An fMRI investigation of speech and tone segmentation.

Recent research strongly indicates that phonological tasks activate a subregion of the inferior frontal gyrus. The purpose of the present fMRI study was to investigate the extent to which activation of this region during phonological processing is due to speech processes per se such as articulatory recoding or to other cognitive task demands such as working memory. Thus, we compared activation patterns during segmentation of speech and tone sequences to a tone discrimination task. In particular, participants performed same/different judgments on pairs of words, pseudowords, and tone sequences that required segmentation of a continuous acoustic signal as well as tone pairs that did not require segmentation. Accuracy and reaction time data showed that speech and tone sequence segmentation conditions patterned more similarly to each other than to tone discrimination pairs. Analyses of group data revealed strong activation of the region at the border of the left inferior and middle frontal gyrus for all three segmentation conditions compared to tone discrimination, but no consistent differences were observed when word and pseudoword segmentation were directly contrasted. Analyses of individual subjects indicated that a large number of participants activated a small area of the middle frontal gyrus during the speech conditions compared to the sequences. These results suggest that a significant portion of active frontal areas is recruited for extracting acoustic information and maintaining it in memory for decision. However, some regions at the border of the inferior/middle frontal gyrus may be unique to speech segmentation.