Table 1.

Differential brain activation of words and errors: Activation elicited by correct word-retrieval was compared to semantic paraphasias (words > semantic errors) and neologisms (words > neologisms). Furthermore, we compared activation elicited by erroneous responses (before therapy) to the same stimuli that were named correctly after therapy (correct trials > former errors; all contrasts: cluster threshold p < .05, k>50, FWE-corrected; voxel threshold p < .01, uncorrected).

	Hemi	Regions	BA	Z	k	x	y	z
Words > Neologisms	R	Inferior frontal	47	4.74	1635	42	28	-19
			45	4.51		59	19	21
		Middle frontal	46	4.19		50	42	20
	L	Frontal (precentral)	6	4.01	499	-50	-3	47
	L	Superior frontal	9	4.00	466	-21	43	39
Words > Semantic Errors	L	Frontal (precentral)	4	3.81	995	-50	-4	44
	R	Inferior frontal	45	3.78	629	59	18	18
Correct Trials > Former Errors (T2> T1)	R	Putamen		4.31	562	30	-14	9
	L	Putamen		4.24		27	-9	-7
	R	Thalamus		3.94		9	-17	1
	R	Inferior frontal*	47	4.11	437	42	28	-17
		Middle frontal	11	4.08		27	37	-17
		Limbic (anterior cingulate)	24	3.92		9	32	1

Hemi = Hemisphere; R = right, L = left; BA = Brodman Area; k = cluster extent; x/y/z = coordinates according to Talairach and Tournoux [36], Z = z-scores of significant voxels within significant clusters are reported (p < .01, k>50, corrected).

* The right IFG activation for the contrast correct trials > former errors (42,28,-17) differs with regard to localization to that from the contrast words vs. semantic errors, while the localization of the IFG related activation of this contrast and the contrast words vs. neologisms almost match perfectly. This might be explained by the fact that the first contrast includes 7 neologisms (and 3 sematic paraphasias) that may "dominate" the activation pattern.