Preparing to switch languages versus preparing to switch tasks: Which is more effective?

A substantial literature relates task-set control and language selection in bilinguals-with "switching" paradigms serving as a methodological "bridge." We asked a basic question: Is preparation for a switch equally effective in the two domains? Bilinguals switched between naming pictures in one language and another, or between the tasks of naming and categorizing pictures. The critical trials used for comparing the two kinds of switching were identical in all respects-task (naming), stimuli, responses-except one: whether the shape cue presented before the picture specified the language or the task. The effect of preparation on the "switch cost" was examined by varying the cue-stimulus interval (CSI; 50/800/1,175 ms). Preparation for a task switch was more effective: Increasing the CSI from 50 to 800 ms reduced the reaction time task switch cost by ∼63% to its minimum, but the language switch cost only by ∼24%, the latter continuing to reduce with further opportunity for preparation (CSI = 1,175 ms). The switch costs in the two domains correlated moderately ( = .36). We propose that preparation for a language switch is less effective, because (a) it must preemptively counteract greater interference during a language switch than during a task switch, and/or (b) lexical access is less amenable to "top-down" control than (components of) task-set. We also investigated the associations between stimuli and the language (or task) where they were last encountered. Associative history influenced performance-but similarly for switches and repetitions-indicating that stimulus-induced associative retrieval of language (or task-set) did not contribute to switch costs. (PsycInfo Database Record (c) 2021 APA, all rights reserved).