ESCA Workshop on Spoken Dialogue Systems Vigsø, Denmark May 30 - June 2, 1995

Speech versus Keying in the Human-Computer Interface

Robert I. Damper, M. A. Tranchant, S. D. Wood

Image, Speech and Intelligent Systems (ISIS) Research Group, Department of Electronics and Computer Science, University of Southampton, Southampton, UK

Speech input is frequently claimed to offer great benefits in human-computer interaction. A study conducted by Poock in the early 1980s, for instance, has purportedly shown a very significant superiority for speech over keying. We have previously argued, however, that this finding was an artifact of a methodological flaw - specifically that the command vocabulary was chosen to suit the requirements of speech input and made little or no concession to the requirements of keying. In earlier experiments designed by us to overcome this putative flaw - by using abbreviated rather than full command strings in the keying condition - the claimed superiority disappeared. There were, however, other differences between our experiments and Poock's, so that other interpretations of these findings are possible. Most obviously, Poock7s subjects carried out a concurrent, secondary task while ours did not. Since speech input is generally considered to be advantageous in such situations, this difference is potentially important. Also, we did not replicate Poock's original experimental condition, so that no direct comparison of results was possible.

In this paper, we describe new experiments - again modelled on those of Poock - in which speech input, full command-string keying and abbreviated command-string keying are compared under conditions of concurrent tasking. We find that speech input is no faster and enormously more error-prone than abbreviated keying, but allows somewhat more of the secondary task to be completed. Full keying has no advantages whatsoever, confirming the methodological flaw in Poock's work. Our subjects perform less well on speech input than did his under equivalent conditions: we attribute this to their generally lower familiarity with speech data entry.

Full Paper

Bibliographic reference.  Damper, Robert I. / Tranchant, M. A. / Wood, S. D. (1995): "Speech versus keying in the human-computer interface", In SDS-1995, 229-232.