| dc.description.abstract | This paper aims at problematizing the treatment offered to polysemy in a frame-based lexical resource. In order to achieve such an objective, we are going to establish an approximation between Frame Semantics and the cognitivist approach to polysemy, aiming at correlating frames and lexical networks. To do so, we depart from some well-known studies about polysemy, like Brèal (1992), Ullman (1964) and Lyons (1977), in order to place the subject of this study in time and space. The second theoretical chapter in this dissertation presents the paradigm known as Cognitive Linguistics, introducing its main commitments, as well as the historical background that surrounds it. In this fashion, we approach the contributions of Evans e Green (2006). We also talk about Frame Semantics (FILLMORE, 1982), a theory situated under the scope of the cognitivist paradigm which considers meaning from the encyclopaedic point of view, not distinguishing either linguistic nor contextual information in the process of meaning assignment. We also approach the FrameNet, an online lexical resource based on the principles of Frame Semantics, in order to problematize the treatment dismissed to polysemy by such a resource. Still in the second theoretical chapter, we present the cognitivist understanding of polysemy, departing, firstly, from Langacker (1987) and Geeraerts (2006), when it comes to the notions of complex category and flexibility, respectively, and, on the second hand, from Lakoff (1987) and Fauconnier and Turner (2006), which offer theoretical models that enable one to approach polysemy, like the radial categories and the conceptual blending. The applied section of this paper consists of applying the radial categories proposed by Lakoff (1987) to the descriptions of the senses related to the noun ataque and to the verb marcar – the already known application of the lexical networks – as well as to the descriptions of the frames evoked by such senses – being this the application we propose in this dissertation. We believe that our proposal of integrating network representations and frames is justifiable once (i) it follows a Cognitive Linguistics principle according to which meaning is a conceptual entity, as well as, in the same way, frames are conceptual in nature, and (ii) it follows a principle of Frame Semantics which advocates that, to know the meaning of a word, one must necessarily be able to assign the frame evoked by such a word. In this case, a frame is a wide knowledge structure that provides encyclopaedic information about the uses of such a word in a speaking community. What our proposal advocates is that the application of radial networks not only to the senses, but also to the frames evoked by them is able to provide different kinds of information to the description of polysemous items: departing from the networks filled with senses, one finds information concerning the lexicon; departing from the networks filled with frames, one has access to wider cognitive structures which underlie these senses, that is, the frames themselves. Our corpus study shows that these words, in their uses related to football, are polysemous, and, thus, they can be considered as complex categories. This result also confirms the cognitivist hypothesis that states that polysemy is a regular phenomenon in language, in opposition to what advocate the ideas presented in the first chapter of this paper, to which polysemy is an epiphenomenon, something secondary in language. | en |
