This article focuses on the integration of rich variability of speech into more formal phonological models. The phonological variation often involves multiple factors, including internal factors such as morphology, syntax, and lexical identity, and external factors such as age, gender, style, register, identity, ethnicity, and socioeconomic class. One of ways employed to deal with non-phonological factors in an optimality-theoretic analysis is to include them in the grammar as so many independent constraints. Alternatively, one can imagine a modular solution where the phonological grammar is independent of the rest of the system and the interactions are modeled in some way that does not necessarily involve constraint ranking. The recent usage-based models of grammar such as exemplar theory have emphasized the role of lexical frequency as a factor in phonological variation. Optimality theory assumes that grammatical constraints interact in terms of strict ranking. The hypothesis has been challenged on both empirical and learning-theoretic grounds. Alternative proposals usually involve numerically weighted constraints as in harmonic grammar and maximum entropy models. These numerical models also include the original variable rules model.
Keywords: phonological analysis; harmonic grammar models; maximum entropy models; optimality theory; lexical identity; variable rules model
Article. 10450 words.
Subjects: Linguistics ; Phonetics and Phonology
Full text: subscription required