The visual equivalency of the Word Intelligibility by Picture Identification Test
Nicholson, Nancy K.
MetadataShow full item record
The Word Intelligibility by Picture Identification Test (WIPI), developed by Ross and Lerman (1970), is widely used to test the speech understanding ability of children. It was designed for administration not only in the auditory mode but also with visual and combined auditory-visual cues. While the four test lists that comprise the WIPI have been shown to be highly equivalent when administered auditorily, visual realtionships have not been established. The present investigation was designed to explore the visual equivalency of the WIPI Test lists as well as the effects of speaker differences and order of presentation on viewer performance. The materials for this investigation consisted of the four 25-word test lists of the WIPI Test. The items were videotaped in black and white under controlled lighting conditions using four female speakers, age 22 to 32 years. All reported having normal hearing and spoke General American English with no articulation abnormalities. They were instructed to read each stimulus item using the carrier phrase, "Show me _____", once every 15 seconds with normal vocal effort. Articulation of each sentence w'as initiated and terminated with a slightly open, relaxed mouth position. Following the videotaping, the sixteen taped lists (four lists from each of four speakers) were grouped into four different randomizations and were edited onto cassette videotapes with a four-second interstimulus interval. Forty students from Northern Illinois University served as subjects for this investigation. None had any previous formal training in speechreading, and all demonstrated normal hearing and vision. The stimulus tapes were viewed by four groups of ten subjects each during four separate sessions. Answer sheets containing the six word choices for each WIPI item were given to the subjects who were instructed to circle the item that was visually perceived. The 16 response sheets from each of the ifO subjects were initially scored according to the total number of words correctly identified. The mean scores and standard deviations for each speaker, test list, and randomization were then computed. In order to determine if there was a significant difference between the visibility of the four WIPI Test lists and to discover whether this visibility was affected by speaker and randomization differences, the data were subjected to an analysis of variance, three-factor repeated measures design (Winer, 1971). A post-hoc Newman-Keuls Multiple Range Test (Winer, 1971) was also performed. Results of the visual observations indicated that, first of all, the order of list presentation did not affect subject perfromance. Second, analysis of the data indicated that the productions of three of the four speakers were visually equivalent while one of the four resulted in statistically different scores. This difference, however, was equivalent to less than one word which was not clinically significant, finally, data analysis indicated that WIPI List 1 and List 2 were statistically different from List 3 and List 4, but in this case, the difference was equivalent to less than a two-word variation. Interpretation of these data leads to the conclusion that the four WIPI Test lists are visually similar enough to be used for general clinical purposes. However, for more accurate results, List 1 and List 2 should be used for visual and auditory-visual comparison since the basic visibility of these lists is equal. The same is true for Lists 3 and With auditory equivalencies already established, the WIPI Test can now successfully be used as a measurement instrument to probe the auditory, visual, and combined auditory-visual speech perception of a clinical population.