M.Sc Thesis

M.Sc StudentBenjamin Eran
SubjectVoice Analysis System Supporting an Electrophysiological
Study of Word Production Initiated by Visual
DepartmentDepartment of Biomedical Engineering
Supervisor PROFESSOR EMERITUS Hillel Pratt


Fourteen subjects (all men) between the ages of 20 and 30 years old took part in this study. All of the subjects performed a naming task of 61 visual stimuli which were selected carefully as proper representatives of the words they symbolize. The drawing symbolized Hebrew bi-syllabic words whose age of acquisition is very early. Therefore, validation of the drawings was done by testing 35 children (15 4-5 years old, and 20 7-8 years old) and 10 adults. All words began with a plosive consonant. A Voice key indicated the response latency which allowed averaging the brain potentials that were time-locked to the response. Thus, it was possible to identify the response-locked components and to estimate their generators by using LORETA.

The results showed that reaction time is correlated to the size of the word pool from which the subject retrieves the words. Thus when the pool is the subject’s nouns vocabulary it takes longer to produce a word in response to a visual stimulus (626 ms), in comparison to a limited pool of sixty one words (577 ms) and word repetition (“one word” pool, ~510ms ).

The study found N170 to have a role in the process of naming. Its detection in both the stimulus-locked and the response-locked brain waveforms indicates its role as a junction between the electrophysiological manifestation of stimulus evaluation and the electrophysiological activity in preparation for articulation. Source estimation revealed occipital brain regions in the stimulus-locked and the response-locked generators estimation. It also showed higher activity in the parietal lobe in the response-locked source estimation than in the stimulus-locked estimation. These electrophysiological findings indicate that these regions take part in both analyzing the stimulus and in the preparation for the articulation. The higher parietal lobe activity as shown in the response-locked source estimation seems to be related to (and derived from) motor activity preceding speech articulation.

We investigated brain regions, in measures of extent of activity and level of activity begins with N170. The results showed the left temporal regions to be more active than the right regions. The time course of activity in the left temporal regions had two stages. The earlier stage showed activity in Broca, Wernicked and Brodman area 21. The later stage showed higher activity in all the regions including Brodman 39 and 42.