LAUSR

Objective Repetitive transcranial magnetic stimulation (rTMS) to map cortical language functions by inducing “virtual lesions” during task performance (e.g., picture-naming) has recently been introduced into pre-surgical diagnostics in brain tumor patients. We investigated to which extent the characteristics of the objects (words) used in the task influence the occurrence of semantically related language errors. Methods In a prospective single center study, we investigated thirteen right-handed, healthy volunteers (female n  = 6/male n  = 7) in 3 consecutive sessions by using 10 Hz, 30 Hz and 50 Hz-rTMS. After determination of the individual threshold reflecting motor cortex excitability, online-rTMS was applied during a picture-naming task: black-and-white drawings ( n  = 50) of everyday objects were presented simultaneously with rTMS (triggered, delay = 0 ms) with rTMS. rTMS bursts lasted for 1.5 s (15 pulses). The rTMS was applied over the frontolateral, parietolateral and superior/middle temporal area of the left hemisphere. The language errors were rated by two independent examiners using post hoc video analysis. The frequency (number of errors per 100 rTMS pulses) of errors related to semantic processing (i.e., anomia and semantic paraphasia) was quantified. The speech arrest or a not identifiable word response were a measure of the error strength. All the objects allowing alternative naming responses (e.g., ball/ football) were excluded. The remaining words were grouped by the number of sillables in one- (1S; N  = 19) and two-sillable words (2S; N  = 25). For each object, the word frequency was assessed (by using google hits in german language) and correlated to the error rates by Pearson’s product moment correlation. Error rates were compared between 1S and 2S using McNemar’s test for paired count data. Results The number of sillables (1S vs. 2S) had no influence on the total error rate (1S: mean N  = 73.9 errors/word vs. 2S: mean N  = 79.9) or the error strength (1S: mean N  = 19.5 vs. 2S: mean N  = 19.5). By contrast, the rate of anomias (1S: mean N  = 4.2 vs. 2S: mean N  = 6.4) increased with the number of sillables ( p 0.01 ), regardless of the word frequency in German language (n.s.). For semantic paraphasias, a weak statistical tendency was observed ( p  = 0.11), presumably due to few events (1S: mean N  = 1.4 vs. 2S: mean N  = 2.2). Regarding semantic paraphasias, the slightly higher rate of errors for 2-sillable words did not meet statistical significance ( p  = 0.11), presumable due to the rareness of the events (mean N  = 1.5 errors/word). Conclusion The complexity of the tested words, more specifically the number of sillables, seems to be significantly influencing the disruption of semantic word expression induced by rTMS. Therefore, using longer words for language mapping might be favourable in order to increase the true-positive rate of mapping results.