Linking usage and grammar: Generative phonology, exemplar theory, and variable rules

Gregory R. Guy

    Research output: Contribution to journalArticlepeer-review

    Abstract

    Rule- and usage-based models in phonology are difficult to reconcile: 'rule'-based approaches (including generative and optimality models) rely on abstraction and seek to account for regularity and generality. Usage-based models, like exemplar theory, rely on concrete representations, eschewing abstraction; they typically seek to account for lexically differentiated phonological phenomena, including variability, gradience and probabilistic properties. An alternative that incorporates both generative productivity and quantitative precision is the "variable rule" (VR) model of sociolinguistic variation. VR preserves advantages of rule-based models, including abstraction and the capacity to represent categorical processes. But VR resolves many limitations of these formalisms using probabilistic quantification: any phonological process or constraint may be associated with a probability, which permits the treatment of variation and gradience. This paper cites evidence from variation in speech style, child language, and reanalysis across the life span showing that speakers have both discrete abstract and nondiscrete, variable elements of phonology. Variable processes provide a nondeterministic but recoverable link between these different representations.

    Original languageEnglish (US)
    Pages (from-to)57-65
    Number of pages9
    JournalLingua
    Volume142
    DOIs
    StatePublished - Apr 2014

    Keywords

    • Abstraction
    • Exemplar theory
    • Generative phonology
    • Lexical exceptions
    • Probabilistic grammar
    • Productivity
    • Quantification
    • Usage-based phonology
    • Variable rules

    ASJC Scopus subject areas

    • Language and Linguistics
    • Linguistics and Language

    Fingerprint

    Dive into the research topics of 'Linking usage and grammar: Generative phonology, exemplar theory, and variable rules'. Together they form a unique fingerprint.

    Cite this