The N400 and Late Positive Complex (LPC) Effects Reflect Controlled Rather than Automatic Mechanisms of Sentence Processing

Abstract

This study compared automatic and controlled cognitive processes that underlie event-related potentials (ERPs) effects during speech perception. Sentences were presented to French native speakers, and the final word could be congruent or incongruent, and presented at one of four levels of degradation (using a modulation with pink noise): no degradation, mild degradation (2 levels), or strong degradation. We assumed that degradation impairs controlled more than automatic processes. The N400 and Late Positive Complex (LPC) effects were defined as the differences between the corresponding wave amplitudes to incongruent words minus congruent words. Under mild degradation, where controlled sentence-level processing could still occur (as indicated by behavioral data), both N400 and LPC effects were delayed and the latter effect was reduced. Under strong degradation, where sentence processing was rather automatic (as indicated by behavioral data), no ERP effect remained. These results suggest that ERP effects elicited in complex contexts, such as sentences, reflect controlled rather than automatic mechanisms of speech processing. These results differ from the results of experiments that used word-pair or word-list paradigms.

Figure 1

Waveforms and spectrograms of three sentences on the left (Si tu vas jouer dehors, n’oublie pas ton manteau.), middle (Ils ont visité la France pendant les vacances.), and right panel (La maîtresse a recopié l’exercice sur le tableau.) in the four degradation conditions. The bottom panel shows the no degradation condition (DL0), the second panel from the bottom shows the “low” degradation condition (DL1), the second panel from the top shows the “medium” degradation condition (DL2), and the top panel shows the “strong” degradation condition (DL3). Waveforms vertical scale range from 0 to −∞ (Unit: dB). Spectrogram vertical scale range from 0 kHz to 5 kHz (linear scale). Waveform and spectrogram horizontal time axis range from 0 s to the sentence duration, i.e., 2.08 s for the left panel, 2.57 s for the middle panel, and 2.70 s for the right panel.

Figure 2

Sequence of stimulus presentation.

Figure 3

Grand averaged event-related potentials (ERPs) to incongruent targets (thick line) and congruent targets (thin line) at each degradation level (no degradation: DL0; “low” degradation: DL1; “medium” degradation: DL2; “strong” degradation: DL3) (N = 20 participants, vertical unit: µV with negativity upward, horizontal unit: ms).

Figure 4

Grand averaged subtraction waveforms between ERP to incongruent targets and ERP to congruent targets at each degradation level (no degradation: DL0; “low” degradation: DL1; “medium” degradation: DL2; “strong” degradation: DL3) (N = 20 participants, vertical unit: µV with negativity upward, horizontal unit: ms).