INVESTIGATING THE ASSUMPTIONS OF AUTOMATICITY AND RELIABILITY IN MASKED MORPHOLOGICAL PRIMING

Abstract

Researchers typically assume that masked morphological priming (e.g. builder-BUILD) occurs below conscious awareness and is thus automatic and pre-attentive. Additionally, the morphological priming effect associated with an individual is assumed to be stable across time, given that the averaged priming effect have been robust across studies. Importantly, establishing the reliability of individual-level priming is imperative to its usefulness in predicting other measures, such as individual differences in language proficiency. These two fundamental assumptions of automaticity and reliability are tested in the present study. A masked morphological priming lexical decision task was conducted with a concurrent memory search task in a dual-task paradigm to reduce attentional resources. To consider the role of semantics in morphological priming, prime-target pairs varied as a function of semantic transparency (e.g., transparent pairs: farmer-FARM, opaque pairs: corner-CORN). Using these data, test-retest and split-half reliabilities were also computed. Results showed a graded transparency priming effect that was not moderated by memory load, and no reliability for individual-level morphological priming. This has important implications for research on individual differences in morphological priming as the instability of individual priming effects suggests that one cannot accurately discern systematic individual differences from random measurement error.