Modification of prosodic cues when an interlocutor cannot be seen : the effect of visual feedback on acoustic prosody production

Speakers alter the way they produce speech according to the communicative situation. Changes are made to enhance the efficiency of information transmission. For instance, when in noisy environments, people speak loudly and produce more energy in higher frequencies (the Lombard effect). This study investigated whether a change in the visual conditions associated with communication would also lead to modification in speech production. More specifically, it examined if auditory prosody would be affected by whether the speaker could see the interlocutor or not. In the experiment, two types of prosodic contrasts were included. The first was ‘prosodic focus’ used by speakers to enhance the perceptual salience of an item. The second was ‘prosodic phrasing’ which refers to the phrasing of a sentence as a question without using an interrogative pronoun. Four speakers were recorded while completing a dialog exchange task in which the interlocutor could or could not be seen. The results showed that the corner-most vowels recorded in narrow focus and echoic question contexts were produced over longer durations and with a greater vowel space (reflected by greater vowel triangle area and vowel triangle dispersions) relative to broad focused renditions across both interaction conditions. With the exception of intensity, no other acoustic or spectral properties appeared to be enhanced at the phonemic level when the interlocutor was not visible to the speaker. This may be due to prosody affecting the utterance at more global levels (e.g., word and utterance levels), rather than at the localized vowel level. That is, modifications may be seen between interactive conditions in terms of pitch contours, pre-focal shortening and intensity profiles when examined across the whole utterance