Detection of Errors by Skilled Typists

Abstract Skilled typists are able to detect and correct many errors which they make in copy text, even when they cannot see their copy; (Long 1976. West, 1967). The present investigation shows that they can also do this when they see neither their copy nor the keyboard they use; and that when they detect that they have made errors they arc usually also able to specify precisely what these have been. Typists sometimes make one, or even two additional correct keystrokes before pausing to signal that they have committed an error. They are nevertheless sometimes still able to report precisely what these errors were When copy obtained from a typewriter using mechanical linkages is examined, the density of impressions of different characters may be taken to indicate how hard particular keys have been struck. A second experiment shows that errors, more frequently than correct responses, are executed with light keystrokes. This, in line with previous work by Megaw (1972). suggests that typists sometimes become aware that they are about to make an error before they complete the keystroke implementing it. They may then attempt to ‘pull back’ incorrect keystrokes, producing fainter impressions on copy The probability that typists will subsequently signal detection of such ‘ faint’ (or‘ pulled’) errors was found to be significantly greater than the probability that they will subsequently signal their detection of other errors, made with ‘firm’ (i.e. ‘unpulled’) keystrokes These data, related to earlier findings by Diehl and Seibel (1962), Klemmer (1971). Long (1975, 1976a,b) and Seibel (1972) offer little support for the assumption that skilled typists may monitor their emitted responses in ‘units’ of greater than one keystroke. They rather suggest that, even al very high response rates (e.g. 172ms or faster), typists read copy, programme and emit responses and evaluate each of their responses for accuracy as concurrent, or partially overlapping activities.