Visual speech fills in both discrimination and identification of non-intact auditory speech in children

Abstract

To communicate, children must discriminate and identify speech sounds. Because visual speech plays an important role in this process, we explored how visual speech influences phoneme discrimination and identification by children. Critical items had intact visual speech (e.g. bæz) coupled to non-intact (excised onsets) auditory speech (signified by /-b/æz). Children discriminated syllable pairs that differed in intactness (i.e. bæz:/-b/æz) and identified non-intact nonwords (/-b/æz). We predicted that visual speech would cause children to perceive the non-intact onsets as intact, resulting in more same responses for discrimination and more intact (i.e. bæz) responses for identification in the audiovisual than auditory mode. Visual speech for the easy-to-speechread /b/ but not for the difficult-to-speechread /g/ boosted discrimination and identification (about 35-45%) in children from four to fourteen years. The influence of visual speech on discrimination was uniquely associated with the influence of visual speech on identification and receptive vocabulary skills.

Figure 1

Baseline results for the auditory mode in the children grouped according to age. Performance was quantified by percent of same responses to the different pairs (e.g., bæ:/−b/æ perceived as same) for discrimination and percent of correct consonant onset responses (e.g., /−b/æz perceived as bæz) for identification. Results—which did not differ across the age groups, the tasks, or the onsets—are consistent with our criterion for excising the onsets and yield a stable baseline for assessing the Visual Speech Fill-in Effect. Error bars are ± one standard error of the mean.

Figure 2

Visual Speech Fill-in Effect, VSFE (i.e., difference in performance for the AV - auditory modes) in children grouped according to age. Discrimination was quantified by percent of same responses to the different pairs (e.g., bæ:/−b/æ perceived as same); identification was quantified by percent of correct consonant onset responses (e.g., /−b/æz perceived as bæz). Results showed a large VSFE (with significant age and task differences) for /b/ but not for /g/. Error bars are ± one standard error of the mean.