Reproducibility of the word frequency effect: comparison of signal change and voxel counting

We determined the reproducibility of both the direction and the effect size of the word frequency effect (WFE) as it relates to associative semantic judgments. Sixteen volunteers were scanned twice. At the group level of analysis, signal change and voxel counting could both reproducibly detect the existence of a WFE. However, signal change data showed less intersession variation, particularly in the left inferior frontal gyrus. The effect size of WFE was well reproduced only with signal change measurements. In consideration of the signal change data, statistical threshold did not have a major effect on the detection or determination of the effect size. In general, while the direction of the WFE was reasonably reproducible at the individual level, the effect size was far less well reproduced. These findings suggest that with existing techniques, fMRI may be used to track changes in brain activation stemming from improvement in language proficiency at the group level but not at the individual level.