A Picture May Be Worth 1,000 Words, but Is It Worth a Letter? Examining Whether the Choice of Label Affects the Perception of Speech Sounds.

Abstract

Researchers often use identification or goodness rating tasks to assess speech perception for different populations. These tasks provide useful information about a listener's willingness to accept a range of acoustically variable stimuli as belonging to the same category and also about assessing how stimuli that are labeled the same may not be perceived as equally good versions of a particular speech sound. Many methodological aspects of these simple tasks have been tested, but one aspect that has not is the choice of label. In this study, we examine response patterns to images versus letters, as studies with different populations (children vs. adults) or different methods (typical behavioral study vs. visual world paradigm) may vary in the type of label used. Eighty-one adult listeners completed phoneme identification and goodness ratings tasks with either images of response options (a picture of a bear and a picture of a pear) or with letter labels (a capital B and P). The results suggest that choice of label does not alter performance within the tasks studied here. In addition, the results did show the expected finding that the slope of the response curve is steeper in an identification task than in a goodness rating task. These results suggest that it is possible to compare across studies that use different response options, a benefit to research and practice because letter labels can be used for nonimageable words and nonwords, whereas images may be best used for participants who are younger or have poorer reading skills.