Phase 1: designing

ROELIEN BASTIAANSE: To reveal the language impairments of a patient with a tumour, we need a so-called language production test. In this video, I will discuss which criteria such a test has to meet and how we design such a test. For a test to be suitable in an intraoperative situation, an item cannot take longer than four seconds for the answer. This is how long the direct electrical stimulation is thought to have an effect. Longer sentences involving embeddings and subclauses are therefore not possible during awake surgery. A test that meets this requirement is a repetition test. The patient just has to repeat the word he just heard.

On the one hand, this is an excellent test for production, since we know exactly what the target is, so errors are easily detected. However, in this test, the patient does not have to retrieve the word himself, like in the picture naming test that we mentioned earlier. Those naming tests are more and more used during awake brain surgery with promising results. This is a piano. As we already know, a test like that should involve both nouns and verbs. Now, how many items do we need? And how do we know which picture to choose? Considering all the brain areas that may need to be mapped, the test should be relatively long to avoid repetition of the items.

Between 50 and 100 items is advisable. If you have those pictures, normally hand-drawn by an artist, we need to be sure that those pictures elicit exactly what we want. Look at these examples of a house, a noun, and swimming, a verb. These are good examples. If a patient sees the pictures, he will come up with one possible description. Now evaluate this item. Is this walking? Hiking? Or strolling? There may be too many possible answers, making it harder for us to judge if the answer is what the patient intends to say.

The next thing to control in order to get a balanced set of pictures is that you do not only include nouns and verbs that are often used, such as house and reading, but also less frequent words, such as abacus and fencing. During surgery, some patients can come up with frequent words, but not with the infrequent ones.

The selection of items made so far may still not survive a so-called piloting phase. Pictures you imagine to be good may have to be thrown out if your participants reacted differently than expected. Therefore, always make sure to have at least 15% more items than you want to have in your final version.

Our neurolinguistic group here at the University of Groningen has created an item list of nouns and verbs and we tested their suitability. We will work with that set of pictures in this course. Moreover, we will offer them as an open source and welcome other groups to work with our materials.