The locus and time course of long-term morphological priming

Abstract

Two experiments investigated the effects of long-term morphological priming in the fragment completion task. Completions for some of the fragments were presented visually during a preceding task, and others were presented auditorily. In addition, some of the target completions were morphologically related to words that were presented visually during the study task, while still others were unrelated to any of the study words. Fragments were most likely to be completed if either the completion or one of its morphological relatives was presented visually during the study task. Analyses of response latencies also indicated that the time course of morphological priming was similar to that of visual identity priming and that both morphological and visual identity priming had earlier influences than auditory identity priming. Overall, the results indicate that morphological priming includes a modality-specific component that reflects the operation of processes that occur relatively early in the time course of processing.

Acknowledgments

This research was supported by Grant HD-01994 from the National Institute of Child Health and Development to Haskins Laboratories. We would like to thank Harald Baayen and an anonymous reviewer for their helpful comments

Notes

¹Note that we could have used cross-modal morphological priming, rather than cross-modal identity priming, as our index of the modality-independent contribution to priming. Also note that identity priming is a special case of morphological priming—one in which only the target morpheme is presented during the priming event. We chose to use cross-modal identity priming to maximise the magnitude of cross-modal priming, and thus provide a conservative test of the hypothesis that (visual-visual) morphological priming includes a modality-specific component.

²Kolmogorov–Smirnov tests for normality confirmed that the four distributions are all non-normal (p<.001 for all four distributions).

³Because there were only eight observations per item per condition, analyses with items treated as random variables were not conducted.