Languages differ in how events are described, but little is known about how semantics interacts with online event processing. This study targets this question examining placement events in Swedish. Swedish has three obligatory placement verbs for events where objects have support from below: sätta ’set’, ställa ’stand’, and lägga ’lay’. Swedish lacks a superordinate general term like English put (Gullberg & Burenhult, 2011). For every placement event the verb choice depends on object properties, and the direction of the object’s extension from the ground. We use event-related potentials (ERPs) and appropriateness ratings of verb usage to investigate the interaction between verb semantics and event properties. Typically violations of semantic congruency positively affect the amplitude of the N400 (Kutas & Hillyard, 1980). Studies also report a centro-parietal positivity (P600) when real-world knowledge is violated and verbs are incongruous to preceding contexts (Kuperberg, 2007, for a review). Previous ERP studies of visually presented images or movies of actions and events have reported an N400 followed by a P600 when the function of an object is violated (e.g., using a screwdriver as a key, Bach, et al., 2009; Balconi & Caldiroli, 2011).
Method: Native speakers (N = 24, 18-35 years) watched still images of placement events followed by sentences visually presented word by word. Sentences described the depicted events while ERPs were recorded and time-locked to the placement verbs. Participants also did an appropriateness rating offline. Object properties (Base/Without base), symmetry (Sym/Asym), and orientation from the ground (Vertical/Horizontal) were varied and sentences with the three different placement verbs were combined with each image in a cross-subject design.
Results: Base was important for appropriateness ratings of verb usage with symmetric objects while orientation was important for asymmetric objects. In contrast, there were no ERP effects to base (Base/Without) for symmetric objects. Asymmetric-base objects showed increased N400s and P600s with verbs incongruent with the depicted events (orientation, e.g., ‘lay’ with vertical glass). Asymmetric-Without base elicited an increased P600 when verbs were incongruent to depicted events when horizontally oriented (e.g., ‘set’ with horizontal avocado), but an increased N400 when verbs were incongruent to the atypical vertical placement of the objects (e.g., ‘lay’ with a vertical avocado).
Discussion: Results showed an increased amplitude of both ERP effects (N400/P600) when placement verbs were incongruent with typical placement scenarios of objects that in the real-world are placed vertically or horizontally (Asymmetric-Base, e.g., a candle; cf. Bach et al., 2009). However, for objects without a base the anterior negativity was increased with a mismatch between the verb and the presented images (the depicted events), while the P600 increased for mismatches between the verb and typical real-world events. These results suggest the anterior N400 and the P600 indeed index different relationships with event processing as previously suggested for images (Sitnikova, et al., 2008). Our results agree with previous studies suggesting that the processing of verb meaning in language cannot be separated from knowledge of object handling in the real world (cf., Van Berkum, et al., 2008).