Distinct representations of phonemes, syllables, and supra-syllabic sequences in the speech production network

Abstract

Functional neuroimaging studies have converged on a core network of brain regions that supports speech production, but the sublexical processing stages performed by the different parts of this network remain unclear. Using an fMRI adaptation paradigm and quantitative analysis of patterns of activation rather than contrast subtractions alone, we were able to identify a set of neural substrates predominantly engaged in phonemic, syllabic, and supra-syllabic levels of processing during speech. Phoneme-level processes were found in the left SMA, pallidum, posterior superior temporal gyrus, and superior lateral cerebellum. Syllable-level processes were found in the left ventral premotor cortex, and supra-syllabic processes related to phonological chunking were found in the right superior lateral cerebellum. Active regions that were not sensitive to sublexical manipulations included primary motor and auditory cortical areas, and medial cerebellum. These results offer a quantitative technique for localizing sublexical neural processes that are difficult to dissociate using non-invasive imaging techniques and provide the beginnings of a "brain map" for language output.

Keywords: SMA; fMRI; premotor cortex; speech.

Figure 1

The four experimental conditions and baseline condition of the fMRI study. Each black box corresponds to a single presentation of a pseudoword displayed on the screen in white on a black background. There are no color indications of syllabic structure in the experimental setup; different colors are used here only to highlight the syllabic boundaries (according to French language rules) and the differences between conditions. In the Identical condition, the same word is repeated six times within a block of trials. In the Reordered condition, two words formed by reordering the same two syllables are repeated three times each. In the Resyllabified condition, two words formed using the same phonemes but different syllabification are repeated three times. In the Variable condition, six unique words with unique syllables are produced. The Baseline condition consisted of silent viewing of a string of X’s.

Figure 2

Predicted activation patterns for putative functional representations. Depending on the pattern of activity across the four experimental conditions, the speech regions can be classified into four groups: a) phonemic, b) syllabic, c) supra-syllabic, d)phonologically-insensitive, and e) task difficulty (see Materials and Methods for details). The bar plots in the center column represent the relative level of signal change across the four conditions (conditions 1 through 4 from left to right). On the right, each model condition activity is assigned an effect size value from 1- 4 representing the relative size of activation in that condition.

Figure 3

Significant voxel-wise activity in the collapsed – baseline contrast. Color scale represents t statistics plotted on surface-rendered canonical SPM brain. The statistical image was thresholded at P_FDR < .05. Left hemisphere is represented by the top two cortical images on the left side and right hemisphere is represented by the top two cortical images on the right side. The figure also includes coronal slice series through the thalamus and basal ganglia (image on the extreme left) and the cerebellum (image on the extreme right) regions. In the slice series, the left hemisphere is shown at the left of the image. The MNI y-axis level of each coronal image is indicated by the adjacent coordinate.

Figure 4

Predicted activity patterns (left column) and the actual patterns of ROIs that match one of the predicted patterns with a corrected p value of less than 0.05. [Abbreviations: vMC = ventral motor cortex; vPMC = ventral premotor cortex; SMA = supplementary motor area; dIFo = dorsal inferior frontal opercularis; pSTg = posterior superior temporal gyrus; Hg = Heschl’s gyrus; amCB = anterior medial cerebellum; splCB = superior posterior lateral cerebellum; spmCB = superior posterior medial cerebellum; Pal = pallidum]

Figure 5

Brain map of ROIs and their best predicted activity pattern match. Color indicates activity pattern found in the ROI. Gray ROIs were active during at least one of the speech – baseline contrasts but did not meet the criteria for a match to one of the predicted patterns (see main text and Table 1 for further information, including region abbreviation definitions).