Discrepancies between orthographic and unrelated baselines in masked priming undermine a decompositional account of morphological facilitation.

Abstract

This study uses the masked priming procedure to compare the decompositionality of regular with irregular English past tense forms relative to both an unrelated baseline and a baseline matched on orthographic similarity to the morphological prime. Morphological facilitation varies with the degree of similarity between related primes and targets. Discrepancies between unrelated and orthographic baselines arise when prime and target match in length and form overlap is high. The outcome demonstrates the key role of baselines in assessments of morphological facilitation and highlights problems of interpretation when evidence of morphological decomposition depends on meeting a statistical criterion for significant morphological facilitation.

‘Fell’ primes ‘fall’, but does ‘bell’ prime ‘ball’? Masked priming with irregularly-inflected primes

“…One thing that is not clear is why the fell-FALL items showed priming while the taught-TEACH items did not, especially since Pastizzo and Feldman (2002) reported quite similar mean values of orthographic similarity for the two conditions (67.9% versus 56.1% of position-specific letters shared across primes and targets), with no evidence provided that the groups of stimuli actually differed statistically on this factor. Further, it is not clear whether the orthographic control primes implemented by Pastizzo and Feldman (2002) were effective in accounting for the orthographic overlap between irregular primes and their targets.…”

“…Further, it is not clear whether the orthographic control primes implemented by Pastizzo and Feldman (2002) were effective in accounting for the orthographic overlap between irregular primes and their targets. The problem here is a general one that relates to the assumptions that researchers must make about the nature of orthographic input coding when designing their orthographic controls.…”

“…Critically, neither study included non-morphological orthographic controls designed to establish to what extent these effects reflected simple letter overlap across morphologically-related primes and targets. 2 This problem with orthographic controls was partially addressed by Pastizzo and Feldman (2002), who compared priming for irregular inflections with high (e.g., fell-FALL) vs. low (e.g., taught-TEACH) orthographic overlap against orthographically matched (e.g., fill-FALL, taunts-TEACH) and completely unrelated baselines (e.g., pair-FALL, slouch-TEACH). Results showed significant priming for the fell-FALL items against the orthographic baseline (but not against the unrelated baseline), and no priming for the taught-TEACH items against either baseline.…”

“…Though extreme examples like ate-EAT are not frequent, this issue applies to any prime-target pair in which there shared letters in different positions (e.g., bought-BUY, taught-TEACH). Because no stimuli were provided in Pastizzo and Feldman (2002), it is not possible to gauge the extent to which their stimuli suffered from this potential problem.…”

“…The problem here is a general one that relates to the assumptions that researchers must make about the nature of orthographic input coding when designing their orthographic controls. Pastizzo and Feldman (2002) reported designing their orthographic controls such that they preserved common letters in common positions across irregular primes and targets (i.e., they assumed a slot-based coding scheme in the design of their controls).…”