Neural signatures of response planning occur midway through an incoming question in conversation

Abstract

A striking puzzle about language use in everyday conversation is that turn-taking latencies are usually very short, whereas planning language production takes much longer. This implies overlap between language comprehension and production processes, but the nature and extent of such overlap has never been studied directly. Combining an interactive quiz paradigm with EEG measurements in an innovative way, we show that production planning processes start as soon as possible, that is, within half a second after the answer to a question can be retrieved (up to several seconds before the end of the question). Localization of ERP data shows early activation even of brain areas related to late stages of production planning (e.g., syllabification). Finally, oscillation results suggest an attention switch from comprehension to production around the same time frame. This perspective from interactive language use throws new light on the performance characteristics that language competence involves.

Figure 1. Grand average ERPs for a representative electrode (Pz).

Critical words are always indicated by red solid lines and equivalent positions by blue dashed lines. Topographical plots are given for the N400 time window (top: 300–500 ms) and two time-windows for the positivity (middle: 600–900 ms; bottom: 900–1200 ms). Colors indicate T-values. Electrodes that show a significant effect in more than 70% of the time window are highlighted in white.

Figure 2. Localizations of the positivities in the ERPs (600–1100 ms) of the main experiment.

Localizations for the effects at TL1 (EARLY vs. LATE) are shown on the left and for that at TL2 (LATE vs EARLY) are shown on the right. Only the left hemisphere is shown since activations occurred mainly in this hemisphere. Colors indicate T-values.

Figure 3. Time-frequency results for a representative electrode (left posterior, see middle).

Colors in all plots indicate the relative difference between raw power in the relevant conditions. In the time-frequency plots, the relative difference is given in transparent colors with the statistically significant cluster overlaid in opaque colors. Topographical plots are given for appropriate time windows and for the 10–12 Hz range. Electrodes that are significant in the time window are highlighted in white.

Figure 4. Localizations of reduced alpha power for the critical words in the main experiment.

Localizations of the reduced alpha power at TL1 (EARLY vs. LATE) are shown at the top and those at TL2 (LATE vs EARLY) are shown at the bottom. Colors indicate T-values.