Abstract
People are highly efficient in recognizing faces. However, it is almost impossible for them to cope with huge datasets of facial images without any computational support. On the other hand, the way people describe the facial features using quite commonly encountered descriptors such as "long nose", "small eyes" and also allude to their feelings according to a specific person like "seems to be nice", may be utilized to enhance automatic face recognition systems. This offers an interesting possibility to incorporate human perception of faces and relations between facial features into machine-made computations. To address this aspect, one can engage the linguistic descriptors and the linguistic modeling. In this study, we present a comprehensive survey of the state-of-the-art studies and elaborate on some promising perspectives of the developments in this area.