
Speciale  modelbyggervariant, 2. modul, Vinter 2011, id:431  
Vejleder:  Stig Andur Pedersen og Anders J. Hede Madsen 
Findes på RUb:  Ja 
In the present thesis, a theory of semantic competence is modeled using tools from epistemic logic. The resulting formal model is used to analyze a problem from the philosophy of language, namely Frege's Dilemma. There are two aims of the thesis: to construct a formal theory of semantic competence, and to show that the formal theory can be used as an useful analytical tool in uncovering the informational structure behind problems from the philosophy of language. The first aim is achieved by, first, deciding for which theory of meaning a theory of semantic competence is wanted. Due to its simplicity, Millianism is chosen. Then various nonformal theories of semantic competence are evaluated with respect to finding one which allows for an objective, intersubjective comparison of competence levels. It is argued that the conceptual theory of (Marconi, 1997) is the best choice: the theory has a clearly defined structure making modeling possible, and is based on empirical studies from cognitive neuropsychology. Following these initial choises, the modeling framework and its philosophical interpretation is presented. The framework used is epistemic logic, and both the propositional and quantified versions are introduced. As a more expressible logical language is required, manysorted quantified epistemic logic is presented, and a novel, general completeness result is shown for manysorted extensions of quantified modal logic. Having thus set the stage of achieving the first aim, a slightly simplified version of the theory of (Marconi, 1997) is modeled. A suitable modelclass is defined and a meaning function is added to capture Millian meaning. Based on the shown completeness result, a sound and complete axiom system is presented, and a logic representing the formal theory is thereby found. The model is then validated. It is shown that both the essential ontological properties as well as the competence types from Marconi's theory are present. It is further shown that the formal counterparts of the competence types from Marconi's theory adhere to the principles dictated by empirical studies. Thereby, the first aim is achieved. To accomplish the second aim, proof of concept is shown. This is done by analyzing an objection to the correctness of the Millian theory of meaning, namely Frege's Dilemma (Frege, 1892). The formal theory is used to analyze both disjuncts of the dilemma, while focusing on the epistemic situation of the agent, i.e. the agent's level of semantic competence. The formal theory of semantic competence allows for multiple notions of semantic competence, each resulting in a unique rendering of the dilemma. Based on these analyses, it it is concluded that once the underlying informational structure of the discussed situations is revealed, neither disjunct proves to be a problem for the Millian theory of meaning. Hereby, the second aim is accomplished. However, I raise an intuitive objection to one of the analyses. It is argued that the objection introduces an unaccounted for parameter, namely contexts. In order to show that this objection is not fatal for the proposed analysis, a chapter is devoted to the construction of a contextual theory of semantic competence. The notion of contexts is incorporated into the models for semantic competence, and the possibilities for finding a complete axiomatic system is discussed, but no completeness result is shown. Therefore, a formal theory, i.e. a logic, for contextual semantic competence is not presented. However, the modeltheoretic machinery is used to reanalyze the problematic case. It is shown that when the situation is modeled in a contextual model, the epistemic analysis of the disjunct again showed the Puzzle about Identity is unproblematic for the Millian view. Overall, the constructed formal theory of semantic competence is shown to elucidate informational aspects of the problems posed to the philosophy of language by Frege's Dilemma. In particular, once the informational structure of the problems is clear, it is shown that each argument is far from being as decisive against Millianism as has been the mainstream view in 20th century philosophy of language.