Phonological Mismatch Initiates Inhibitory Control of Failed Predictions During Sentence Comprehension

False predictions during sentence comprehension are a frequent phenomenon. Recent research has shown that in highly constrained sentences, inhibitory mechanisms are engaged to suppress false predictions. However, little is known about what specifically leads to this inhibition. Sixty-eight monolingual English-speaking adults participated in the current study to examine what triggers the inhibition of predicted words. We utilized the cross-modal lexical priming paradigm. Participants performed a visual lexical decision task (LDT) immediately after listening to incomplete sentences and sentences containing violations that did not match their prediction. In Experiment 1, violation sentences ended in pseudowords that contained a phonological mismatch, while in Experiment 2, violation sentences ended in environmental sounds that contained a semantic mismatch. To examine whether the predicted word was inhibited in each case, we compared LDT reaction times (RTs) to predicted words across sentence conditions. Results showed that LDT RTs to the predicted word were significantly slower after pseudowords, but not environment sounds. Taken together with previous work, this suggests that lexico-semantic information may not be required to trigger inhibition. While pseudoword violations-stimuli that resemble real words but lack meaning-inhibit false predictions, semantic mismatch alone may be insufficient to elicit inhibition.