Understanding Speech

Production Studies

Looking closely, speaking is quite a complex task. We have to think of what we want to talk about, plan an appropriate way of saying what we want to communicate, select the right words, build a syntactic structure and put the words into the right place, and finally coordinate the articulatory muscles to produce the planned words and sentences in the correct order without spluttering. Most of these processes happen automatically and within a few hundred milliseconds.

Speaking under noisy circumstances can be a bit more challenging. Speakers normally tend to counteract concurrent noise by changing pitch and intensity of their voice. We are interested in the possible effects (automatic or strategic) that adverse communication settings and sensorineural perception problems of speakers with hearing impairment might have on central cognitive processes implied in planning and formulating sentences. Do speakers perhaps "simplify" their language under noise?

In order to test this question empirically, we developed different experimental methods to elicit spontaneous and controlled speech under different acoustic background conditions. In our language production experiments, speakers are presented with a variety of materials, mostly pictures, and we ask the participants to describe the materials in different ways.

For example, we recently conducted a study where a couple of subjects with and without hearing impairment were asked to describe simple drawings from the OLACS corpus in one sentence. The acoustic background was manipulated using the communication and acoustics simulator (CAS) facility at the Oldenburg "Haus des Hörens" audiological research centre.