Comparison of three sentence‐level tests for evaluating audiovisual performance of subjects using a cochlear implant

Abstract

Postlingually deafened adults implanted with the Symbion cochlear prosthesis have been tested for audiovisual reception of isolated sentences. Three different sets of sentence materials have been used: (a) CID everyday sentences, including live and recorded versions; (b) CUNY sentences developed by Boothroyal et al. (CUNY Rep. RCI10), presented from videodisc; and (c) IEEE/Harvard sentences, from in‐house recordings. The tests were conducted in three modes: vision (speechreading) alone, sound alone, and sound plus vision. No feedback was given and, except for the live‐voice CID tests, sentences were never repeated for a given subject. Results indicate that all three materials are sensitive to detecting whether the cochlear implant provides a benefit to speechreading, but the CID and CUNY sentences frequently exhibit ceiling effects that limit their ability to differentiate the degree of benefit. The more challenging IEEE/Harvard sentences have reduced ceiling effects and provide other useful performance measures. Specifically, the benefit to speechreading provided by the prosthesis at speech‐to‐noise ratios of 0–3 dB is roughly one‐half that provided in quiet. [Work supported by NIH.]