Different Electrophysiological Signatures of Similarity-induced and Stroop-like Interference in Language Production

Contextual similarity between targets and competitors, whether semantic or phonological, often leads to behavioral interference in language production. It has been assumed that resolving such interference relies on control processes similar to those involved in tasks such as Stroop. This article tests this assumption by comparing the electrophysiological signatures of interference resulting from a contextual similarity versus a Stroop-like manipulation. In blocks containing two items, participants repeatedly named pictures that were semantically related, phonologically related, or unrelated (contextual similarity manipulation). In straight blocks, the pictures were named by their canonical names. In reverse blocks, participants had to reverse the names (Stroop-like manipulation). Both manipulations led to behavioral interference, but with different electrophysiological profiles. Whole-scalp stimulus-locked and response-locked analyses of semantic and phonological similarity pointed to a system with global modularity with some degree of cascading and interactivity, whereas the effect of phase reversal was sustained and of the opposite polarity. More strikingly, a representational similarity analysis showed a biphasic pattern for Stroop-like reversal, with earlier higher similarity scores for the reverse phase flipping into lower scores ~500 msec poststimulus onset. In contrast, contextual similarity induced higher similarity scores up to articulation. Finally, response-locked mediofrontal components indexing performance monitoring differed between manipulations. Correct response negativity's amplitude was lower in the phonological blocks, whereas a pre-correct response negativity component had higher amplitude in reverse versus straight blocks. These results argue against the involvement of Stroop-like control mechanisms in resolving interference from contextual similarity in language production.