This analysis revealed a RE/LH advantage, F(1, 56) = 34.80, p < .0001, and differences regarding the group of participants, F(3, 56) = 3.65, p < .02. However, the effect of hemisphere varied with the group, as revealed by the interaction of the factors Group and Ear/Hemisphere, F(3, 56) = 4.38, p < .01. This interaction is illustrated in Figure 14. As seen in this figure, all groups except for the right tinnitus patients exhibited a RE/LH advantage. Contrast analyses comparing left- and right-ear performance for each group confirmed the RE/LH advantage for every group except for the right tinnitus group, F(1, 56) = 13.45, p < .001, in BT group; F(1 ,56) = 21.50, p < .0001, in LT group; F(1, 56) = 21.41, p < .0001, in CO group; and F(1, 56) < 1, in RT group. Figure 2 reveals that the absence of lateral differences in the RT group was principally the consequence of a low RE/LH performance. In addition, because we hypothesized a greater RE/LH advantage for LT patients compared to controls, we performed a contrast analysis on the Group by Ear/Hemisphere interaction that compared these two groups. This analysis did not confirm that LT patients demonstrated a greater RE/LH advantage than normal controls, F(1, 56) = 1.93, p > .15.

In order to ensure that these results were not biased by differences in participants’ overall accuracy, we performed an analysis using the laterality index (LI) {(RE-LE)/(RE+LE)}. This analysis revealed significant differences between groups, F(3, 56) = 4.78, p < .005 (bilateral tinnitus: LI = 16.4; right tinnitus: LI = -1.0; left tinnitus: LI = 29.0; controls: LI = 13.8). Moreover, contrast analyses comparing each group confirmed the smaller laterality index for right tinnitus patients compared to bilateral tinnitus patients, F(1, 56) = 4.77, p < .04, left tinnitus patients, F(1, 56) = 14.18, p < .0005, and controls, F(1, 56) = 5.18, p < .03. These contrast analyses also revealed a higher laterality index for left tinnitus patients compared to controls, F(1, 56) = 5.45, p < .03. Thus, these analyses confirmed the previous ones based on left- and right-ear/hemisphere data.

In order to test the influence of the stimulus position within each triplet, we performed an additional analysis where the effect of the rank of the word recalled within the triplet (R1, R2, and R3) was tested. This analysis included Group as between-subjects factor, and Ear/Hemisphere of recall and Rank as within-subjects factors.

This ANOVA confirmed the above-mentioned effects of Group and Ear/Hemisphere and further revealed that the last word of the triplet was reported more accurately (R3 = 5.5) than the two others (R1 = 3.7; R2 = 3.3), F(2, 112) = 29.49, p < .0001. No interaction of the rank and any other factor occurred.

Eventually, we analyzed side-inversions errors (i.e., errors consisting in reporting a word-stimulus presented to the non-cued side). We called “RE for LE inversion” a situation where a participant recalled a word presented to the right ear, although he had to report words presented to the left ear, and vice-versa for “LE for RE inversion.” This analysis included Group as between-subjects factor and Inversion-Side as within-subjects factor.

The ANOVA only revealed that more errors of the “RE for LE inversion” type (M = 4.40) were observed than of the opposite type (M = 2.75), F(1, 56) = 16.09, p = .0002. This result was consistent with the RE/LH advantage observed in most of our subjects.