Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Composition by prepositional phrases comparing semantic relations in Brazilian Portuguese: an EEG-ERP study
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Marije Soto1,2, Mayda Mayda Rangel Gomes Peres2; 1Department for Brazilian Sign Language, Letters Faculty, Federal University of Rio de Janeiro, 2Graduate Program in Linguistics, Federal University of Rio de Janeiro
Introduction: In this EEG/ERP study, we investigate combinations of noun phrases with prepositional phrase (PP) modifiers in Brazilian Portuguese, contrasting two types of semantic relation: (i) PPs specifying material (ex. “fruteira de bambu”, i.e. a bowl [made of] bamboo) and (ii) PPs specifying the function (ex. “serrote de bamboo”, i.e. a (little) saw [made for cutting] bamboo). These semantic relations may be distinguished by being feature vs. thematic based, respectively. Specifically in the second category, there seems to be the need to compute some hidden meaning in order to understand that ‘bamboo’ is the object for the purpose of cutting. Thus, seemingly identical structures convey distinct meanings, perhaps engaging different processes. In a MEG study, Flick et al. (2021) compared similar semantic distinctions in compounds (such as ‘metal cabinet’ and ‘trophy cabinet) with thematic relations showing greater activation in the left posterior temporal lobe, 100–200 ms after phrasal head onset. We hypothesized that the processing of function meaning may involve coercion; that is, an action or implicit verbal event (i.e. for [cutting] in the case of “serrote de bamboo”) is computed, leading to processing costs. A study by Neufeld et al. (2016) suggests that in the same time window, combinatory processing can be captured with EEG/ERP. We also expected these different semantic relations to affect the N400 component. Methods: We contrasted 2 relation types and a non-combinatory control condition with non-word items (e.g., “livro de nslpk”, “book of nslpk”), we varied the preposition in distractor items (ex. “caixa com bonbons”, box with chocolates). We applied a norming experiment (298 participants) in order to control familiarity and semantic intepretation. We selected 84 pairs (the same target word in two semantic contexts) that fit the semantic interpretation of either material (“is made of”) or function (“is made for x-ing”), with over 90% agreement and which were comparable for familiarity judgements on a 5 point Likert scale. Ratings of 3 (+- 1) were preferred for not being too idiomatic nor too unfamiliar. We collected EEG data with 32 electrodes from 26 university students (M=12) in a phrase picture matching task. As a follow-up, we collected reaction times from self-paced reading and accuracy from a similar task, but no EEG, with 57 participants. Results and discussion: We found that combinatory items, in contrast to combinations with non-words, yielded greater negativity after the third word onset around 200ms (fronto-central channels) and around 400ms (strongest at parietal channels). However, there was no effect for semantic relation (material vs. function). This might be due to methodological issues (e.g. spatial dissociation is possible for MEG, but not EEG), or perhaps coercion and semantic integration are equally costly. An alternative explanation is that there is no difference in online computation, but rather specific task demands may engage pragmatic inference posterior to phrasal processing. Results from the behavioral picture matching study seem to suggest this, as reading times for the PP head were significantly longer for the function interpretation, as we had predicted.
Topic Areas: Syntax and Combinatorial Semantics, Meaning: Lexical Semantics