Halting in Single Word Production: A Test of the Perceptual Loop Theory of Speech Monitoring

L. Robert Slevc and Victor S. Ferreira
University of California, San Diego

The concept of a pre-articulatory editor or monitor has been used to explain a variety of patterns in the speech error record. The _perceptual loop theory_ of monitor function (Levelt, 1983) claims that inner speech is monitored by the comprehension system, which detects errors by comparing the comprehension of formulated utterances to originally intended concepts. This implies that the monitor should be sensitive primarily to semantic information, and that its main task is to decide if the comprehended meaning of a formulated utterance is close enough to the intended meaning so that it
can be articulated. Alternatively, a pre-articulatory monitor might detect errors via a comparison at a phonological level of representation, in which case the monitor should be sensitive primarily to phonological information. Or, as suggested by evidence such as the _mixed error effect_ (e.g. Dell & Reich, 1981), the monitor might be sensitive to both semantic and
phonological information. Monitor-based explanations of such effects assume that errors that are similar to the intended production are less likely to be detected by the monitor - and therefore relatively more likely to be produced - than errors that are dissimilar to the intended production.

A defining feature of perceptual loop theories is that inner speech is monitored by the same comprehension system as external speech; thus it should be possible to simulate monitor function by having speakers alter their production based on externally presented speech signals (e.g. Oomen & Postma, 2002). Three experiments used this logic to assess the perceptual loop theory of monitor function by looking at differences in the ability to inhibit word production in response to _stop signals_ - words - that varied in terms of their semantic or phonological similarity to the intended word.

Subjects named pictures and sometimes heard (Experiment 1) or saw (Experiments 2 and 3) a word different from the picture name, which served as a signal to halt their naming response. When the signal was phonologically similar to the picture name, subjects had significantly more difficulty halting speech than when the signal was phonologically dissimilar
to the picture name. Semantic similarity had no effect on halting performance. Ability to inhibit words appears to be more sensitive to phonological than to semantic similarity, suggesting that errors are detected and avoided by comparing at a phonological rather than at a semantic level.

References

Dell, G. S., & Reich, P. A. (1981). Stages in sentence production: An analysis of speech error data. _Journal of Verbal Learning and Verbal Behavior_, 20, 611-629.

Levelt, W. J. M. (1983). Monitoring and self-repair in speech. _Cognition_, 14, 41-104.

Oomen, C. C. E., & Postma, A. (2002). Limitations in processing resources and speech monitoring. _Language and Cognitive Processes_, 17 (2), 163-184.