual characteristics from samples of speech of different durations so as to identify the Arabic dialect in which they are produced. Our goal is to determine the minimum duration threshold that is required to achieve correctly a perceptual identification task. Results - analyzed in terms of reaction time (RT), Correct identification rate and errors analysis - are discussed in the field of experimental research and will be used to draw up the experimental design of an ongoing fMRI study dealing with the perceptual treatment of linguistic distance applied to the Arabic dialectal continuum.