Monitoring speech production and comprehension: Where is the second-language delay?

Research on error monitoring suggests that bilingual Dutch–English speakers are slower to correct some speech errors in their second language (L2) as opposed to their first language (L1). But which component of self-monitoring is slowed down in L2, error detection or interruption and repair of the error? This study charted the time course of monitoring in monolingual English speakers and bilingual Dutch–English speakers in language production and language comprehension, with the aim of pinpointing the component(s) of monitoring that cause an L2 disadvantage. First, we asked whether phonological errors are interrupted more slowly in L2. An analysis of data from three speech error elicitation experiments indeed showed that Dutch–English bilinguals were slower to stop speaking after an error had been detected in their L2 (English) than in their L1 (Dutch), at least for interrupted errors. A similar L2 disadvantage was found when comparing the L2 of Dutch–English bilinguals to the L1 of English monolinguals. Second, monolingual English speakers and bilingual Dutch–English speakers performed a picture naming task, a production monitoring task, and a comprehension monitoring task. Bilingual English speakers were slower in naming pictures in their L2 than monolingual English speakers. However, the production monitoring task and comprehension monitoring task yielded comparable response latencies between monolinguals in their L1 and bilinguals in their L2, indicating that monitoring processes in L2 are not generally slower. We suggest that interruption and repair are planned concurrently and that the difficulty of repairing in L2 triggers a slow-down in L2 interruption.