This paper considers the processing of an important class of spatial lexemes –called Internal Localization Nouns (ILNs)– which are used for designating the different parts of entities. It is grounded on a detailed linguistic analysis of the studied lexemes in terms of geometrical, functional and contextual/pragmatic properties. The experiment described in this work mainly focuses on ILNs calling for orientation notions. It consists in a pointing task which makes systematically vary the geometrical and functional properties of entities as well as the position in which they are displayed. The impact of these changes in the interpretation is evaluated through the measurement of response latencies. This protocol provides interesting data on the part played by several complex parameters –such as gravity, geometrical and functional saliency, motion, canonical use, etc. – in the processing of orientational ILNs.