Use of a Sequential Strategy in Intelligibility Testing

Abstract

In a recent experiment involving intelligibility testing, the response curve (intelligibility function) was found to flatten off sharply at moderately high intelligibilities. A sequential strategy of the up-down-transformed-response (UDTR) type was used in order to concentrate observations in the symmetric region of the curve. The transformed-response curve showed the departure from symmetry at a much higher intelligibility level. Precision of estimation for the 50% level was found to be higher than for the simple up-and-down procedure. The technique is designed for situations in which it is difficult or impossible to vary the step size.