Głębia eksplanacyjna w językoznawstwie generatywnym

Głębia eksplanacyjna w językoznawstwie generatywnym

The aim of this dissertation is to account for the concept of explanatory depth in generative linguistics. In the rst chapter some general remarks concerning the methodological speci city of the philosophy of linguistics are presented. The second chapter di erentiates the meanings of explanatory depth in the context of philosophical models of scienti c explanation: inferential (deductive-nomological model of C. Hempel and P. Oppenheim, uni cationist model of P. Kitcher, the notion of explanatory depth developed by J. Watkins), causal-mechanical (C. Hitchcock and J. Woodward), and explanatory analysis of cognitive phenomena developed by D. Marr. Several kinds of explanatory depth based on these accounts are identi ed. An explanation can achieve depth in distinct explanatory dimensions. In the deductive tradition the deeper explanation contains more abstract (theoretical) laws, principles, or theories within its premises. In the causal-mechanical approach an explanation is deeper if it provides more detailed analysis of causal factors or elements of a mechanism. In formal sciences the deeper explanation involves more primitive concepts or axioms that unify many di erent facts or underlie de nitions of other notions. Finally, in cognitive sciences an explanation in terms of computational level of analysis is sometimes thought to be more fundamental and deeper than algorythmic or implementational one. Chapter 3 highlights the meaning and scope of the term generative grammar . The chapter characterizes also main issues in the generative tradition, especially the poverty of the stimulus argument, the so-called evolutionary fable , Chomsky's methodological naturalism, competence vs performance distinction, and generative concepts of Internal/External language, grammar, and Universal Grammar. Chapter 4 presents the typology of linguistic data according to main research methods: o -line data (utterances, intuitive judgments) and on-line data that are provided by psycho- and neurolinguistic experiments. I suggest that we can speak of two types of interpretation of the theory of grammar weak interpretation based on o -line data and a strong one that is based on on-line data. The methodology of introspective judgments and elicited formal methods are brie y compared. Intuitive judgments are themselves acts of performance and cannot serve as a direct image of competence. Nevertheless, formal methods have their own limitations and introspective data remain epistemically primitive. Chapter 5 gives a broad outline of the development of Chomsky's generative grammar from the methodological point of view. The aim of this part is to show di erent dimensions of explanatory depth in the changing generative paradigm, and to show how it was intertwined with the notions of uni cation as well as simplicity. In the development of generative grammar three main periods can be distinguished: formal, cognitive (Standard Theory, Extended Standard Theory, Principles and Parameters Theory) and minimalist. In the early period explanation and simplicity were mutually related. The General Linguistic Theory was explanatory adequate if it provided evaluation procedure for grammars in terms of simplicity. After the cognitive turn these formal notions were given cognitive interpretation within the logical problem of language acquisition (i. e. Plato's problem). Explanation in its strictly methodological sense involved unifying disparate grammatical facts. The chapter depicts the uni cation of grammatical principles, phrase rules, and transformational rules accomplished by Chomsky. This historical preview shows also the concomitant degeneration of the grammatical apparatus of Government and Binding Theory which gave rise to many puzzles (in the sense of T. Kuhn). Grammatical theory in this period reached a level of a uni ed but highly con gurational description. The current version of generative grammar the Minimalist Program removes much of the 1 descriptive machinery of the preceding theory and turns representational/con gurational character of the generative procedure into a derivational one. It is argued that especially this property makes these inquiries deep in a new way because laws governing the operations which build the grammatical structure are more basic than the relations between elements of grammatical con gurations. The methodological analysis of the relation between Government and Binding Theory and Minimalist Program reveals in many cases that the change resembles revolution rather than correspondence. Explanatory depth can be captured with some quali cations in terms of explanatory extensions (in the sense of P. Kitcher). In chapter 6 Chomsky's transformational grammar is compared with some generative alternatives (Head-Driven Phrase Structure Grammar, Lexical-Functional Grammar, Combinatory Categorial Grammar). I argue that these theories can be treated as explanatory in some ways, however, they lack this kind of explanatory depth that characterize Chomsky's transformational grammar. This is due to their relatively stable grammatical apparatus as well as to their surface-oriented nature. The last chapter tends to identify the kind of explanatory depth which refers to General Theory of Language (i. e. theory of performance). Di erent conceptions of the relation between grammar and performance are presented. I argue that declarative generative theories can be described as instrumentalistic because they focus mainly on the descriptive and predictive power of the grammatical theory. Chomsky can be seen as a realist, however, in the special explanationist variety of scienti c realism according to which the theory that best explains some phenomenon can be said to be true, and everything that it postulates can be said to be real. The chapter previews some examples of psycholinguistic explanations: some linguists argue that perceptual strategies act in parallel with grammar (T. Bever), others that grammar is to a signi cant extent performance-driven (J. Hawkins). Next, I explore the idea of giving the strong interpretation to grammatical terms. It depends on the ontological status of grammar which is sometimes treated as an abstract system in the sense of Marr's computational level of analysis or as a mental object (state of the brain) which interacts with performance systems (Chomsky), or is simply identi ed with the parser/generator (C. Philips). Grammar as parser can have a strong interpretation and can attempt to achieve the reductive (psycholinguistic) depth. In the former cases the issue is more problematic. Although there are problems, such as the Granularity Mismatch Problem between linguistic and neurobiological theories, D. Embick and D. Poeppel rightly suggest that it should also be possible to nd some determinability connections between computational-representational and neurobiological levels of description. Some general conclusions follow from the analysis. The uni cation of grammatical facts is the hallmark of generative explanations from the beginning of the enterprise. Chomsky's style of theorising is like the one encountered in natural sciences. He always tries to subsume or unify once formulated generalizations and often reformulates them signi cantly in the face of recalcitrant data. Some of the former notions are no longer formulable in a new frame. These reformulations are often arranged mainly in aprioristic and somewhat arbitrary fashion as they cannot be experimentally con rmed or discon rmed. All these properties contribute to the conceptual elasticity of the generative paradigm which should be regarded as an advantage. In the Minimalist Program, however, the tendency to un cation, reduction, and simplicity makes the discussion more rigid. Several kinds of explanatory depth are present in generative linguistics. Grammatical explanations of the rst period were subsumed under the more general/theoretical laws of Principles and 2 Parameters Theory. It can be seen as a kind of theoretical depth. The Minimalist Program attempts to derive these generalizations from the deeper laws of derivation and from the minimal design speci cations. In this case the depth comes from the derivational nature of the generative procedure. The depth in Chomsky's generative grammar, however achieved only on conceptual grounds, should not be seen as a formal property. It seems that the neurobiological interpretation of the theory is not inconsistent with Chomskyan methodological naturalism as far as it does not assume radical physicalist reductionism.