On April 25, Guillermo Montero-Melis from the Centre for Research on Bilingualism at Stockholm University is visiting the lab. He’ll talk to us about “How cross-linguistic differences are reflected in the way we judge similarity – the case of caused motion in Spanish and Swedish”. Abstract below.
My PhD project is framed within the linguistic relativity hypothesis, that differences among languages cause differences in non-linguistic cognition. The study investigates whether different lexicalization patterns of caused motion events in Spanish and Swedish are associated with differences in how speakers of these languages judge similarity among these events. Caused motion events (e.g. ‘He pushed the table into the cave’) typically contain more information about motion than voluntary motion events, yet they have received less attention in the literature. We used 32 video clips that crossed different values of four semantic components: Path, Manner in which the agent causes the object to move, Manner in which the object moves, and Direction (see Hickmann & Hendriks, 2010 for stimuli description). Fourty-four native speakers of each of the two languages (total N=88) were randomly assigned to either the verbal-first or the nonverbal-first condition. In the verbal-first condition, participants first described the events and then carried out a spatial arrangement task in which they had to place the 32 stimuli events on the screen so that their proximity was proportional to their similarity (Goldstone, 1994; Kriegeskorte & Mur, 2012). In the nonverbal-first condition, participants carried out both tasks in the opposite order. The study thus defines a 2-by-2 factorial design with language (Spanish, Swedish) and order (verbal-first, nonverbal-first) as between-subjects factors. The verbal description task showed that speakers of Swedish were more likely to encode Manner-related information in their descriptions than speakers of Spanish, whereas no such difference was found in their encoding of Path, thus replicating earlier results in the motion domain. Results from the non-verbal arrangement task suggest that speakers of Spanish and Swedish were guided by different conceptual components when judging event similarity. These components largely agree with language-specific lexicalization patterns. Cross-linguistic differences seem to hold across both orders, although some of the main trends interact with order. These results suggest more important cross-linguistic differences than earlier studies gauging similarity judgments (e.g., Gennari, Sloman, Malt, & Fitch, 2002), which failed to detect differences in the absence of previous linguistic encoding. The results will be discussed in the light of the novelty of the materials and the design introduced in this study.
Time permitting I will also preliminarily present an automatized approach to analyzing the linguistic data from the event descriptions. Applying a number of algorithms widely used in NLP seems to help us uncover structure in the data that very much agrees with the hand-coded data. This approach has two advantages: it automatizes the work-intense coding process, and it allows us to extract patterns making minimal theoretical assumptions.
Gennari, S. P., Sloman, S. A., Malt, B. C., & Fitch, W. T. (2002). Motion events in language and cognition. Cognition, 83(1), 49–79. doi:16/S0010-0277(01)00166-4
Goldstone, R. L. (1994). An efficient method for obtaining similarity data. Behavior Research Methods, 26(4), 381–386. doi:10.3758/BF03204653
Hickmann, M., & Hendriks, H. (2010). Typological constraints on the acquisition of spatial language in French and English. Cognitive Linguistics, 21(2), 189–215. doi:10.1515/COGL.2010.007
Kriegeskorte, N., & Mur, M. (2012). Inverse MDS: Inferring dissimilarity structure from multiple item arrangements. Frontiers in Perception Science, 3, 245. doi:10.3389/fpsyg.2012.00245