Here the patient is presented with a noun (visual word) for 3 seconds, and is required to think of a related verb (e.g.: chair > sit). The version used by Bonelli and John Duncan's group is included. A third condition–"repetition"–is also included, where the patient simply subvocally repeats each word. Each block starts with a brief cue (e.g. "G: Generate" action or "R: Repetition"). Trails are 3s long, and are presented in blocks of 10 trials for a total of 120 trials. The repetition condition is not used in the standard analysis they published (though was modeled in analysis).

It is important to note that they use a very specific analysis procedure. They lateralized language using a frontal region of interest, and when laterality was uncertain they then used the verb generation task to determine laterality. An implicit baseline is used, with only the task conditions modeled in analysis.