Charles Reiss

Formal and Empirical Arguments concerning Phonological Acquisition

Smolensky (1996a) has proposed an ingenious solution to the well-known comprehension/production dilemma in phonological acquisition. In this article we argue that Smolenskys model encounters serious difficulties with respect to the parsing algorithm proposed and the learnability of underlying representations. Drawing on the generative literature in phonological acquisition, as well as the work of phoneticians and psycholinguists, we offer alternative parsing algorithms and examine their implications for learnability and the initial ranking of Optimality Theory constraints. Finally, we propose that the resolution of the comprehension/production dilemma lies not in the phonological domain (linguistic competence), but in the domain of the implementation of linguistic knowledge (performance).

Phonology as Cognition

This paper attempts to ground phonology within psychology. That is, we are interested in phonology as a branch of the study of mental representation, the psychology of mind. In order to develop this ‘phonology of mind’ we need to understand the relationship between form and substance in linguistic representation. A coherent account of this distinction has yet to be proposed for either phonology or syntax. We attempt to contribute to this necessary inquiry in the domain of phonology by first defining ‘form’ and ‘substance’, then critiquing some recent work which implicitly or explicitly touches on the relationship between the two. We will argue that current trends in phonology fail to offer a coherent conception of form and substance and are also inconsistent with basic principles of science. Since we are not proposing a complete alternative model of phonology, we invite the reader to reflect on how our proposals could be implemented or on how our assumptions (which we believe are widely shared in principle, if not in practice) should be modified. It has proven quite useful for linguists to conceive of a grammar as a relationship between i) a set of symbols—entities such as features and variables, constituents like syllables, feet, NPs, etc.; and ii) a set of computations—operations whose operands are drawn from the set of symbols, such as concatenation, deletion, etc. The set of symbols and relations together describe the formal properties of the system. Relevant questions in discussing formal properties include ‘Is the system ruleand/or constraint-based?’; ‘Do operations apply serially or in parallel?’; and ‘Are there limits on the number of operands referred to in the course of a given phonological computation?’ The issue of substance essentially arises only with respect to the set of symbols and the extent to which their behavior in phonological computation is driven by what they symbolize. For the sake of simplicity we restrict ourselves in this discussion to the set of phonological primitives known as distinctive features and to the representations which can be defined as combinations of distinctive features. We will concentrate in this paper on this notion of substance in phonological representation. In brief the question we are interested in is the following:

The Subset Principle in Phonology: Why the tabula can't be rasa

We demonstrate the logical necessity of assuming innate knowledge for language acquisition using toy grammars. The implications are applied to reconceptualizing the Subset Principle in terms of features, rather than, say, segments. Both syntactic and phonological issues are discussed, but the focus is on the acquisition of phonological inventories. We point out that our logical arguments converge with recent empirical results.

Deriving the Feature-Filling/Feature-Changing Contrast: An Application to Hungarian Vowel Harmony

The article explores an alternative to the interpretive procedure adopted in SPE and proposes a unified interpretive procedure for all languages. The proposal solves long-standing problems by making it unnecessary to refer to a third value of binary features [F], to introduce negation into lexical representations (e.g., [NOT + rd]), or to introduce a feature filling/feature changing diacritic on rules.The article provides a metric for comparing extensionally equivalent rule systems and argues that the most concise formulation is not always the correct one, by appealing to crosslinguistic evidence.The proposal is illustrated by application to the target/trigger relations in Hungarian vowel harmony.

Optimality Theory from a cognitive science perspective

The appeal to Richness of the Base in Optimality Theory may undermine many OT analyses, as well as OTs status as a psychological, and not merely computational theory. Constraint violability in OT theorizing is critiqued as arising from a confusion between epistemological and ontological issues. Following Kiparsky (1973), conspiracies are explained as epiphenomena due to the nature of language change and language acquisition.

Language Change Without Constraint Reranking

The notion that language change consists purely of constraint reranking is incoherent. Languages must also differ lexically. Cases of output-output correspondence in the literature actually reflect diachronic analogy, which is not a property of synchronic grammars, but is rather the result of lexical restructuring during acquisition.