Slide Slam S14 Sandbox Series
Agreement processing as a domain-general mechanism? An artificial grammar study
Simona Mancini1, Karen Arellano2, Jordi Martorell3; 1BCBL, Basque Center on Cognition Brain and Language
The domain-specificity or generality of core linguistic operations has been the object of extensive theoretical and experimental study. Support for the domain-general view comes from studies implementing different types of artificial grammar (AG) learning paradigms (see Siegelman et al., 2017), which suggest the presence of common mechanisms across domains (e.g., tracking of statistical regularities and sequential processing). We here present the design of a behavioral study aimed to further explore the role of domain-general mechanisms in language by focusing on agreement. Specifically, we plan to test how agreement rules are learned and identified in a visuo-spatial, shape-based AG. We capitalize on Basque agreement patterns, which prototypically show double agreement between verbs and both subjects and objects. Moreover, to better understand the impact of individual differences, we assess to what extent linguistic background (e.g., proficiency) is associated with AG performance. Spanish-Basque bilinguals (N=54) differing in their Basque proficiency will be tested on a visuo-spatial AG based on abstract geometrical shapes (combinations of circles, squares and triangles in different colors). Trials consist of sequences composed by 4 shapes mapped to grammatical functions (adverb, subject, object, and verb). Subject and object shapes fit the verb shape in a puzzle-like manner, mimicking the agreement relations established between verbs and both subjects and objects in Basque. Shapes’ ordering in each sequence follows Basque word order (subject-object-verb), with the adverb shape occupying random pre-verbal positions. During training, participants will perform a self-paced task with correct trials, advancing each shape at their own pace by pressing the spacebar. In the test phase, the self-paced task will comprise correct trials, as well as incorrect ones displaying single instances of agreement (i.e., between verb and either subject or object shapes). An acceptability judgment (AJ) task will be performed after each test trial to evaluate whether the sequence followed the rules of the Basque-like AG. Similar self-paced reading and AJ tasks will be administered with Basque sentences containing both correct and incorrect number agreement relations. Response times (RTs) at final verb-shape position in training trials will show whether progressive learning of visuo-spatial agreement leads to faster RTs, possibly reflecting participants’ prediction of the upcoming element. RTs at verb-shape position in test trials will evidence whether encountering an agreement violation increases RTs compared to agreement-correct trials, suggesting participants’ learning of AG rules. Offline accuracy from the AJ will show whether participants successfully learned the Basque-like AG (i.e., accuracy above chance). Correlations will be run between (i) AG and Basque-sentence performance, and (ii) AG performance and Basque proficiency. If agreement relations are processed similarly across domains, we expect a positive correlation between Basque-like AG and Basque-sentence performance. In addition, if successful learning of the AG is associated with Basque proficiency, greater accuracy and faster RTs in the AG task should emerge as speakers’ proficiency in Basque increases. The results of our study will provide a better understanding of the potential domain-general basis of core language operations like agreement and of the impact of individual-level differences.