![]() ![]() The way in which a distributional semantic model captures the statistical regularities in word co-occurrence patterns has important implications for how well its output (e.g., word embeddings) aligns with human judgments and for its plausibility as a model of human cognition (Kumar, 2021 Lake & Murphy, 2021). Leveraging semantic similarity from these models has opened an exciting new frontier of research in psychological science and beyond, including in attitudes and emotions (Caluori et al., 2020 Eichstaedt et al., 2021 Vo & Collier, 2013), cultural similarities and differences (Jackson et al., 2021), creativity (Beaty & Johnson, 2021 Dumas et al., 2020 Gray et al., 2019 Green, 2016 Heinen & Johnson, 2018 Johnson et al., 2021 Prabhakaran et al., 2014), and more (see Jackson et al., 2021, and Lake & Murphy, 2021, for recent reviews). While these values do not have symbolic meaning themselves, they can be used to derive semantic similarity between words and texts. These numerical word representations are referred to as word vectors or word embeddings (Günther et al., 2019 Lake & Murphy, 2021). By exploiting the statistical regularities in word co-occurrence patterns in large corpora, each word can be represented by a high-dimensional numerical vector. A word’s distribution or co-occurrence with other words across a large corpus (i.e., body of text) determines its meaning. ![]() For example, the words teacher and educator often co-occur with the words student, classroom, and school, and consequently have similar meaning according to distributional semantics theory. Thus, words that tend to occur in the same contexts have similar meaning. The core principle of distributional semantics theory is that “you shall know a word by the company it keeps” (Firth, 1957, p. It is the extent to which a narrative connects divergent ideas. The new construct is termed divergent semantic integration ( DSI). The goals of the current paper are to (1) develop a new conceptualization of one component of creativity in narratives by integrating creativity theory and distributional semantics theory, (2) examine the psychometric properties of this new construct across diverse narrative texts and diverse participants, and (3) maximize accessibility by providing a tutorial and access to automated assessment of this construct with an open-source web application. Identifying the key components of creativity in narratives interests a broad array of researchers and practitioners including psychologists (D’Souza, 2021 Zedelius et al., 2019) and linguists (Mozaffari, 2013), as well as employers (Florida, 2014), educators (Graham et al., 2002 Vaezi & Rezaei, 2019), creative writers (Bland, 2011), and other practitioners (Barbot et al., 2012). Creativity is among the most valuable attributes in the US workforce, and consequently, automated assessment of creativity is a top priority (Florida, 2014 Lichtenberg et al., 2008). To facilitate new discoveries across diverse disciplines, we provide a tutorial with code (osf.io/ath2s) on how to compute DSI and a web app ( osf.io/ath2s) to freely retrieve DSI scores.ĭeveloping a reliable and automated metric that captures creativity in narrative text has potentially far-reaching and consequential implications. The integration of creativity and distributional semantics theory has substantial potential to generate novel hypotheses about creativity and novel operationalizations of its underlying processes and components. Critically, DSI scores generalized across ethnicity and English language proficiency, including individuals identifying as Hispanic and L2 English speakers. BERT DSI scores showed equivalently high predictive power for expert and nonexpert human ratings of creativity in narratives. BERT DSI scores demonstrated impressive predictive power, explaining up to 72% of the variance in human creativity ratings, even approaching human inter-rater reliability for some tasks. The best-performing model employed Bidirectional Encoder Representations from Transformers (BERT), which generates context-dependent numerical representations of words (i.e., embeddings). Across nine studies, 27 different narrative prompts, and over 3500 short narratives, we compared six models of DSI that varied in their computational architecture. ![]() ![]() We termed the new construct divergent semantic integration ( DSI), defined as the extent to which a narrative connects divergent ideas. We developed a novel conceptualization of one component of creativity in narratives by integrating creativity theory and distributional semantics theory. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |