Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
The role of animacy in computing thematic relations during online sentence comprehension
Poster E39 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall
Chia-Hsuan Liao1, Hsiang-Ching Chiu1; 1National Tsing Hua University
INTRODUCTION: Although unpredicted words usually elicit a larger N400 response relative to predicted ones (Kutas & Federmeier, 2011), results from “role-reversed” sentences (“The millionaire that the servant fired…”) are controversial. While the absence of N400 responses to role reversal situations have been replicated in many languages (Kolk et al. 2003; Kuperberg et al., 2007), some studies do observe an N400 effect (Bornkessel-Schlesewsky et al. 2011). The current study tested whether animacy of the noun phrases could modulate the N400 responses in role-reversal anomalies. We hypothesized that when the noun phrases have different animacy features, it would be easier to evaluate their thematic relations to update predictions. In particular, (at least in Mandarin) sentences with an inanimate subject and an animate object would be more marked compared with the other way around, and we expected their processing profiles to be different. EXPERIMENT: 28 native speakers of Mandarin read sentences word by word during EEG recording. 120 stimuli were presented. We used the SOV ba construction in Mandarin, with the morpheme ba providing reliable cues about the thematic roles of the noun phrases before the presence of the verb. A two (Thematic role assignment: Canonical vs. Reversal) by two (Animacy feature: Animate-Inanimate vs. Inanimate-Animate) experiment was set up to approach the question. We’ve kept the semantic relatedness between the subject and object identical between the two types of Animacy features. The predictability of the target verb in Canonical conditions, regardless of the Animacy features, was 52% (Animate-Inanimate: The pitcher ba the ball threw out, meaning “the pitcher threw out the ball”, Inanimate-Animate: The alarm clock ba the roommate woke up, meaning “the alarm clock woke up the roommate”). Sentences in Reversal conditions were created by reversing the subject and the object in Canonical conditions, and the predictability of the verb was 0%. None of the target verbs were repeated throughout the experiment. Participants performed a plausibility judgement task at the end of each sentence. ERP analyses were time-locked to the onset of the verb for each condition. RESULTS: ERP results showed a significant interaction between Thematic role assignment and Animacy feature at the N400 time window. Follow-up analyses showed a reduced N400 response in Canonical Inanimate-Animate condition relative to the other three conditions. By contrast, at the P600 time window, we only observed a significant Thematic role assignment main effect, with Reversal conditions being more positive than Canonical conditions. Taken together, the P600 main effect suggests that role reversal anomalies are difficult to recover. More importantly, animacy of the noun phrases could modulate the computation of thematic relations online. Participants were sensitive to the marked animacy features of the noun phrases, and they could make advantage of such features to update predictions rapidly. The findings add to the literature by showing that animacy features could be one of the factors that contribute to the inconsistent role reversal findings. It also provides insights into how event relations are represented in the mind.
Topic Areas: Meaning: Combinatorial Semantics, Meaning: Lexical Semantics