Re-examining the SLIP Task: Completely Lexical vs. Completely Non-Lexical

Abstract

For years, the SLIP task, a methodology used to elicit speech errors in a laboratory setting, has added invaluable evidence in developing models of language production. It has lead to the finding that speech errors more commonly result in real words than in non-words, called the ‘lexical bias effect’. In turn, it has been used to support the idea of a self-monitor that monitors for speech errors, and is more likely to let real words slip past than non-words, as well as a theory of feedback between levels of processing, which increases the number of lexical errors, and more recently a theory that combines the self-monitor and feedback. However, past usage of the SLIP task has never directly compared results for a completely lexical context to that of a completely non-lexical context. The purpose of the present study is to fill this gap in the literature by using two levels of context, one block where no non-words are presented, and another where no real words are presented, as well as an outcome condition, where the intended error outcome of a target is either lexical or non-lexical. 52 participants read word pairs and the number and variety of speech errors and other responses they made were recorded. The results found no significant evidence for a difference in the number of exchanges between any conditions, however there were effects for the number of corrections made and for the number of times participants failed to respond. These results can add support to existing models of language production, particularly in the role of feedback in the selection of lexical concepts after an error is detected by the monitor.