On the efficacy of cross-modal vs. all-visual tasks in priming studies

Abstract:
In a series of articles, McKoon, Ratcliff and colleagues have argued that priming results obtained in experiments investigating reactivation effects of anaphors during sentence processing have been artifactual. However, Nicol, Fodor and Swinney (1994) pointed out that McKoon and Ratcliff (1994) used a all-visual task which was not necessarily comparable to the cross-modal task, and that this could explain their lack of results. We here specifically address the issue of the efficacy of cross-modal vs. all-visual design in studying priming during sentence processing, by directly comparing two sets of lexical decision experiments differing essentially in presentation mode.

1. The cross-modal experiment : In this experiment, each stimulus sentence was presented auditorily at a normal speech rate, and was preceded with a short context-neutralizing discourse (see (i), which also illustrates where visual lexical decision probes were temporally presented at positions 1 and 2). Visual probes were related to the subject of the first clause, thus examining whether probe words related the "strict" reading of the reflexive in the reconstructed VP were primed during sentence presentation. RTs were significant shorter to related compared to unrelated probes, but irrespective of probe position, F(1,20) = 14.78, p<.01.

2. The all-visual experiments : Here, the words of the sentence were presented centered on the screen on a word-by-word basis, with 300 ms presentation and 200 ms pause between each word. Probe words were presented in capital letters and in a different color off-center on the screen, and with the same duration. No significant main effects of probe position (p>.39), probe relation (p>.45) or significant interactions between the factors (p>.82) were found.

We also ran another all-visual experiment with probes related to the sloppy subject, with the same probe positions. Again, there were no significant main effects of probe position (p>.66), probe relation (p>.50) or significant interactions between the factors (ps>.98).

3. Conclusions : The two experiments had essentially identical linguistic stimuli materials, and differed mainly with respect to presentation mode, and rate of sentence presentation (as necessitated by all-visual design). The striking absence of priming in the latter task compared to the first provides strong support for Nicol et al.'s original contention concerning the non-comparability of the all-visual vs. the cross-modal paradigm in the study of priming in sentence processing.

Examples

(i) Postbudet kjøpte et silkeslips til seg selv, og flykapteinen ⁽¹ ) gjorde også det ⁽²⁾ .
"The mailman bought a silk tie for himself, and the air line pilot did also so."
Probes: FRIMERKE/BILMERKE, "stamp"/"car brand"