Phonology-driven decomposition

This project asks whether decomposition is affected by phonologically-conditioned morphological alternations, in which the alternants realize the same value of a given morpho-syntactic feature and can be accounted for by assuming language-wide, regular phonological operations triggered by the surrounding phonological context. For example, the realization of this prefix “in” alternates between [im] and [in], on whether the following root-initial phonological segment is a labial consonant or not: i[n]tollerable vs. i[m]possible. These alternations are generally dealt with as resulting from general phonological processes (rules and/or constraints) operating on a single underlying form. In the example above, it is generally assumed that the underlying form of the prefix in is [in], in which the nasal becomes [m] via an assimilation process triggered in front of labial sounds. One way to test this is to elicit the priming response to word pairs in which each alternant occurs either as a prime or as a target: e.g., intollerable-IMPOSSIBLE and impossible-INTOLLERABLE. In doing this, we ask two different, but related questions. The first question asks whether decomposition is able to detect the morpho-phonological relationship between alternating realizations of the same morpheme. The second question asked whether decomposition is able to detect the morpho-phonological asymmetry between phonologically-conditioned alternants, whereby one alternant (i.e., [im] phonologically derives from the other (i.e., [in]).

Alternations such as the one exemplified above represent the simplex case of phonological conditioning in morphology, as they involve exactly one phonological operation. If the priming response were found to be sensitive to such simplex cases, it may be probed as a way to test competing phonological accounts for more complex phonologically-conditioned alternations, in which more than one phonological operation interact with one another. Derivational theories assume that phonological processes are ordered serially so that the surface form results from multiple, intermediate stages in which the underlying form undergoes rule in a fixed order (Chomsky & Halle 1968). Representational theories assume, instead, that phonological processes all occur in parallel and compete with one another for output selection ( Optimality Theory: Prince & Smolensky 2008). These two theories of phonology therefore make different predictions with respect to the number of grammatical operations involved in the computation, which have been argued to affect processing time and therefore response times ( derivational theory of complexity, DTC; Miller & Chomsky1963, Phillips 1996). On one hand, derivational theories predict that the parser analyzes the stimulus by applying the rules “in reverse” to go back to the underlying representation (which is assumed to be the long-term representation). As such, decomposition procedures are predicted to take more time depending on the the number of rules to undo to get to the corresponding long-term representation. On the other hand, representational theories predict that multiple phonological processes are all evaluated in parallel. Therefore, decomposition procedures are predicted to be unaffected by the number of processes involved.