Twisting tongues to test for conflict-monitoring in speech production

Abstract

A number of recent studies have hypothesized that monitoring in speech production may occur via domain-general mechanisms responsible for the detection of response conflict. Outside of language, two ERP components have consistently been elicited in conflict-inducing tasks (e.g., the flanker task): the stimulus-locked N2 on correct trials, and the response-locked error-related negativity (ERN). The present investigation used these electrophysiological markers to test whether a common response conflict monitor is responsible for monitoring in speech and non-speech tasks. Electroencephalography (EEG) was recorded while participants performed a tongue twister (TT) task and a manual version of the flanker task. In the TT task, people rapidly read sequences of four nonwords arranged in TT and non-TT patterns three times. In the flanker task, people responded with a left/right button press to a center-facing arrow, and conflict was manipulated by the congruency of the flanking arrows. Behavioral results showed typical effects of both tasks, with increased error rates and slower speech onset times for TT relative to non-TT trials and for incongruent relative to congruent flanker trials. In the flanker task, stimulus-locked EEG analyses replicated previous results, with a larger N2 for incongruent relative to congruent trials, and a response-locked ERN. In the TT task, stimulus-locked analyses revealed broad, frontally-distributed differences beginning around 50 ms and lasting until just before speech initiation, with TT trials more negative than non-TT trials; response-locked analyses revealed an ERN. Correlation across these measures showed some correlations within a task, but little evidence of systematic cross-task correlation. Although the present results do not speak against conflict signals from the production system serving as cues to self-monitoring, they are not consistent with signatures of response conflict being mediated by a single, domain-general conflict monitor.

Keywords: ERP/EEG; N2; cognitive control; error-related negativity (ERN); flanker task; monitoring; speech production; tongue twister.

FIGURE 1

Behavioral results across both tasks. (A) Mean error rates and reaction times for congruent and incongruent trials in the flanker task. (B) Mean error rates for three different types of errors in the tongue twister task as well as the proportion of errors that were self-corrected across each tongue twister condition. Error bars correspond to standard error across participants.

FIGURE 2

Stimulus-locked ERPs and scalp topographies of differences for both tasks. Trials were baseline-corrected -100 ms before stimulus onset. (A) Flanker task – congruent and incongruent trials ERPs at electrode FCz as well as scalp topographies of the mean amplitude (upper)and current source density (lower) differences between 250 and 325 ms. (B) Tongue-twister and non-tongue twister trials as well as scalp of the mean amplitude (upper) and current source density (lower) differences in 100 ms windows between 50 and 450 ms.

FIGURE 3

Response-locked ERPs and scalp topographies of differences for both tasks. ERPs are plotted for electrode FCz, and were baseline-corrected -300 to -200 ms before response onset. (A) ERPs for errors and correct trials at FCz in the flanker task and the scalp topographies of the mean amplitude (upper) and current source density (lower) differences between errors and correct trials between 0 and 100 ms after a response. (B) ERPs for errors and correct trials for the tongue twister task at FCz as well as scalp topographies of the mean amplitude (upper) and current source density (lower) differences between errors and correct trials between 0 and 100 ms after a response.