Context effects in language comprehension: The role of emotional state and attention on semantic and syntactic processing

Abstract

Semantics and syntax are core components of language. The prevailing view was that processing of word meaning and syntactic processing happens in isolation from other systems. In light of proofed interactions between language and other systems, especially with perception, action and emotion, this view became untenable. This article reviews Event-related potential studies conducted at the Donders Centre for Cognition exploring the interplay between language comprehension and a person's emotional state. The research program was aimed at an investigation of the online effects of emotional state on semantic processing and syntactic processing. To this aim we manipulated mood via film fragments (happy vs. sad) before participants read neutral sentences while their EEG was recorded. In Part 1, it is shown that mood impacts online semantic processing (as indicated by N400) and the processing of syntactic violations (as indicated by P600). Part 2 was directed at a further determination of the mechanisms underlying these interactions. The role of heuristics was examined by investigating the effects of mood on the P600 to semantic reversals. The results revealed that mood affects heuristic processing. The next step consisted of an assessment of the role of attention, in the mood-by-semantics and mood-by-syntax interaction. This was accomplished by recording EEG while manipulating attention via task next to emotional state. Participants performed a semantic or syntactic judgment task vs. a letter-size judgment task. The main ERP results were as follows: (i) attention interacts with the mood effect on semantic processing and syntactic processing, respectively, (ii) the effects of mood on semantic processing and syntactic processing are reliable, and (iii) the mood effects on semantic processing are not fixed but context-dependent. In Part 3 the effects of mood on the processing of script knowledge and general world knowledge are presented. Part 4 closes with a discussion of the mechanisms involved in the mood-by-language interactions and recommendations for future research. Regarding the underlying mechanism we propose that heuristics based on semantic expectancies or syntactic expectancies play a key role in the mood-by-language interactions. The results support the view that language takes place in continuous interaction with other (non-language) systems.

Keywords: N400 cloze effect; P600 effect; abstract symbol theories; attention; embodied theories; emotion; heuristic processing; language comprehension.

FIGURE 1

Visual sketch of the mood induction procedure.

FIGURE 2

Modified Figure 2 from Chwilla et al. (2011). Study 1: Grand ERP waveforms for the happy mood condition, time-locked to the onset of the critical noun superimposed for the two levels of cloze probability (high, low) for all midline sites and a representative subset of lateral sites. N400 was measured in the time window from 300 to 500 ms. Negativity is plotted upwards.

FIGURE 3

Modified Figure 3 from Chwilla et al. (2011). Study 1: Grand ERP waveforms for the sad mood condition, time-locked to the onset of the critical noun superimposed for the two levels of cloze probability (high, low) for all midline sites and a representative subset of lateral sites. N400 was measured in the time window from 300 to 500 ms. Negativity is plotted upwards.

FIGURE 4

Modified Figure 2 from Vissers et al. (2010). Study 2: Grand ERP waveforms for the happy mood condition, time-locked to the onset of the critical verb superimposed for the two levels of syntactic correctness for all midline sites and a representative subset of lateral sites. The dashed rectangles indicate the time window (600–800 ms) in which P600 amplitude was measured. Negativity is plotted upwards.

FIGURE 5

Modified Figure 3 from Vissers et al. (2010). Study 2: Grand ERP waveforms for the sad mood condition, time-locked to the onset of the critical verb superimposed for the two levels of syntactic correctness for all midline sites and a representative subset of lateral sites. The dashed rectangles indicate the time window (600–800 ms) in which P600 amplitude was measured. Negativity is plotted upwards.

FIGURE 6

Modified Figure 2 from Vissers et al. (2013). Study 3: Grand ERP waveforms for the happy mood condition, time-locked to the onset of the critical verb superimposed for the two levels of condition for all midline sites and a representative subset of lateral sites. P600 was measured in the time window from 600 to 800 ms. Negativity is plotted upwards.

FIGURE 7

Modified Figure 3 from Vissers et al. (2013). Study 3: Grand ERP waveforms for the happy mood condition, time-locked to the onset of the critical verb superimposed for the two levels of condition for all midline sites and a representative subset of lateral sites. P600 was measured in the time window from 600 to 800 ms. Negativity is plotted upwards.

FIGURE 8

Modified Figure 8 from Verhees et al. (2015). Study 4: Grand ERP averages at Pz for both mood and task conditions.

FIGURE 9

(A,B) Modified Figure 6 from Verhees et al. (2015). Comparison of the mean P600 amplitudes (incorrect-correct) for the happy and sad mood condition (A) for the syntactic judgment task and (B) for the letter-size judgment task.

FIGURE 10

Study 5: Grand ERP waveforms at Cz for both mood and task conditions (Chwilla and Tromp, 2013).

FIGURE 11

Study 5: Comparison of the mean N400 amplitudes (implausible-plausible) for the happy and sad mood condition, for the semantic task for the electrodes of the right hemisphere (Chwilla and Tromp, 2013).

FIGURE 12

Grand ERP waveforms for the midline sites to the critical word of the script-related vs. unrelated triplets separately for the happy mood condition (A) and the sad mood condition (B) (Menn and Chwilla, 2017).