Abstract
We present new evidence that word translation involves semantic mediation. It has been shown that participants react faster to small numbers with their left hand and to large numbers with their right hand. This SNARC (spatial-numerical association of response codes) effect is due to the fact that in Western cultures the semantic number line is oriented from left (small) to right (large). We obtained a SNARC effect when participants had to indicate the parity of second-language (L2) number words, but not when they had to indicate whether L2 number words contained a particular sound. Crucially, the SNARC effect was also obtained in a translation verification task, indicating that this task involved the activation of number magnitude.
Notes
1 It might be objected that the significant effect of number magnitude in Experiment 2 is evidence for semantic mediation. However, the effect we observed is not a “regular” semantic number magnitude effect (i.e., longer RTs at the upper end of the scale). RTs were especially long for the word “trois” (three). This is probably due to the incompatibility between the orthographic representation of the word (which includes the letter “s”) and the phonological representation (which does not include the phoneme /s/, as the end-letter “s” is a silent letter in French). Participants had to give a “no” response, despite the presence of the letter “s” in the orthographic stimulus, which resulted in longer RTs. A similar incompatibility was present for the stimuli “cinq” and “dix” (five, ten): These stimuli do not contain the letter “s” but do contain the sound /s/.
2 As expected, the position of the L1 word on the screen did not make a difference for the presence of the SNARC effect. The coefficient of number magnitude for the trials with L1 in the upper position, dRT = 19.38 − (4.92 × number magnitude), did not differ significantly from the coefficient of number magnitude for the trials with L2 in the upper position, dRT = 17.29 − (5.21 × number magnitude), t(24) = 0.14.