Speech recognition-can user problems be predicted?

Some of the existing theories which form the basis for a viable account of automatic speech recognition (ASR) use are discussed. A number of factors which fall under the general umbrella of user problems are considered in terms of vocabulary/dialogue design, task performance and extraneous factors. The models of ASR use are discussed with reference to psycholinguistics, multiple resources, working memory and the principles of compatibility, derived from ergonomics. >