December 2022

Prajneya Kumar, a Dual Degree 4th year CLD Student working under the supervision of Dr. Vishnu Sreekumar at the Memory and Neurodynamics Labs (MANDALab, KCIS) made a poster presented on Word Memorability: A Cross-Cultural Perspective at the 9th Annual Conference of Cognitive Science (ACCS9) held at IIT Delhi  from 8 to 10 December.

Research work as explained by the authors Prajneya Kumar; Weizhen Xie,  National Institute of Neurological Disorders and Stroke of the National Institutes of Health, USA; Wilma Bainbridge, Department of Psychology, University of Chicago, USA and Vishnu Sreekumar

Some experiences are more memorable than others. For example, there is a high degree of consistency between people in terms of what images they tend to remember well. While some image features contribute partly to memorability, they do not explain all the variance in memorability scores, leading to the conclusion that memorability is a singular and intrinsic feature of an image. Understanding the features that contribute to memorability can help develop models that can predict memorability or even modify images to enhance their memorability. Words too have intrinsic memorability and semantic properties like animacy, size, and utility contribute significantly to word memorability when tested using free-recall. However, we do not currently know whether similar features explain memorability measured using cued-recall. Furthermore, if semantic features explain word memorability and given that semantic knowledge is built over experience, are highly memorable words in English also highly memorable when translated to a different language, i.e., is word memorability cross-cultural? In the current study, we ask whether memorability is a cross-cultural phenomenon by employing word stimuli in both English and Hindi and investigate whether features like distinctiveness within a semantic network contribute to target word memorability in a cued-recall task. To gain a preliminary understanding of the word properties that contribute to memorability, we first analyze a cued-recall dataset. We then translate the English word pool used in the prior memorability study into Hindi in order to determine whether word memorability generalises across languages and cultures. Finally, we hypothesize that word memorability is related to centrality of words in semantic networks. For instance, the availability of multiple semantic associates in memory may facilitate target memorability of these central words. On the contrary, fewer associates may lend the word a more unique meaning and therefore may enhance associative encoding processes. To test these ideas, we first construct semantic networks independently for English and Hindi and compare the neighborhoods of the same concepts in the two networks. We then present a preliminary analysis of the correlation between network properties and memorability of English words. Finally, we describe the ongoing data collection efforts to measure memorability of Hindi words.