An fMRI study of overt vs. covert production of mono- and multisyllabic words

By L.I. Shuster, and S.K. Lemieux

Functional magnetic resonance imaging (fMRI) has been used to study normal speech and language processes as well as disorders of speech and language. One challenge in using fMRI, however, is that movement, particularly near the field of view, introduces motion artifact. Movement may produce spurious signals or it may distort the fMRI images. Since the speech articulators are near the brain, their movement poses particular difficulty for analyzing fMRI data. One way that investigators have dealt with this issue is to use silent “speech,” or speech that is not vocalized; however, there are several problems with this approach, including the loss of the normal auditory and tactile/kinesthetic feedback that typically accompanies speech. An alternative procedure is to reduce motion artifact during the data analysis process. However, this method has been applied only to fairly brief movements. The purposes of this investigation were twofold. The first purpose was to provide further information regarding whether vocalized speech and silent “speech” produce the same pattern of fMRI activation. The second purpose was to determine whether speech involving more motor sequencing involves greater activation of brain areas thought to be responsible for this sequencing, in particular the left insular cortex, than does speech requiring less planning.