The role of parameters in phonology: a critical account

Abstract

Parameters have been used in generative phonology for the last fifteen years, but there is no unified account of what they have to say and how they work. The purpose of this paper is to question the empirical and heuristic value of parameters in most of their uses and to elucidate the relationship between Markedness Theory and parameters. I shall attempt to show that approaches labeled Principles and Parameters (PP) are but a reformulation of Markedness Theory (MT), vitiated by the same defects and leading to the same explanatory failure; that the PP reformulation may in fact be less adequate than former models of MT, since it is even less refutable—mainly because it is less articulated, organized and formalized; and that the function of parameters in current phonological literature seems mainly rhetorical due in part to what seems to be a refusal to define precisely what is a possible candidate for a parameter. We would perhaps want to make a special proviso for the use of parameters in metrical theory, as opposed to their use for segmental and syllabic structures, and in a limited way for Optimality Theory ( Prince and Smolensky, 1993, Optimality Theory: Constraint Interaction in Generative grammar. Piscataway, N.J.: Rutgers Center for Cognitive Science), for which, although provided with more formal definitions, parameters play a more marginal role, whereas MT is still used as a justification. In the case of metrical theory, parameters have a rather strong and restrictive meaning, whereas in the latter cases they are of a more metaphorical nature. Since markedness theory has not been abandoned, and is still often referred to in the phonological literature, as we will see, universals of grammar in the form of principles and parameters take over, and sometimes mix with, markedness theory, leading to an undesirable situation characterized by parallel, informal and redundant discourses on alleged universals. Statements of PP can be reduced either to ‘primitives’ of markedness theory, that is, implicational laws, or to the simple meaning of ‘any conceivable choice’. It will be suggested that fruitful research for universals and ‘core’ phonology could only begin once we admit that current approaches have mainly concentrated on peripheral properties, and that insight into the specific linguistic capacities of humans could not be attained without factoring out universals for which physiological explanations are readily available and which partake only trivially in the language faculty (for a similar point of view, cf. Singh. 1996. ‘Natural Phonology: A view from the outside’, Natural Phonology: The State of the Art ed. by Bernhard Hurch & Richard Rhodes, 1–38. Berlin: Mouton). The focus will be mainly on three questions: (1) what precisely is the relationship between MT and PP? (2) If (or when) they are the same, is PP only an informal version of a former MT? (3) If not, what is the content of PP (if any)? In the first section, I will briefly sketch the concept of parameter, while the second section will point out the manner in which markedness and markedness theory are referred to, from its formalization in Kean. The Theory of Markedness in Generative Grammer. Bloomington, Ind.: Indiana Univ. Linguistics Club (1980) to a theory of phonological complexity (Calabrese. 1991. The Notion of Phonological Complexity in Phonological Theory: A new approach to Markedness theory. manuscript, Harvard University), default rules of underspecification theories (Archangeli and Pulleyblank. The Content and Structure of Phonological Representations. manuscript, University of Colorado and University of Southern California 1986, Paradis, and Prunet. The Special Status of Coronals. New York: Academic Press), feature geometry, syllabic structures and Optimality Theory. The third section examines definitions and examples given in PP approaches, and will point out the absence of any articulated theory of phonological parameters as well as of any constraining definition of the term, with the consequence of its use as a device to avoid refutability of propositions about universals that do not have anything to do with core grammar. The fourth section points to the more successful use of parameters in metrical theory, and the fifth discusses the explanatory value of MT and PP.